Trapped Optically Active (E)-Cycloheptene Generated by Enantiodifferentiating Z-E Photoisomerization of Cycloheptene Sensitized by Chiral Aromatic Esters

Article abstract of DOI:10.1021/ja991315l

Highly strained, optically active (E)-cycloheptene (1E) was prepared for the first time in the enantiodifferentiating geometrical photoisomerization of the (Z)-isomer (1Z) sensitized by chiral benzene-tetracarboxylates at -40 to -80°C. Low-temperature irradiations of 1Z in the presence of the chiral sensitizer, and subsequent stereospecific trapping of the optically active photoproduct, 1E, through a Diels-Alder reaction with 1,3-diphenylisobenzofuran or by oxidation with OsO₄, afforded the cycloadduct or trans-1,2-cycloheptanediol, respectively. The enantiomeric excesses (ee's) of the two products were subsequently determined by chiral HPLC or GC. The ee of the product, which was used as a measure of the efficiency of chirality transfer in the excited state, was found to depend critically not only on the chiral sensitizer employed but also on the temperature and solvent employed. Thus, the ee of the product was doubled in an extreme case simply by changing the solvent from dichloromethane to hexane. Furthermore, the product chirality could be switched over a relatively narrow range of temperature as a consequence of the significant contribution of the entropy term in the enantiodifferentiating isomerization within the exciplex intermediate. Sensitization with (-)-bornyl benzenetetracarboxylate in hexane at -80°C gave an ee value of 77%, which is the highest ee ever obtained for an asymmetric photosensitization. Based on the differential activation enthalpy and entropy for the enantiodifferentiating process and the fluorescence quenching experiments with C₅-C₈ cycloalkenes, the origin of the highly efficient enantiodifferentiation and a detailed mechanism for the enantiodifferentiating photoisomerizations are discussed.

Full text of DOI:10.1021/ja991315l

10702
J. Am. Chem. Soc. 1999, 121, 10702-10710
Trapped Optically Active (E)-Cycloheptene Generated by
Enantiodifferentiating Z-E Photoisomerization of Cycloheptene
Sensitized by Chiral Aromatic Esters
Ralf Hoffmann and Yoshihisa Inoue*
Contribution from the Inoue Photochirogenesis Project, ERATO, JST, 4-6-3 Kamishinden,
Toyonaka 565-0085, Japan
ReceiVed April 23, 1999
Abstract: Highly strained, optically active (E)-cycloheptene (1E) was prepared for the first time in the
enantiodifferentiating geometrical photoisomerization of the (Z)-isomer (1Z) sensitized by chiral benzene-
tetracarboxylates at -40 to -80 °C. Low-temperature irradiations of 1Z in the presence of the chiral sensitizer,
and subsequent stereospecific trapping of the optically active photoproduct, 1E, through a Diels-Alder reaction
with 1,3-diphenylisobenzofuran or by oxidation with OsO4, afforded the cycloadduct or trans-1,2-cyclohep-
tanediol, respectively. The enantiomeric excesses (ee’s) of the two products were subsequently determined by
chiral HPLC or GC. The ee of the product, which was used as a measure of the efficiency of chirality transfer
in the excited state, was found to depend critically not only on the chiral sensitizer employed but also on the
temperature and solvent employed. Thus, the ee of the product was doubled in an extreme case simply by
changing the solvent from dichloromethane to hexane. Furthermore, the product chirality could be switched
over a relatively narrow range of temperature as a consequence of the significant contribution of the entropy
term in the enantiodifferentiating isomerization within the exciplex intermediate. Sensitization with (-)-bornyl
benzenetetracarboxylate in hexane at -80 °C gave an ee value of 77%, which is the highest ee ever obtained
for an asymmetric photosensitization. Based on the differential activation enthalpy and entropy for the
enantiodifferentiating process and the fluorescence quenching experiments with C5-C8 cycloalkenes, the origin
of the highly efficient enantiodifferentiation and a detailed mechanism for the enantiodifferentiating
photoisomerizations are discussed.
Introduction
reaction and subsequently trapped stereospecifically to afford
a stable Diels-Alder adduct with trans-configuration. After the
first photochemical synthesis of 1Z was achieved by Kropp
through triplet photosensitization,5 Inoue et al.3,6 investigated
the reaction kinetics of 1E generated by singlet-sensitized
photoisomerization in detail and determined the lifetime (τ) of
1E at low temperatures by trapping the product with acidic
methanol. Under these conditions, τ ) 9.7 min at 1 °C and 68
min at -15 °C. Squillacote et al. performed an NMR spectro-
scopic characterization and were able to study the thermal
stability of this labile olefin.7 More recently, Krebs et al. reported
the synthesis of (E)-1,1,3,3,6,6-hexamethyl-1-sila-4-cyclohep-
tene and its optical resolution by means of chiral gas chroma-
tography.8 The anomalous stability of this hetero-(E)-cyclohep-
tene (τ1/2 ) 5.7 d at 123 °C) is mainly attributed to the enlarged
ring size as a result of the incorporated silicon atom. However,
no attempts to prepare optically active (E)-cycloheptene have
hitherto been made.
Chirality transfer from optically active sensitizers to prochiral
substrates in the excited state is an intriguing and attractive
process in photochemistry.1 Thus, the enantiodifferentiating
geometrical photoisomerizations of cyclooctene and its deriva-
tives, sensitized by chiral aromatic esters, have been actively
investigated in recent years.2 In these photoisomerizations, the
rotational relaxation of the alkene double bond occurring within
an exciplex intermediate constitutes the enantiodifferentiating
process, producing the constrained chiral (E)-isomer in a
relatively high enantiomeric excess (ee). Because the chirality
of (E)-cyclooctene originates from its distorted structure (10
kcal/mol higher in energy than the (Z)-isomer), caused by the
hindered rope-jumping motion of the methylene chain bridging
the double bond, the much more strained substrate (E)-
cycloheptene (1E) (27 kcal/mol higher in energy than the (Z)-
isomer 1Z) is quite an interesting target molecule from a
mechanistic and synthetic point of view.3 Compared with its
higher homologues, little is known about (E)-cycloheptene. 1E
was first generated by Corey et al.4 in a thermal elimination
In view of the thermal instability of (E)-cycloheptene (1E)
and the relatively high ee values obtained in the enantiodiffer-
(1) (a) Rau, H. Chem. ReV. 1983, 83, 535. (b) Inoue, Y. Chem. ReV.
1992, 92, 741. (c) Everitt, S. R. L.; Inoue, Y. In Organic Molecular
Photochemistry; Ramamurthy, V., Schanze, K., Eds.; Marcel Dekker: New
York, 1999; pp 71-130.
(2) (a) Inoue, Y.; Yamasaki, N.; Yokoyama, T.; Tai, A. J. Org. Chem.
1992, 57, 1332. (b) Inoue, Y.; Yamasaki, N.; Yokoyama, T.; Tai, A. J.
Org. Chem. 1993, 58, 1011. (c) Inoue, Y.; Matsushima, E.; Wada, T. J.
