Solvent and temperature effects on diastereodifferentiating paterno-buechi reaction of chiral alkyl cyanobenzoates with diphenylethene upon direct versus charge-transfer excitation

Article abstract of DOI:10.1021/jo101332x

In the Paterno-Buechi reaction of chiral p-cyanobenzoates (1) with 1,1-diphenylethene (2), we revealed that the excited charge-transfer (CT) complex formed upon selective excitation at the CT band is distinctly different in structure and reactivity from the conventional exciplex generated through the direct excitation of acceptor 1 which subsequently associates with donor 2. Thus, the favored diastereoface upon photocycloaddition, as well as the temperature- and solvent-dependent behavior of the product's diastereoselectivity, were highly contrasting, often opposite, to each other upon direct versus CT excitation. From the activation parameters obtained by the Eyring analyses of the diastereoselectivity, we are able to infer that the conventional exciplex is relatively flexible and susceptible to the environmental variants, whereas the CT complex is better π-π stacked and more rigid in the ground state and also in the excited state, leading to the significantly smaller differential activation enthalpies and entropies. More interestingly, the signs of the differential activation parameters determined for direct and CT excitation are consistently opposite to each other and the isokinetic temperatures calculated therefrom differ significantly, unambiguously revealing the distinctly different nature in structure and reactivity of these two excited-state complex species. Thus, the combined use of irradiation wavelength, temperature, and solvent provides us with a convenient, powerful tool not only for elucidating the mechanistic details of photoreaction but also for critically controlling the stereochemical outcomes of photochirogenic reaction.

Full text of DOI:10.1021/jo101332x

pubs.acs.org/joc
Solvent and Temperature Effects on Diastereodifferentiating
ꢀ
€
Paterno-Buchi Reaction of Chiral Alkyl Cyanobenzoates with
Diphenylethene upon Direct versus Charge-Transfer Excitation
Kazuyuki Matsumura, Tadashi Mori,* and Yoshihisa Inoue*
Department of Applied Chemistry, Osaka University, 2-1 Yamada-oka, Suita 565-0871, Japan
tmori@chem.eng.osaka-u.ac.jp
Received July 6, 2010
ꢀ
€
In the Paterno-Buchi reaction of chiral p-cyanobenzoates (1) with 1,1-diphenylethene (2), we revealed
that the excited charge-transfer (CT) complex formed upon selective excitation at the CT band is distinctly
different in structure and reactivity from the conventional exciplex generated through the direct excitation
of acceptor 1 which subsequently associates with donor 2. Thus, the favored diastereoface upon
photocycloaddition, as well as the temperature- and solvent-dependent behavior of the product’s
diastereoselectivity, were highly contrasting, often opposite, to each other upon direct versus CT excitation.
From the activation parameters obtained by the Eyring analyses of the diastereoselectivity, we are able to
infer that the conventional exciplex is relatively flexible and susceptible to the environmental variants,
whereas the CT complex is better π-π stacked and more rigid in the ground state and also in the excited
state, leading to the significantly smaller differential activation enthalpies and entropies. More interestingly,
the signs of the differential activation parameters determined for direct and CT excitation are consistently
opposite to each other and the isokinetic temperatures calculated therefrom differ significantly, unam-
biguously revealing the distinctly different nature in structure and reactivity of these two excited-state
complex species. Thus, the combined use of irradiation wavelength, temperature, and solvent provides us
with a convenient, powerful tool not only for elucidating the mechanistic details of photoreaction but also
for critically controlling the stereochemical outcomes of photochirogenic reaction.
Introduction
viewpoints,2 as the photochemical route to oxetane is more
advantageous in several aspects than conventional thermal
1
ꢀ
€
The Paterno-Buchi reaction, i.e., [2 þ 2] photocycloaddi-
tion of a ketone to an alkene leading to oxetane formation, has
attracted much attention from the mechanistic and synthetic
3
ꢀ
€
ones. Mechanistic studies on Paterno-Buchi reaction were
carried out in considerable detail to reveal the vital role of tem-
perature in determining the product selectivity.4 The knowl-
edge accumulated through these studies provides the crucial
clues to effectively control the distribution and stereochemistry
ꢁ
(1) (a) Paterno, E.; Chieffi, G. Gazz. Chim. Ital. 1909, 39, 341–361. (b)
Buchi, G.; Inman, C. G.; Lipinsky, E. S. J. Am. Chem. Soc. 1954, 76, 4327–
€
ꢀ
of oxetane products. More recently, asymmetric Paterno-
€
Buchi reactions of chirally modified substrates were also studied
4331.