Am. Chem. Soc. 1998, 120, 10687.
(4) Corey, E. J.; Carey, F. A.; Winter, R. A. E. J. Am. Chem. Soc. 1965,
87, 934.
(5) Kropp, P. J. J. Am. Chem. Soc. 1969, 91, 5783.
(6) (a) Inoue, Y.; Ueoka, T.; Hakushi, T. J. Chem. Soc. Perkin Trans. 2
1984, 2053. (b) Inoue, Y.; Hagiwara, S.; Daino, Y.; Hakushi, T. J. Chem.
Soc., Chem. Commun. 1985, 1307.
(7) Squillacote, M.; Bergman, A.; De Felippis, J. Tetrahedron Lett. 1989,
30, 6805.
(3) Inoue, Y.; Ueoka, T.; Kuroda, T.; Hakushi, T. J. Chem. Soc., Perkin
Trans. 2 1983, 983.
(8) Krebs, A.; Pforr, K.-I.; Raffay, W.; Tho¨lke, B.; Ko¨nig, W. A.; Hardt,
I.; Boese, R. Angew. Chem., Int. Ed. Engl. 1997, 36, 159.
10.1021/ja991315l CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/06/1999

Trapped Optically ActiVe (E)-Cycloheptene
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10703
Scheme 1
gave the adduct 3 or the oxidation product 4, respectively. No
significant enantiomeric loss is expected to occur during either
trapping procedure, since these reactions are known to proceed
stereospecifically and are also carried out at low temperatures.
Even in the presence of residual 1Z, only 1E was selectively
trapped by 2, affording the adduct 3 as sole product, and 1Z
did not give a Diels-Alder product in an appreciable yield when
reacted with 2 under the conditions employed. Similarly, no
kinetic resolution in the trapping process is expected to occur
in the absence of any specific interaction or complex formation
between the trapping agent and chiral sensitizers in the groud
state. In fact, all of the temperature dependence experiments
gave good straight lines in the Eyring plots, as described below.
Such results would not be obtained if the two independent
processes, i.e., the enantiodifferentiating Z-E isomerization in
the exciplex and the kinetic resolution in the trapping reaction,
were jointly responsible for the determination of the ee.
In contrast, OsO4 oxidized both 1Z and 1E to give the cis-
and trans-1,2-cycloheptanediols, respectively. The enantiomers
of the Diels-Alder adducts (R,R)-3 and (S,S)-3 were satisfac-
torily separated by chiral HPLC, while the ratio of the cis-diol
and the enantiomeric trans-diols (S,S)-(+)-4 and (R,R)-(-)-4
were determined by chiral GC.
To assign the absolute configurations of the primary product
1E and the Diels-Alder adduct 3, a preparative scale photo-
sensitization of 1Z with (-)-bornyl 1,2,4,5-benzenetetracar-
boxylate (5) was carried out, and the configuration of 1E
produced was correlated to the known absolute configuration
of trans-diol 4 by chemical derivatization.9 Half of the irradiated
solution was treated with OsO4, as above, and the resulting
trans-diol was separated from the reaction mixture by means
of preparative GC on an achiral column. The isolated trans-
diol was dissolved in dichloromethane and subjected to pola-
rimetric measurement, giving a negative optical rotation (R )
-0.008° at 589 nm and -0.015° at 367 nm; error (0.001°).
This proved the predominant formation of (R,R)-(-)-4, and
therefore the excess of (S)-1E over (R)-1E in the original
photoproduct. The other half of the irradiated solution was
treated with 2 to yield the adduct 3. Based on the absolute
configuration of the major photoproduct (S)-1E, the predominant
enantiomer was assigned to be (R,R)-3, which was in turn shown
to elute as the first fraction in the chiral HPLC analysis.
Photosensitization and E/Z Ratio. In previous studies
concerning the enantiodifferentiating photoisomerization of
cyclooctene and its derivatives,2,10 a wide variety of optically
active polyalkyl benzenepolycarboxylates were employed as
chiral singlet sensitizers. Since one of the most interesting
aspects of this work is the examination of the effects of the
ring size and strain of (E)-cycloalkenes on their photochemical
and stereochemical behavior, we employed the same chiral
sensitizers which have been extensively studied and are known
to give satisfactory chemical and optical yields in the cy-
clooctene case, i.e., (-)-tetrabornyl, (-)-tetramenthyl, and (-)-
tetrakis(8-phenylmenthyl) 1,2,4,5-benzenetetracarboxylates
(5-7) (Scheme 1).2
entiating photoisomerization of (Z)-cyclooctene sensitized by
chiral benzenepolycarboxylates,2 the asymmetric photochemical
approach is perhaps the most direct and logical choice for the
generation and trapping of optically active 1E. Here we will
demonstrate that optically active 1E can be efficiently generated
with high ee’s (up to 77%) and can be stereospecifically trapped
by a subsequent Diels-Alder reaction with diphenylisobenzo-
furan or oxidation with OsO4, as illustrated in Scheme 1. The
smaller ring size and higher strain energy of 1E than those of
(E)-cyclooctene are thought to be responsible for the much
stronger temperature and solvent dependence observed for the
ee values of 1E.
Results
Trapping Experiments and Determination of the Absolute
Configurations of the Products. Chromatographic and spec-
troscopic methods are well established for the direct determi-
nation of the optical purity or ee of chiral compounds at ambient
and higher temperatures; e.g., the ee of moderately stable (E)-
cyclooctene has been determined by chiral GC.2 In the present
study, we first planned to separate and preferabaly isolate the
enantiomers of (E)-cycloheptene by chiral GC or HPLC at a
low temperature. However, after discussing the technical
problems associated with the low-temperature GC and HPLC
with the experts from Shimadzu and Daicel Co., we could not
find an appropriate instrument or chiral column which can be
operated at temperatures lower than at least -20 °C. At the
same time, we tried to solve the problem by using the trapping
technique. To preserve the chirality of the photoproduct, an
enantiomeric mixture of 1E produced in the asymmetric
photosensitization should be efficiently and stereospecifically
converted to the corresponding enantiomers of a thermally stable
adduct, the ee of which is then determined by conventional
methods and is regarded as being equivalent to that of the
photoproduct 1E. As described below, the Diels-Alder and
oxidation reactions proceeded even at low temperatures, leading
to the corresponding products in good to excellent yields with
stereoretention. Hence, we decided to use the trapping technique
for the determination of the enantiomeric excess of the (E)-
isomer. We notice that this is an indirect method, and certainly
the isolation of optically active (E)-cycloheptene and the direct
measurement of its chiroptical properties are intriguing future
targets.
The time course of the photoisomerization was examined first.