(2) (a) D’Auria, M.; Racioppi, R. Eur. J. Org. Chem. 2010, 3831–3826.
(b) Abe, M. J. Chin. Chem. Soc. 2008, 55, 479–486. (c) D’Auria, M.;
Emanuele, L.; Racioppi, R. Adv. Photochem. 2005, 28, 81–127. (d) Griesbeck,
A. G. Synlett 2003, 451–472. (e) Vargas, F.; Rivas, C. Int. J. Photoenergy
2000, 2, 97–101. (f) Bach, T. Synlett 2000, 1699–1707. (g) Bach, T. Synthesis
1998, 683–703. (h) Jones, G. Org. Photochem. 1981, 5, 1–122. (i) Horspool,
W. M. Photolysis of Carbonyl Compounds, in Specialist Periodical Reports,
Photochemistry, vol. 11, 283-300. (Ed. Bryce-Smith, D.) RSC Publishing, UK
(1981). (j) Arnold, D. R. Adv. Photochem. 1968, 6, 301–423.
(3) (a) Hoffmann, N. Chem. Rev. 2008, 108, 1052–1103. (b) Jimenez,
M. C.; Miranda, M. A. Oxygen-containing functions, in Specialist Periodical
Reports, Photochemistry, vol. 37, 149-174. (Ed. Albini, A.) RSC Publish-
ing, UK (2009). (c) See also: Roth, H. D. Angew. Chem., Int. Ed. Engl. 1989,
28, 1193–1207.
(4) Buschmann, H.; Scharf, H.-D.; Hoffmann, N.; Esser, P. Angew.
Chem., Int. Ed. Engl. 1991, 30, 477–515.
DOI: 10.1021/jo101332x
r
Published on Web 07/27/2010
J. Org. Chem. 2010, 75, 5461–5469 5461
2010 American Chemical Society

Matsumura et al.
JOCFeatured Article
extensively to find that a wide variety of chiral oxetanes were
produced in generally high diastereoselectivities.5 A consider-
able amount of effort has been devoted to the manipulation of
SCHEME 1. Diastereodifferentiating [2 þ 2] Photocycloaddi-
tion of Chiral Alkyl Benzoates 1 to 1,1-Diphenylethene 2 upon
Direct and Charge-Transfer Band Excitation
ꢀ
€
regio- and diastereoselectivities in asymmetric Paterno-Buchi
reactions,6 and in particular, the use of allylic strain and
hydrogen-bonding interaction were demonstrated to be highly
effective in controlling the stereochemical consequence of the
diastereodiffeentiating [2 þ 2] photoreaction.7 However, the
ꢀ
€
precise control of the face selectivity of Paterno-Buchi reac-
tion is still a challenging task. The excited-state dynamics of a
donor-acceptor system can be manipulated by changing the
excitation wavelength, but the focus is rather on the photo-
physical aspects in the literature,8 and the importance of excita-
small or controversial, except for the clear excitation-wavelength
dependence observed in the diastereodifferentiating [2 þ 2]
photocycloaddition of stilbene to chiral fumarate.12
ꢀ
€
tion wavelength on Paterno-Buchi reaction has not been fully
recognized or examined in detail.9
In this study, we closely investigated the solvent and tempera-
ture effects on the diastereodifferentiating photocycloaddition
of chiral alkyl cyanobenzoates 1 to diphenylethene 2 to elucidate
the nature and the difference of two excited-state complex spe-
cies generated upon direct and CT excitation. Interestingly, the
CT, rather than direct, excitation led to lower diastereoselec-
tivities in general, which is in keen contrast to the photocycload-
dition of stilbene to chiral fumarate, where the CT excitation
consistently affords better diastereoselectivities. Although the
photoreaction of stilbene with fumarate is not very clean
(accompanying the E-Z isomerization and dimerizaiton of
stilbene), the reaction of 1 with 2 exclusively affords oxetane 3,
allowing more detailed examinations of the product selectivities.
The differential activation parameters obtained by the Eyring
analysis of the diastereoselectivities of 3 obtained under a variety
of conditions provide us with rich insights into the photocy-
cloaddition mechanism and the nature of excited-state inter-
mediates, which will be discussed below.
We have recently investigated the temperature dependence
behavior of the diastereodifferentiating [2 þ 2] photocycloaddi-
tion of chiral alkyl benzoates (1) to 1,1-diphenylethene (2)
(Scheme 1).10 The stereochemical outcome was critically
affected by the mode of excitation, and the direct excitation of
acceptor 1 and the selective excitation at the charge-transfer
(CT) band gave the same oxetane but in entirely different dias-
tereoselectivities, indicating the existence of two distinct excited-
state complex species that are not equilibrated each other. The
CT-band excitation of he ground-state complex has been a tar-
get of intensive studies,11 but the observed difference in photo-
chemical outcome upon direct versus CT excitation was rather
(5) (a) D’Auria, M.; Racioppi, R. Cur. Org. Chem. 2009, 13, 939–954. (b)
Demuth, M.; Mikhail, G. Synthesis 1989, 145–162. (b) See also: Gotthardt,
H.; Lenz, W. Angew. Chem., Int. Ed. Engl. 1979, 18, 868.
(6) (a) Adam, W.; Stegmann, W. J. Am. Chem. Soc. 2002, 124, 3600–3607.
€
(b) Adam, W.; Stegmann, W.; Weikotz, S. J. Am. Chem. Soc. 2001, 123,
2452–2453. (c) Hoffmann, N.; Buschmann, H.; Raabe, G.; Scharf, H. D.
Tetrahedron 1994, 50, 11167–11186. (d) Buschmann, H.; Hoffmann, N.;
Scharf, H. D. Tetrahedron: Asymmetry 1991, 2, 1429–1444. (e) Buschmann,
H.; Scharf, H.-D.; Hoffmann, N.; Plath, M. W.; Runsink, J. J. Am. Chem.
Soc. 1989, 111, 5367–5373. (f) Pelzer, R.; Juetten, P.; Scharf, H. D. Chem.
Ber. 1989, 122, 487–491. (g) Hoffmann, N.; Scharf, H.-D.; Runsinkt, J.
Tetrahedron Lett. 1989, 30, 2637–2638. (h) Runsink, J.; Koch, H.; Nehrings,
A.; Scharf, H. D.; Nowack, E.; Hahn, T. J. Chem. Soc., Perkin Trans. 2 1988,
49–55. (i) Herzog, H.; Koch, H.; Scharf, H.-D.; Runsinkt, J. Tetrahedron
1986, 42, 3547–3558. (j) Koch, H.; Runsink, J.; Scharf, H. D. Tetrahedron
Lett. 1983, 24, 3217–3220. (k) Jarosz, S.; Zamojski, A. Tetrahedron 1982, 38,
1453–1456.
Results and Discussion
Ground-State Interaction of Cyanobenzoate 1a with Diphe-
nylethene 2. It was reported that despite no clear indication of
the ground-state interaction between electron-deficient aro-
matic esters, such as dimethyl tere- and isophthalate, and
several olefinic donors in the UV spectra, these donor-accep-
ꢀ
€
tor pairs smoothly underwent the Paterno-Buchi reaction
to give the corresponding oxetanes upon photoexcitation.13
Because such interactions are elusive in general, it is likely that
the failure to detect the ground-state interaction was due to the
insufficient formation of the CT complex under the experi-
mental conditions employed. Indeed, we were able to observe
weak but appreciable UV spectral changes only by mixing
donor 2 and acceptor 1a at relatively high concentrations
(∼0.2 M). The new absorption band, visible at the red edge of
the original spectrum, can be ascribed to the formation of
donor-acceptor or charge-transfer complex. Association
constants (KCT) were determined in a variety of solvents by
using the modified Benesi-Hildebrand method; the original
Benesi-Hildebrand treatment14 was avoided in the present
case, as the method is not suitable for accurately determining a
€
(7) (a) Bach, T.; Jodicke, K.; Kather, K.; Frohlich, R. J. Am. Chem. Soc.
1997, 119, 2437–2445. (b) Bach, T.; Jodicke, K.; Kather, K.; Hecht, J. Angew.
€
€
Chem., Int. Ed. Engl. 1995, 34, 2271–2273. (c) Griesbeck, A. G.; Bondock, S.
J. Am. Chem. Soc. 2001, 123, 6191–6192. (d) Adam, W.; Stegmann, V. R.
Synthesis 2001, 1203–1214. (e) Adam, W.; Peters, K.; Peters, E. M.;
Stegmann, V. R. J. Am. Chem. Soc. 2000, 122, 2958–2959. (f) Bach, T.;
Bergmann, H.; Harms, K. J. Am. Chem. Soc. 1999, 121, 10650–10651. (g)
Abe, M.; Kawakami, T.; Ohata, S.; Nozaki, K.; Nojima, M. J. Am. Chem.