Dichloromethane solutions containing 1Z (10 mM) and sensi-
tizer 5 or 6 (2 mM) were irradiated for varying periods of time
at -70 °C under an argon atmosphere, using a 300-W high-
pressure mercury lamp. After a designated period of irradiation,
To establish procedures for the photolysis and subsequent
trapping at low temperatures, preparative scale irradiations of
1Z with methyl benzoate as an achiral singlet sensitizer were
carried out. The addition of the very reactive diene 1,3-
diphenylisobenzofuran (2) or OsO4 to the irradiated solution at
-70 °C, followed by gradual warming to room temperature,
(9) Posternak, T.; Reymond, D.; Friedli, H. HelV. Chim. Acta 1955, 38,
205. Mousseron, M.; Richaud, R. Bull. Soc. Chim. Fr. 1946, 643 and 647.
(10) (a) For 1-methylcyclooctene, see: Tsuneishi, H.; Hakushi, T.; Tai,
A.; Inoue, Y. J. Chem. Soc., Perkin Trans. 2 1995, 2057. (b) For
1,3-cyclooctadiene, see ref 2b.

10704 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Hoffmann and Inoue
0.5% ee with a standard deviation of (0.5% ee in favor of (S,S)-
3, which eluted second, and each ee value was usually deter-
mined by five HPLC runs. Since the standard deviation of the
ee determined was also in the range of (0.5%, the total error
can be up to (1% in a few cases. The irradiation bath temper-
ature was constant within (0.5 °C. The ee values of 3 produced
in the photosensitizations of 1Z with 5, 6, and 7 in some solvents
of different polarity at -43 to -80 °C are listed in Table 1.
For a more quantitative treatment of the temperature depen-
dence, the Eyring equations for the formation rate constants (kS
and kR) of (S)- and (R)-1E were combined to give the expression
ln(kS/kR) ) -∆∆Hq/RT + ∆∆Sq/R, in which the differential
activation enthalpy ∆∆Hq ) ∆HqS - ∆HqR and the differential
activation entropy ∆∆Sq ) ∆SqS - ∆Sq . Thus, ln(kS/kR) should
R
give a linear plot against 1/T. The relative rate constant kS/kR is
experimentally equal to (100 + %ee)/(100 - %ee). The
temperature-dependent ee values obtained in the photosensiti-
zation with 5 and 6 in several different solvents were analyzed
according to the above kinetic equation to give the plots shown
in Figures 2 and 3.
Figure 1. (E)- to (Z)-cycloheptene ratio (1E/1Z), as evaluated by 3/1Z
ratio, as a function of irradiation time upon sensitization of 1Z with
(-)-bornyl (5, b) and (-)-menthyl (6, 2) 1,2,4,5-benzenetetracarboxy-
late in dichloromethane.
each solution was poured into dichloromethane containing an
excess of trapping agent 2, cooled to -70 °C. The resultant
mixture was left overnight in the bath, gradually warming to
room temperature, and then analyzed by chiral GC to give the
ratio of adduct 3 and remaining 1Z. As can be seen from Figure
1, both benzenetetracarboxylate sensitizers 5 and 6 resulted in
very similar 3/1Z ratios which rapidly increased with increasing
irradiation time, reaching a plateau of 0.1 at 25-40 min, and
then gradually decreased upon prolonged irradiations of up to
For each sensitizer-solvent combination, a good straight line
plot was drawn, indicating that a single enantiodifferentiation
mechanism is operating in this temperature range, giving high
ee values of up to 77%. It is noted that (S)-1E is favored at
lower temperatures, irrespective of the chiral sensitizer and
solvent employed. As is frequently observed in the enantio-
differentiating photoisomerization of cylcooctene,2 the product
chirality was switched from S to R by raising the temperature
upon photosensitization with (-)-menthyl benzenetetracarboxy-
late 6 (Figure 3). Product chirality is also expected to be inverted
at higher temperatures in the case of sensitizer 5 (Figure 2).
Beyond this critical point, called the equipodal temperature
(which decreases with increasing solvent polarity), the ee of
the product increases with increasing temperature, a behavior
which is not anticipated in conventional asymmetric syntheses.
It is also interesting to note that the slopes of the regression
lines for cycloheptene are much steeper than those reported for
the photoisomerization of cyclooctene with the same sensitizers.2
By changing the solvent from polar dichloromethane to ether
and then to nonpolar hexane, the regression line is significantly
shifted upward for both sensitizers, favoring the (S)-enantiomer.
Simultaneously, the slope of the plot increases with decreasing
solvent polarity. In an extreme case, as a combined effect of
the upward shift and the steeper slope, the ee of the product
obtained upon sensitization with 5 is enhanced from 30% in
dichloromethane at -75 °C to 77% in hexane at -80 °C (Figure
3). This is the highest ee ever reported for an enantiodifferen-
tiating photosensitized reaction. In contrast, such a marked
solvent effect has not been observed for the photosensitized
isomerization of cyclooctene or its derivatives.2
90 min. The apparent photostationary-state E/Z ratio, (E/Z)pss
,
of 0.1 is comparable to the corresponding ratio ((E/Z)pss ) 0.10)
reported for cyclooctene at 25 °C,11 but it is considerably lower
than that observed for cyclooctene at -78 °C (0.38).11 This
lower (E/Z)pss ratio at a comparable temperature may be attrib-
uted to the higher strain observed for 1E than for (E)-cyclo-
octene, and the gradual decrease upon prolonged irradiation is
ascribed to secondary photoreactions of the highly strained 1E.
Temperature Dependence of Ee and Activation Param-
eters. Photochemistry offers an excellent method for conducting
experiments over a very wide temperature range without
affecting the reaction mechanism. Using this advantage, we have
investigated the temperature dependence of the ee of the product
in enantiodifferentiating photosensitized Z-E isomerizations of
cyclooctene and its derivatives,2 revealing intriguing behavior
over a temperature range from 50 to -90 °C. In the present
study, such a wide temperature range could not be applied due
to the thermal instability of 1E at ambient temperatures; e.g., τ
) 9.7 min at 1 °C and 45 s at 25 °C.3 We therefore investigated
the temperature dependence of the enantiodifferentiating pho-
toisomerization between -43 and -80 °C, employing the
representative benzenetetracarboxylate sensitizers shown in
Scheme 1.
To investigate the effect of the highly congested, electron-
donating chiral ester groups in sensitizer, we studied the
photosensitization of 1Z with (-)-8-phenylmenthyl benzene-
tetracarboxylate (7), which is known to form an intramolecular
exciplex with charge-transfer characteristics between the ben-
zenetetracarboxylate core (acceptor) and the 8-phenyl group
(donor).2b Interestingly, this sensitizer gave very high ee values
not only at low temperatures (66% ee at -45 °C and 74% ee at
-70 °C) but also at 25 °C (46% ee, extrapolated). Figure 4
illustrates the temperature dependence of the ee obtained with
sensitizers 5, 6, and 7 in hexane. The menthyl ester 6 gives
In a typical experiment, five quartz tubes containing a solution
(10 mL) of 1Z (10 mM) and sensitizer (2 mM) were placed
around a 300-W high-pressure mercury lamp, and the whole
system was immersed in a temperature-controlled bath. Photo-
sensitizations were carried out in a variety of solvents with
differing polarities, as detailed in Table 1. After irradiation at
low temperature and the subsequent trapping procedure with 2,
the ee of adduct 3 produced was determined by HPLC on a
Daicel Chiralpak AD column, which afforded a satisfactory
separation between the enantiomers.12 Calibration with racemic
3 indicated that this method gives a small systematic error of
(11) Yamasaki, N.; Yokoyama, T.; Inoue, Y.; Tai, A. J. Photochem.