Soc. 2004, 126, 2838–2846. (h) Abe, M.; Fujimoto, K.; Nojima, M. J. Am.
Chem. Soc. 2000, 122, 4005–4010.
ꢁ
(8) (a) Nicolet, O.; Banerji, N.; Pages, S.; Vauthey, E. J. Phys. Chem. A
2005, 109, 8236–8245. (b) Levy, D.; Arnold, B. R. J. Phys. Chem. A 2005, 109,
2113–2119. (c) See also: Mohammed, O. F.; Vauthey, E. J. Phys. Chem. A
2008, 112, 3823–3830. (d) Wang, Y.; Haze, O.; Dinnocenzo, J. P.; Farid, S.;
Farid, R. S.; Gould, I. R. J. Org. Chem. 2007, 72, 6970–6981.
(9) Gotthardt, H.; Lenz, W. Tetrahedron Lett. 1979, 20, 2879–2880.
(10) Matsumura, K.; Mori, T.; Inoue, Y. J. Am. Chem. Soc. 2009, 131,
17076–17077.
(11) (a) Gonzalez-Bejar, M.; Stiriba, S.-E.; Miranda, M. A.; Perez-Prieto,
J. Org. Lett. 2007, 9, 453–456. (b) Haga, N.; Takayanagi, H.; Tokumaru, K.
J. Chem. Soc., Perkin Trans. 2 2002, 734–745. (c) Sun, D.; Hubig, S. M.;
Kochi, J. K. J. Org. Chem. 1999, 64, 2250–2258. (d) Hubig, S. M.; Sun, D.;
Kochi, J. K. J. Chem. Soc., Perkin Trans. 2 1999, 781–788. (e) Bosch, E.;
Hubig, S. M.; Lindeman, S. V.; Kochi, J. K. J. Org. Chem. 1998, 63, 592–601.
(f) Haga, N.; Nakajima, H.; Takayanagi, H.; Tokumaru, K. J. Org. Chem.
1998, 63, 5372–5384. (g) Takahashi, Y.; Ohaku, H.; Nishioka, N.; Ikeda, H.;
Miyashi, T.; Gormin, D. A.; Hilinski, E. F. J. Chem. Soc., Perkin Trans. 2
1997, 303–308. (h) Haga, N.; Nakajima, H.; Takayanagi, H.; Tokumaru, K.
Chem. Commun. 1997, 1171–1172.
(12) (a) Saito, H.; Mori, T.; Wada, T.; Inoue, Y. J. Am. Chem. Soc. 2004,
126, 1900–1906. (b) Saito, H.; Mori, T.; Wada, T.; Inoue, Y. Org. Lett. 2006,
8, 1909–1912.
(13) (a) Shigemitsu, Y.; Nakai, H.; Odaira, Y. Tetrahedron 1969, 25,
3039–3043. (b) Shigemitsu, Y.; Katsuhara, Y.; Odaira, Y. Tetrahedron Lett.
1971, 12, 2887–2890. (c) Kateuhara, Y.; Shigemitsu, Y.; Odaira, Y. Bull.
Chem. Soc. Jpn. 1971, 44, 1169.
(14) Benesi, H. A.; Hildebrand, J. H. J. Am. Chem. Soc. 1949, 71, 2703–
2707.
5462 J. Org. Chem. Vol. 75, No. 16, 2010

Matsumura et al.
JOCFeatured Article
TABLE 1. Association Constants (KCT) of Charge-Transfer Complex
1a has a chiral handle. The electronic CD is frequently used
as a tool for analyzing the conformational behavior of chiral
molecules in solution.20 Thus, the CD spectral changes of a
methylcyclohexane solution of 1a upon addition of 2 were
examined. Although no appreciable induced CD associated
with the complexation developed at 25 °C under the same
conditions employed for the photoreaction, a fairly weak but
appreciable negative Cotton effect was observed for the CT
band at around 320-330 nm decreasing the temperature down
to -25 °C (Figure S2, Supporting Information). Note that
cyanobenzoate 1a and the related (S,S)-bis(2-methylpropyl)
terephthalate were not fluorescent at 25 °C in the presence and
absence of donor 2; nevertheless, achiral methyl terephthalate
was subjected to the fluorescence study.13
Diastereodifferentiating [2 þ 2] Photocycloaddition of 1a to 2
upon Direct versus CT Excitation. Photoirradiation of a
concentrated solution of 1a and 2 (0.2 and 1.0 M, respectively)
afforded diastereomeric oxetane 3a in moderate to good com-
bined yields in all of the solvents examined (Table 2). The pho-
toreaction was essentially very clean, and only small amounts of
minor byproduct were detected by HPLC analysis; therefore,
the low material balance and modest chemical yield would be
ascribed at least in part to the less accurate evaluation of
consumed 1a in particular at low conversions. The diastereo-
meric excess (de) of photoadduct 3a was determined by HPLC
with satisfactory accuracy (error (3%), and the effects of
irradiation wavelength, temperature, and solvent on the de of
3a obtained upon photoirradiation of a mixture of 1a and 2
were systematically examined.
(a) Effect of Irradiation Wavelength. The effect of irradia-
tion wavelength on the product’s de was examined with a
toluene solution of 1a and 2 at low conversions.10 The de of
3a was kept almost constant upon irradiation at 254-290 nm
but gradually decreased by shifting the irradiation wave-
length to 310-320 nm, and eventually the epimeric product
was obtained upon excitation at 330 nm. When irradiated at
340 nm, the starting material was recovered even after pro-
longed irradiations. We initially investigated the wavelength
effect on the diastereoselectivity of 3a at 254 nm (from a low-
pressure mercury lamp) and at 313 nm (from a high-pressure
mercury lamp fitted with a Pyrex filter), and the temperature-
dependent de values obtained at each wavelength were
subjected to the Eyring analysis by plotting the relative rate
constant forming diastereomeric 3a as a function of recipro-
cal temperature. The de values obtained upon excitation at
254 nm gave a good linear Eyring plot over the entire tem-
perature region employed. However, the de data obtained at
313 nm led to a curved plot (Figure S3, Supporting Infor-
mation), revealing operation of two distinct diastereodiffer-
entiation mechanisms that are switched at the middle of the
temperature range employed (ca. 15 °C). The Eyring plot for
the de’s obtained at 330 nm became normal again, giving a
single straight line of oppositely signed slope without show-
ing any curvature over the entire temperature range. Hence,
the spectral overlap between 1a and the CT complex is likely
to be responsible for the curved Eyring plot and the mixed
diastereodifferentiation mechanism observed at 313 nm. We
therefore chose the excitation wavelengths of 290 and 330 nm
of 1a with 2 in Some Solventsa
dipole moment dielectric
(μ)/debye
solvent
constant/ε λedge/nmb KCT/M-1
methylcyclohexane
toluene
tetrahydrofuran
acetonitrile
0
2.02
2.38
7.39
337
343
328
338
0.04 (0.03)c
0.8
0.02
0.37
1.68
3.94
37.5
0.2
aDetermined by the modified Benesi-Hildebrand method for solu-
tions of [1a] = [2] = 0.2-0.07 M at 25 °C. For data in methylcyclohex-
ane, see ref. 10. bEdge of the absorption spectrum of CT complex
(defined as the wavelength at which the absorbance becomes 0.005),
which was obtained by subtracting the spectral sum of 1a and 2 from the
observed spectrum. cAt 50 °C.
low K value. Thus, we recorded the UV spectral changes upon
simultaneous reduction of the donor and acceptor concentrations
from 0.2 M to 0.07 or 0.08 M, keeping the 1:1 stoichiometry. The
deviation of the observed spectrum from the spectral sum of each
component at a given concentration indicates the net absorption
of the CT complex, which appeared at 300-340 nm (Figure S1,
Supporting Information). The absorbance change in the CT band
region was plotted against the product of the donor and acceptor
concentrations,15 followed by the curve fitting procedure, to afford
the association constants listed in Table 1. The detailed discussion
on the evaluation of such low KCT values for weak CT complexes
may be found in the literature.16
The association constant for the 1a/2 pair is comparable to or
slightly lower than the typical values reported for the conven-
tional CT complexes formed between aromatic donors and
tetracyanoethene or chloranil.17 The relatively low KCT value
found for 1a/2 pair may be attributed at least in part to the steric
hindrance of the acceptor’s ester moiety. The KCT values are
known to be significantly affected by the steric inhibition bet-
ween donor and acceptor, in addition to the electrostatic and
stacking interactions.18 The KCT value became considerably
larger in polar acetonitrile than in less polar methylcyclohexane
or tetrahydrofuran, indicating that the charge delocalization
within the complex plays some role to stabilize the complex. The
much larger KCT value in toluene may be attributed to the
effective stabilization of the complex through the π-stacking
interaction with toluene molecules. The complexation thermo-
dynamic parameters could not be determined, as the difference
in KCT determined at 25 and 50 °C (in methylcyclohexane) was
so small to give a reliable van’t Hoff plot. Due to the low KCT
values in all the solvents, relatively high donor/acceptor con-
centrations were used to keep a satisfactory amount of the com-
plex in the solution. However, by irradiating the CT band at 330
nm through a band-path filter (with a xenon light source), the
selective excitation of the CT complex was possible (vide infra).