Photobiol., A: Chem. 1989, 48, 465 and unpublished results on photosen-
sitization at low temperatures.
(12) Determination of the ee by chiral GC is not possible. We observed
at least partial racemization of 3 upon purification on a preparative GC,
which is attributed to a retro[4 + 2] reaction; cf. mass spectra.

Trapped Optically ActiVe (E)-Cycloheptene
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10705
Table 1. Temperature and Solvent Effects upon Enatiodifferentiating Photoisomerization of 1Z Sensitized by (-)-Bornyl (5), (-)-Menthyl
(6), and (-)-8-Phenylmenthyl (7) 1,2,4,5-Benzenetetracarboxylates
entry
sensitizer
substrate
solvent
pentane
irradiation time/min
temp/°C
% ee
ref
1
2
3
4
5
6
7
8
9
5
cyclooctene
25
-40
-78
-45
-74
-80
-43
-68
-80
-43
-45
-55
-68
-75
-80
-50
-65
-77
-43
-55
-75
25
-40
-60
-43
-55
-70
-45
-55
-76
-50
-65
-75
25
-9.6
7.1
2a
2a
2a
21.6
52.1
67.0
68.6
45.6
63.8
76.8
35.7
35.8
44.3
53.6
56.2
58.3
16.9
21.1
30.8
46.2
49.7
66.9
11.5
27.3
30.9
-1.6
6.3
28.5
-16.1
-6.8
13.8
-14.3
-0.7
4.8
49.5
27.3
16.1
66.0
70.3
74.1
1
pentane
hexane
ether
45
45
45
45
45
40
45
50
90
45
45
40
90
90
90
45
45
45
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
this work
2a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
dichloromethane
methylcyclohexane
pentane
6
cyclooctene
2a
2a
1
hexane
45
45
45
50
60
45
90
90
90
this work
this work
this work
this work
this work
this work
this work
this work
this work
2b
ether
dichloromethane
pentane
7
cyclooctene
-40
-87
-45
-55
-70
2b
2b
1
hexane
45
45
45
this work
this work
this work
Figure 3. Temperature dependence of enantiomeric excess (ee) of 1E
upon sensitization of 1Z with (-)-menthyl 1,2,4,5-benzenetetracar-
boxylate (6) in hexane (b), ether (2), and dichloromethane (9); plots
of ln(kS/kR), or ln [(100 + % ee)/(100 - %ee)], versus reciprocal
temperature.
Figure 2. Temperature dependence of enantiomeric excess (ee) of 1E
upon sensitization of 1Z with (-)-bornyl 1,2,4,5-benzenetetracarboxy-
late (5) in hexane (b), ether (2), and dichloromethane (9); plots of
ln(kS/kR), or ln [(100 + % ee)/(100 - %ee)], versus reciprocal
temperature.
significantly lower ee’s in this temperature range as compared
with the bornyl ester 5, although the slopes for both sensitizers
do not differ greatly. In contrast, the use of the 8-phenylmenthyl
ester 7 results in much smaller temperature dependence of the
ee, probably due to a reduced conformational freedom within
the sensitizer. It is reasonable that sensitizers 5 and 6 afford
the corresponding (S)-enantiomer of (E)-cycloalkenes at low
temperatures, irrespective of the ring size, although the con-
gested sensitizer 7 is expected to give (S)-1E but to switch over
to give (R)-(E)-cyclooctene at temperatures below -114 °C (the

10706 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Hoffmann and Inoue
sensitization and enantiodifferentiation mechanisms are operat-
ing in the two systems. The enantiodifferentiating step of the
sensitized photoisomerization is believed to be the rotational
relaxation of the double bond within the exciplex, formed
between the excited chiral sensitizer and cyclooctene.2 This
mechanism, originally proposed for cyclooctene2a and later for
1,3-cyclooctadiene10b and 1-methylcyclooctene,10a can be de-
scribed as follows. The initial quenching of the singlet-excited
sensitizer (1S*) by the substrate 1Z leads to the formation of
an encounter exciplex [1Z‚‚‚1S*]. Rotational relaxation around
the double bond of [1Z‚‚‚1S*] gives a relaxed diastereomeric
exciplex pair [(R)-1p‚‚‚S*] and [(S)-1p‚‚‚S*], which spontane-
ously fall apart, liberating the corresponding enantiomers of
twisted cycloheptene singlet (R)-1p and (S)-1p, which in turn
decay stereospecifically to (R)-1E and (S)-1E or to 1Z. Upon
prolonged irradiation, the system reaches a photostationary state
as a consequence of the concomitant E-to-Z photoisomerization.
In principle, the quenching of the chiral sensitizer by enantio-
meric (E)-cyclooctene can occur at different rates, but in
actuality the quenching process has been demonstrated to be
non-enantiodifferentiating as a result of the very fast quenching
rate constants which approach diffusion control.2a Experimen-
tally, the possibility of enantiodifferentiating quenching can be
proved or disproved by examining the dependence of the ee of
the product upon conversion, or by using racemic 1E as the
starting material, although the latter check cannot be used in
the present study because of its instability. We therefore
performed the photosensitization of 1Z with 5 in ether for
different irradiation times (40-90 min) at -43 to -80 °C,
giving an excellent linear fit, as shown in Figure 2 (see entries
10-15 in Table 1). Thus, it is shown that the varied irradiation
time and extent of conversion do not appreciably affect the ee
of the product.
Figure 4. Temperature dependence of enantiomeric excess (ee) of 1E
upon sensitization of 1Z with (-)-bornyl (5, b), (-)-menthyl (6, 2),
and (-)-8-phenylmenthyl (7, 9) 1,2,4,5-benzenetetracarboxylates in
hexane; plots of ln(kS/kR), or ln [(100 + % ee)/(100 - %ee)], versus
reciprocal temperature.
equipodal temperature for the photosensitization of cyclooctene
with 7).2b
From the slope and intercept of the regression lines in Figures
2-4, the differential enthalpy and entropy of activation were
calculated to give the ∆∆Hq and ∆∆Sq values for each sen-
sitizer-solvent combination. As can be seen from the activation
parameters listed in Table 2, the ∆∆Hq and ∆∆Sq values
determined for 1 are much larger and depend more on the
solvent polarity than the corresponding values reported for
cyclooctene.2a,b
Fluorescence Quenching. Since sensitizers 5 and 6 emit
weak fluorescence in solution, we carried out quantitative
fluorescence quenching experiments of 5 and 6 with 1Z in order
to obtain information concerning the quenching process and
exciplex formation, and these results can be compared with those
for cyclooctene and cyclopentene. As exemplified for menthyl
benzenetetracarboxylate 6 in aerated hexane at room tempera-
ture, fluorescence at 330 nm was gradually quenched upon the
stepwise addition of 1Z, and a new emission attributable to an
exciplex emerged at a longer wavelength, accompanied by an
isoemissive point at 385 nm (see Figure 5). The exciplex
emission spectrum, shown in the inset of Figure 5, was obtained
by subtracting the intensity-normalized original spectrum at [1Z]
) 0 M from the final spectrum at [1Z] ) 0.64 M.
According to the conventional Stern-Volmer equation, F/F0
) 1 + kqτ[1Z], the relative fluorescence intensity (F/F0) at 330
nm in the presence (F) and absence (F0) of 1Z was plotted
against the quencher concentration, giving a good straight line,
as shown in Figure 6. For comparison purposes, similar
quenching experiments were repeated with sensitizers 5 and 6,
using the quencher olefins 1Z and cyclopentene in several
solvents. The Stern-Volmer constant (kqτ), obtained as the slope
of the plot, is listed for each sensitizer-quencher combination
in Table 3.
The fluorescence quenching experiments not only support the
proposed mechanism, which involves the efficient quenching
of an excited singlet sensitizer and the subsequent formation of
exciplex intermediate, but also afford quantitative information
concerning the quenching kinetics and the properties of the
exciplex formed. As shown in Table 3, all of the C5-C8
cycloalkenes efficiently quench the fluorescence of sensitizers
5 and 6 at quenching rate constants (kq) of 109-1010 M-1 s-1
.
The very large kq values for (E)-cyclooctene, which are close
to the diffusion limit, have been attributed to its higher strain
and lower singlet energy and are believed to be responsible for
non-enantiodifferentiating quenching.2a The much weaker ex-
ciplex emission observed at longer wavelength for (E)-cy-
ex
clooctene (λmax ≈ 455 nm) indicates extensive conformational
relaxation of an exciplex which has stronger charge-transfer
character.2a The Stern-Volmer constants (kqτ) obtained for 1Z
and cyclopentene in the present study are generally of the same
order of magnitude, irrespective of the solvent used. However,
a closer examination reveals that the kq for C5-C8 (Z)-
cycloalkenes gradually increases with decreasing ring size, from
(1.1-1.4) × 109 M-1 s-1 for C8 to (1.8-2.7) × 109 M-1 s-1
for 1Z and then to (3.1-3.2) × 109 M-1 s-1 for the C5
cycloalkene. Contrastingly, the exciplex fluorescence maximum
(λmaxex) shifts slightly to shorter wavelengths, from 410-415
nm for C8 to 408-410 nm for 1Z and then to 395-405 nm for
the C5 cycloalkene, probably reflecting the reduced conforma-
tional relaxation for the smaller sized cycloalkenes. More evident
bathochromic shifts of λmaxex observed in polar solvents, which
vary from 408 nm in hexane, 418 nm in ether, and 435 nm in
dichloromethane, unambiguously demonstrate the charge-
transfer character of the exciplex. A similar polar solvent-
Discussion
Sensitization and Enantiodifferentiation Mechanisms. The
similarities observed between the fluorescence quenching
parameters, the temperature-dependence profile of ee, and the
absolute configuration of the product obtained in the present
asymmetric photosensitization of 1Z and those reported for
cyclooctene2a,b lead us to the conclusion that essentially the same

Trapped Optically ActiVe (E)-Cycloheptene
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10707
Table 2. Differential Activation Parameters for Enantiodifferentiating Photoisomerizations of Cyclooctene and Cycloheptene (1) Sensitized by
(-)-Bornyl (5), (-)-Menthyl (6), and (-)-8-Phenylmenthyl (7) 1,2,4,5-Benzenetetracarboxylates in Some Solvents
5
6
7
ET(30)a/
∆∆Hq/
∆∆Sq/
∆∆Hq/
∆∆Sq/
∆∆Hq/
∆∆Sq/
substrate
solvent
kcal mol-1 kcal mol-1 cal mol-1 K-1 kcal mol-1 cal mol-1 K-1 kcal mol-1 cal mol-1 K-1
cycloocteneb pentane
ether
31.0
34.5
31.0
-0.77
-0.59
-1.28
-2.43
-1.90
-1.48
-0.95
-3.00
-2.28
-3.26
-8.68
-6.36
-4.92
-3.63
-0.61
-1.55
0.62
3.90
1c
pentane
hexane
methylcyclohexane
ether
dichloromethane
31.0
-2.18
-9.62
-1.16
-1.90
(30.9)d
34.5
-1.72
-1.37
-8.16
-6.67
40.7
a Solvent polarity parameter; Reichardt, C. SolVents and SolVent Effects in Organic Chemistry, 2nd ed.; VCH: Weinheim, 1988. b References 2a
and 2b; revised ∆∆Sq values. c This work. d ET for cyclohexane.
is known that the exciplex intervening in this enantiodifferen-
tiating photosensitization has a degree of charge-transfer
character. This allows for charge separation between the radical
anionic sensitizer and the radical cationic 1Z in polar solvents,
which diminishes chiral recognition in the exciplex intermediate.
In conclusion, it should be emphasized that the sensitizer-
substrate distance in the exciplex intermediate is critical, and
changes that lead to an increased separation remove any
enantiodifferentiation in the excited state.
Temperature Switching of Product Chirality. The unusual
effect of temperature on the ee of the product is perhaps the
most intriguing part of this enantiodifferentiating photoisomer-
ization. In most asymmetric thermal and photochemical reac-
tions, the optical yield is generally believed to be enhanced by
lowering the reaction temperature. Indeed, a variety of asym-
metric reactions, including some photochemical examples,13
have been demonstrated to follow this empirical rule, although
only relatively narrow ranges of temperature have been em-
ployed in the thermal reactions. This widespread belief that the
Figure 5. Fluorescence quenching of (-)-menthyl 1,2,4,5-benzene-
tetracarboxylates 6 with 1Z in hexane and the exciplex emission (inset)
obtained by spectrum subtraction.
optical yield intrinsically increases at lower temperatures is
based on the assumption that the chiral recognition is predomi-
nantly governed by the enthalpic term, which is a reflection of
the magnitude of steric hindrance of the attacking reagent, and
the contribution of the entropy term is simply disregarded in
most cases. However, the temperature dependence study of the
enantiodifferentiating photoisomerization of cylooctene has
revealed that an unprecedented switching of product chirality
can be caused simply by changing the irradiation temperature.2
This temperature-switching phenomenon can only be interpreted
in terms of a significant contribution of the entropy term in the
enantiodifferentiating step. Thus, the nonzero ∆∆Sq, which
possesses the same sign as ∆∆Hq, is the origin of their
temperature-switching effect. The present study on 1Z further
indicates that such an apparently unusual switching behavior,
originally observed for cyclooctene, is not exceptional but is
instead a more common phenomenon which should be observed
in a wide variety of asymmetric (photo)chemical reactions if a
sufficiently wide temperature range is employed. In this respect,
photochemical asymmetric synthesis appears to be more prom-
ising, since the reaction mechanism does not, in principle, vary
over a wide temperature range.
Figure 6. Stern-Volmer plot for the fluorescence quenching of (-)-
menthyl 1,2,4,5-benzenetetracarboxylate 6 with 1Z in hexane.
induced bathochromic shift is also reported for (Z)-cyclooctene,
ex
which shows a λmax of 410 nm in pentane and 445 nm in
acetonitrile (Table 3).2b
As a logical consequence of the ring size of the quencher, a
less hindered approach to the active site of the sensitizer is
expected for the smaller substrate, and this was confirmed
experimentally by the 2-3 times faster quenching by the
smaller-sized cycloalkenes. The smaller ring size of 1Z enables
a closer approach to the chiral sensitizer and the subsequent
formation of more intimately interacting exciplex, which should
be responsible, at least in part, for the generally higher ee values
obtained with 1 than with cyclooctene. In contrast, the ee of
the product is significantly reduced by the use of polar solvents.
From fluorescence quenching experiments in polar solvents, it
Activation Parameters and Enthalpy-Entropy Compen-
sation. When discussing the mechanism, efficiency, and factors
controlling the enantiodifferentiation in the excited state, a global
(13) (a) Henin, F.; Mortezaei, R.; Muzart, J.; Pete, J.-P. Tetrahedron
Lett. 1985, 26, 4945. (b) Piva, O.; Pete, J.-P. Tetrahedron Lett. 1990, 31,
5157. (c) Seebach, D.; Oei, J.-A.; Daum, H. Chem. Ber. 1977, 110, 2316.
(d) Jarosz, S.; Zamojski, A. Tetrahedron 1982, 38, 1447. (e) Boyd, D. R.;
Campbell, R. M.; Coulter, P. B.; Grimshaw, J.; Neill, D. C. J. Chem. Soc.,
Perkin Trans. 1 1985, 849. (f) Demuth, M.; Palomer, A.; Sluma, H.-D.;
Dey, A. K.; Kru¨ger, C.; Tsay, Y.-H. Angew. Chem. 1986, 98, 1093.

10708 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Hoffmann and Inoue
Table 3. Fluorescence Quenching of (-)-Bornyl (5) and (-)-Menthyl (6) 1,2,4,5-Benzenetetracarboxylates with C5-C8 Cycloalkenes in Some
Solvents at Room Temperature
c
sensitizer
quencher
solvent
pentane
pentane
hexane
ET(30)a/kcal mol-1
kqτb/M-1
kq /109 M-1 s-1
λmaxex d/nm
ref
5
(E)-cyclooctene
(Z)-cyclooctene
1Z
cyclopentene
(E)-cyclooctene
(Z)-cyclooctene
31.0
31.0
31.0
31.0
31.0
31.0
45.6
31.0
34.5
40.7
31.0
1.80
0.21
0.35
0.62
2.08
0.28
0.83
0.53
0.34
0.52
0.63
9.0
1.1
1.8
3.1
10.4
1.4
4.2
2.7
nde
nd
∼455
415
410
395
∼455
410
445
408
418
435
405
2a
2a
this work
this work
2a
hexane
6
pentane
pentane
acetonitrile
hexane
ether
dichloromethane
hexane
2a
2b
1Z
this work
this work
this work
this work
cyclopentene
3.2
b
a Solvent polarity parameter; see footnote a of Table 2. Stern-Volmer constant. c Quenching rate constant, calculated for the hexane solutions
by using the same lifetime (τ 0.2 ns) determined for 5 and 6 in pentane (ref 2a). d Exciplex fluorescence maximum. e Not determined.
efficient chiral recognition process. Interestingly, while the data
points available for 1 are limited in number and rather scattered,
the activation parameters obtained with the two cycloalkenes
fall essentially on the same regression line, which has a
correlation coefficient of 0.978, giving the expression ∆∆Hq
) 0.257∆∆Sq + 0.02. It is noted that the data points for the
highly congested sensitizers, carrying 8-phenylmenthyl and
8-cyclohexylmenthyl auxiliaries, deviate significantly from the
regression line and have not been used in the regression analysis.
From the slope of the line, we can calculate an isoenantiodif-
ferentiating temperatures (Tiso) of 257 K, at which no enatiod-
ifferentiation takes place. The single regression line for both
cycloalkenes unambiguously demonstrates that essentially the
same enantiodifferentiation mechanism operates in the asym-
metric photosensitization of cycloalkenes even beyond the
apparent differences of chiral auxiliary, sensitizer, substrate, and
solvent used, while the plots for the congested sensitizers that
deviate from this line suggest the operation of a different type
Figure 7. Enthalpy-entropy compensation plot for cycloheptene (1)
(b), cyclooctene (O), and 1-methylcyclooctene (4).
of exciplex or a different enantiodifferentiation mechanism.
comparison of the differential activation parameters for cyclo-
heptene 1 and cyclooctene becomes interesting. As can be seen
Conclusions
from the data in Table 2, both the ∆∆Hq and ∆∆Sq values
obtained for 1 are unexpectedly greater (by a factor of 2-3)
than those reported for cyclooctene with the same chiral
sensitizers in similar solvents, indicating a much more efficient
enantiodifferentiation process for the smaller-sized 1, which is
demonstrated above in the fluorescence quenching experiments.
Specifically, the larger ∆∆Sq values may indicate larger
conformational changes upon the enantiodifferentiating rota-
tional relaxation of 1 than those experienced by cyclooctene in
the corresponding exciplex.
According to the Gibbs-Helmholtz equation, the enthalpy
and entropy terms contribute to the differential activation free
energy ∆∆Gq, and thus to the ee of the product. As a
consequence of ∆∆Hq and ∆∆Sq obtained for 1 and cyclooctene
both having the same sign, the regression lines inevitably cross
the x-axis, as shown in Figures 2 and 3. At this point, which
we have called the equipodal temperature and which is
characteristic to the specific solvent-sensitizer combination, the
enthalpy (∆∆Hq) and entropy (T∆∆Sq) terms cancel each other.
It is interesting to examine the general validity of the
compensatory enthalpy-entropy relationship, which was ob-
served for the enantiodifferentiating photoisomerization of
cyclooctene. In Figure 7, the ∆∆Sq values obtained in this study
and those reported for cyclooctene2a,b and 1-methylcyclooctene10b
are plotted against the ∆∆Hq values. First, it should be pointed
out that, compared to the plots for cyclooctene and methylcy-
clooctene, where the data are accumulated near the origin, the
corresponding data points for 1 are located in much more
negative regions of ∆∆Hq and ∆∆Sq, indicating a highly
In the present study, we have revealed several important and
novel facts concerning the mechanistic and synthetic aspects
of asymmetric photosensitization as well as on the generation,
configuration, stability, and reactivity of optically active 1E.
The low-temperature photosensitization of 1Z with chiral alkyl
benzenetetracarboxylates affords optically active 1E in good
chemical and optical yield, which is optically and thermally
stable below -40 °C. 1E can be efficiently and stereospecifically
trapped by a Diels-Alder reaction with diphenylisobenzofuran,
or by oxidation with osmium tetraoxide, producing the corre-
sponding adduct or oxidation product. As compared with the
cyclooctene case, the ee of the product is generally higher under
comparable conditions and is more sensitive to both temperature
and solvent polarity. These features, which are characteristic
for cycloheptene, are totally compatible with the faster quench-
ing rate and the highly solvent polarity-dependent exciplex
emission observed in fluorescence quenching experiments. The
smaller ring size of 1Z enables a faster quenching of the chiral
sensitizer and the subsequent formation of a more intimately
interacting exciplex which has greater charge-transfer character,
eventually affording the highest ee values of 70-77% using a
conventional chiral sensitizer in nonpolar solvents at -70 to
-80 °C. This rather simple strategy for the enhancement of the
sensitizer-substrate interaction through a reduction in the steric
hindrance is likely not to be restricted to the enantiodifferen-
tiating photosensitization of cycloalkenes but should be
widely applicable to the general asymmetric photosensitization
systems.

Trapped Optically ActiVe (E)-Cycloheptene
J. Am. Chem. Soc., Vol. 121, No. 46, 1999 10709
resolved, showing a coupling of 12.7 Hz, which is characteristic of the
trans protons. The 13C NMR gives further support to the proposed
structure. 3: m/z (70 eV) 270 (M+ - cycloheptene, retro-Diels-Alder
Although there are some dissimilarities between the asym-
metric photosensitizations of cycloheptene and cyclooctene (as
discussed above), it is crucial to point out similarities between
the two systems in order to give an overall picture. Thus, the
same chiral sensitizer affords the (E)-isomer with the same
absolute configuration for both cycloheptene and cyclooctene
at low temperature, where the entropic contribution is mini-
mized. The present asymmetric photosensitizing system is yet
another example of the temperature-switching behavior of the
chirality of the product, which we originally reported for
cyclooctene. The different values for the ee and activation
parameters obtained for the two cycloalkenes are consolidated
by the global fit to the single enthalpy-entropy compensation
plot, indicating that the same mechanism is operating in both
enantiodifferentiating photoisomerization systems.
1
reaction, 100), 207 (50); H NMR (400 MHz, CDCl3) δ (ppm) 0.75
(m, 1 H), 1.11 (m, 1 H), 1.4-1.7 (m, 7 H), 1.96 (ddd, J ) 3.4/5.9/12.7
Hz, 1 H), 2.04 (m, 1 H), 2.52 (ddd, J ) 3.4/5.9/12.7 Hz, 1 H), 7.15-
7.25 (m, 4 H), 7.36 (ddd, J ) <1/6.8/7.8 Hz, 1 H), 7.42 (ddd, J )
<1/6.8/7.8 Hz, 1 H), 7.44-7.54 (m, 4 H), 7.61 (dd, J ) <1/7.8 Hz, 2
H), 7.76 (dd, J ) <1/7.8 Hz, 2 H); 13C NMR (100 MHz, CDCl3) δ
(ppm) 24.64, 28.06, 28.36, 28.81, 29.36, 51.06, 52.04, 89.10, 90.04,
117.11, 121.41, 125.84, 125.98, 126.61, 126.83, 127.15, 127.96, 128.13,
128.30, 129.64, 129.80, 130.33, 132.97, 137.41, 138.29, 144.48, 150.18.
Trapping of 1E by Osmium Tetraoxide:16 Formation of trans-
1,2-Cycloheptanediol (4) (Accompanied by the cis-Isomer). Except
for a change of solvent to diethyl ether, the same irradiation conditions
as above were employed in the preparation of 4. After the irradiation,
the solutions were poured into a cooled flask containing OsO4 (130
mg, 0.51 mmol) and 0.2 mL of pyridine in diethyl ether (10 mL). The
resulting mixture was allowed to warm overnight to room temperature.
The brown precipitate formed was collected and refluxed for 3 h with
0.7 g of Na2SO3 in an ethanol-water mixture. The resultant mixture
was poured into water and extracted with ether (4 × 40 mL), and the
combined ether extracts were dried over MgSO4. GC analysis indicated
that the crude mixture obtained after evaporation of the solvent
contained cis- and trans-1,2-cycloheptanediol with a combined purity
of 96%. The isomeric diols showed Rf values of 0.33 and 0.38 upon
TLC analysis over silica gel with ethyl acetate as the eluent. The cis-
and trans-isomers were separated by column chromatography on silica
gel. The cis-diol (13.5 mg, mp 46 °C (lit.17 mp 48 °C), 21% yield
referred to OsO4) was shown to be pure by NMR, whereas the isolated
trans-diol 4 (5.6 mg) was still contaminated with the cis-isomer (cis:
trans ) 1:4). The 1H and 13C NMR spectra of the trans-diol 4 prepared
photochemically was in good agreement with those of the authentic
sample prepared independently by the H2O2 oxidation of 1Z.18 cis-
Cycloheptanediol: 1H NMR (400 MHz, C6D6) δ (ppm) 1.22 (m, 1 H),
1.44 (m, 1 H), 1.53 (m, 1 H), 1.68 (m, 1 H), 1.83 (m, 1 H), 2.2-3.3
(s, broad, 1 H, OH), 3.72 (m, 1 H, CHOH); 13C NMR (100 MHz, C6D6)
δ (ppm) 22.10 (2 C), 28.00 (1 C, C-5), 31.11 (2 C), 73.91 (2 C, CHOH).
trans-Cycloheptanediol 4: 1H NMR (400 MHz, C6D6) δ (ppm) 1.27
(m, 2 H), 1.45 (m, 2 H), 1.82 (m, 1 H), 2.9-3.1 (s, broad, 1 H, OH),
3.40 (m, 1 H, CHOH); 13C NMR (400 MHz, C6D6) δ (ppm) 22.50 (2
C), 26.80 (1 C, C-5), 32.82 (2 C), 78.02 (2 C, CHOH).
Finally, we wish to emphasize that this and previous studies
concerning asymmetric photosensitization are gradually reveal-
ing a detailed mechanism and factors controlling the enantio-
differentiating step, as well as the chirality and ee of the product,
and future work must ultimately lead us to a maximum optical
yield.
Experimental Section
General. Melting points were measured with a Yanaco MP-500-D
and are uncorrected. Optical rotations were determined by using a
Perkin-Elmer Polarimeter 341 with a thermostated 10 cm cell. Mass
spectra were recorded by using a GC/MS instrument, HP-5890A/JEOL
JMS-DX-303. 1H and 13C NMR spectra were measured at 400 and 100
MHz, respectively, in CDCl3 or C6D6, using a JEOL EX400 spectrom-
eter. Absorption spectra were obtained on a JASCO V-560 spectro-
photometer, and fluorescence spectra were recorded on a Hitachi F-4500
spectrofluorimeter. GC analyses were performed on a Shimadzu CBP-1
(25 m × 0.25 mm i.d.) column or a chiral Supelco â-DEX 120 (30 m
× 0.25 mm i.d.) column using a Shimadzu GC-14A instrument fitted
with a C-R6A integrator. Preparative GC was run on an SE-30 (1 m)
column using a GL Sciences GC instrument. HPLC analyses were
performed on a Wakosil 5-SIL (4.6 × 150 mm) column or a Daicel
Chiralpak AD (4.6 × 250 mm) column using a JASCO HPLC system
fitted with UV (254 nm) and RI detectors.
Solvents were dried (over CaH2, Mg, or KOH) and distilled under
an argon atmosphere prior to use. Cycloheptene was fractionally
distilled. The sensitizers were synthesized according to a literature
procedure from 1,2,4,5-benzenetetracarboxylic chloride and the com-
mercially available optically active alcohols.14 The trapping agents OsO4
and 1,3-diphenylisobenzofuran are commercially available and were
used without further purification.
Time Dependence of the Formation of 1E. A dichloromethane
solution (30 mL), containing 1Z (35 µL, 0.30 mmol, 9.9 mM) and 5
(49.2 mg, 0.061 mmol, 2.03 mM) or 6 (48.0 mg, 0.059 mmol, 1.96
mM), was prepared. After this stock solution was distributed into six
quartz tubes (3 mL each), each solution was irradiated at -70 °C
under an argon atmosphere with a 300-W high-pressure mercury
lamp for a designated time of 5, 15, 25, 40, 60, or 90 min, and the
irradiated solution was poured into a cooled dichloromethane solution
(0.5 mL) containing 0.43 mg of dodecane, which was used as an internal
standard for the GC analysis, and 1.18 mg (0.0044 mmol) of the
trapping agent 2. The resultant mixture was kept at -70 °C and
gradually warmed to room temperature overnight. This was then
analyzed by GC. For normalization, the starting concentration of 1Z
relative to dodecane was determined separately by an external standard
method.
Standard Procedure for the Analytical-Scale Irradiation. A
solution (50 mL), containing 1Z (10 mM) and chiral sensitizer 5, 6, or
7 (2 mM), was distributed into five quartz irradiation tubes (10 mm
outside diameter). All tubes were sealed under argon with a septum
cap and were fixed near a 300-W high-pressure mercury lamp at a
distance of ca. 1 mm from the lamp. The whole system was immersed
in a cooling bath filled with methanol, which was stabilized at a desired
temperature below -40 °C, after which the solutions were irradiated
Preparative-Scale Irradiation and Trapping of 1E: Formation
of (5RS,5aRS,10aRS,11SR)-5,11-Epoxy-5a,6,7,8,9,10,10a,11-octahy-
dro-5,11-diphenyl-5H-cyclohepta[b]naphthalene (3). Ten quartz tubes
(10 mm outside diameter) containing a deaerated dichloromethane
solution (100 mL in total) of 1Z (192 mg, 2 mmol) and methyl benzoate
(136 mg, 1 mmol) were irradiated for 1 h at -78 °C with a 300-W
high-pressure mercury lamp. The irradiated solutions were poured into
a precooled flask containing 118 mg (0.44 mmol) of 1,3-diphenyl-
isobenzofuran. The resultant solution was subsequently allowed to warm
to room temperature over a period of several hours. After the removal
of the solvent and other low-boiling components from the mixture under
reduced pressure, the residue was purified on a silica gel column, using
a cyclohexanes-ethyl acetate (6:1) eluent, to give 47.2 mg (6% yield
based on 1Z used and 29% based on 1,3-diphenylisobenzofuran) of
the Diels-Alder adduct 3 as a colorless solid, mp 148 °C. The TLC
(eluent, Rf ) 0.8) was developed with an oxidizing molybdenum/cerium
solution. The 1H NMR spectrum of this compound is in good agreement
with the previous report.15 However, our spectrum is much more
(15) Coates, R. M.; Last, L. A. J. Am. Chem. Soc. 1983, 105, 7322.
(16) Leitlich, J. Tetrahedron Lett. 1978, 38, 3589.
(14) (a) Inoue, Y.; Yamasaki, N.; Yokoyama, T.; Tai, A.; Ishida, A.;
Takamuku, S. J. Chem. Soc., Chem. Commun. 1989, 1270. (b) Yamasaki,
N.; Inoue, Y.; Yokoyama, T.; Tai, A.; Ishida, A.; Takamuku, S. J. Am.
Chem. Soc. 1991, 113, 1933.
(17) Criegee, R.; Marchand, B.; Wannowius, H. Liebigs Ann. Chem.
1942, 550, 99 and 128.
(18) AutorenkollektiV Organikum, 16th ed.; VEB Deutscher Verlag der
Wissenschaften: Dresden, 1985; pp 258-289.

10710 J. Am. Chem. Soc., Vol. 121, No. 46, 1999
Hoffmann and Inoue
for 45-90 min. To quantitatively trap the 1E produced, the irradiated
solutions were cooled to at least -70 °C and then poured into a cooled
flask containing ca. 14 mg (0.05 mmol) of 1,3-diphenylisobenzofuran.
When a solvent other than dichloromethane was used for the photolysis,
the irradiated solution was poured into a precooled solution of the
trapping agent in dichloromethane (50 mL).
HPLC column. From repeated injections, it was shown that a racemic
mixture of 3 gave an average ee of 0.5% in favor of the (S,S)-isomer,
which eluted as the second fraction.
Alternatively, 1E could be trapped by OsO4 oxidation. After the
oxidation and the subsequent workup described above, the product was
analyzed as follows. The cis-diol, which always accompanied as the
major product from 1Z, was separated from the trans-diol 4 by means
of HPLC on a Wakosil 5-SIL column. The pure 4 obtained was then
analyzed by chiral GC on a Supelco â-DEX 120 column. Direct chiral
GC analysis of the oxidized mixture was not possible on the same chiral
GC column, as a result of an overlap of the peaks of the cis- and trans-
diols.
Upon completion of the thermal cycloaddition reaction, the resultant
mixture containing residual trapping agent was exposed to sunlight until
the color of diphenylisobenzofuran disappeared.19 The solvent was then
evaporated under reduced pressure, and the oily residue was separated
on a GPC column (Jaigel 1-H and 2-H, Japan Analytical Industry Co.).
The cycloadduct 3 was further purified by an additional small-scale
liquid chromatography, affording a very pure sample suitable for chiral
HPLC (Chiralpak AD, 0.46 × 250 mm). Usually, the enantiomeric
excess of 3 was determined as an average of five injections to the chiral
Acknowledgment. We are grateful to Dr. Simon Everitt for
suggestions and alterations arising from a preliminary reading
of this manuscript.
(19) Sasaki, T.; Kanematsu, K.; Kayakawa, K.; Sugiura, M. J. Am. Chem.
Soc. 1975, 97, 355.
JA991315L