The ground-state interaction was further examined by
circular dichroism (CD) spectroscopy,19 as cyanobenzoate
(15) Arnold, B. R.; Euler, A.; Fields, K.; Zaini, R. Y. J. Phys. Org. Chem.
2000, 13, 729–734.
(16) (a) Zaini, R.; Orcutt, A. C.; Arnold, B. R. Photochem. Photobiol.
1999, 69, 443–447. (b) Exner, O. Chemom. Intell. Lab. Syst. 1997, 39, 85–93.
(17) (a) Frey, J. E.; Andrews, A. M.; Ankoviac, D. G.; Beaman, D. N.; Du
Pont, L. E.; Elsner, T. E.; Lang, S. R.; Zwart, M. O.; Seagle, R. E.; Torreano,
L. A. J. Org. Chem. 1990, 55, 606–624. (b) Frey, J. E.; Aiello, T.; Beaman, D. N.;
Combs, S. D.; Fu, S.-L.; Puckett, J. J. J. Org. Chem. 1994, 59, 1817–1830. (c)
Frey, J. E.; Aiello, T.; Fu, S.-L.; Hutson, H. J. Org. Chem. 1996, 61, 295–300.
(18) Rathore, R.; Lindeman, S. V.; Kochi, J. K. J. Am. Chem. Soc. 1997,
119, 9393–9404.
(20) (a) Mori, T.; Inoue, Y.; Grimme, S. J. Org. Chem. 2006, 71, 9797–
9806. (b) Mori, T.; Inoue, Y.; Grimme, S. J. Phys. Chem. A 2007, 111, 7995–
8006. (c) Mori, T.; Inoue, Y.; Grimme, S. J. Phys. Chem. A 2007, 111, 4222–
4234.
(19) (a) Hembury, G. A.; Borovkov, V. V.; Inoue, Y. Chem. Rev. 2008,
108, 1–73. (b) Berova, N., Nakanishi, K., Woody, R. W. Circular Dichroism:
Principles and Applications, 2nd ed.; Wiley: New York, 2000.
J. Org. Chem. Vol. 75, No. 16, 2010 5463

Matsumura et al.
JOCFeatured Article
TABLE 2. Diastereodifferentiating [2 þ 2] Photocycloaddition of (S)-1-
Methylpropyl p-Cyanobenzoate 1a to 1,1-Diphenylethene 2 upon Direct
and Charge-Transfer Band Excitation in a Variety of Solventsa
direct and CT excitation, is thought to proceed in the singlet
manifold, as judged from the triplet quenching experiments
shown below. This view also agrees with the results of previous
studies on the photocycloaddition reactions of electron-deficient
excitation irrad temp/
modeb
%
%
%
solvent
time/h °C convc yieldd dee
aromatic esters with olefinic donors.13 Thus, the Paterno-Buchi
reaction of 1a and 2 upon direct and CT excitation was exa-
mined in the presence of (E)-1,3-pentadiene (0.2-0.4 M), whose
triplet energy, ET=248 kJ mol-1, is reasonably lower than that
ꢀ
€
methylcyclohexane
direct
3
10
3
50
25
0
-25
-50
50
25
0
-25
-50
50
25
0
-25
-50
50
25
0
-25
-50
50
25
-50
50
25
-50
50
31
31
33
32
32
20
23
31
26
26
18
25
18
26
12
12
11
11
9
5
4
4
4
77
64
63
43
34
3
4
3
24
26
74
65
68
54
37
-11
-4
4
6
21
57
54
32
of methyl p-cyanobenzoate, ET=301 kJ mol-1 22
. The addition
CT
of diene lowered the conversion but enhanced the de in both
excitation modes (Table S1, Supporting Information). On the
contrary, the yield and de of 3a obtained upon direct and CT
excitation were almost unaffected (within the experimental
error) in the presence of molecular oxygen added as a triplet
quencher,23 suggesting that the triplet species are not the major
species in the present photoreaction.10 As the Eyring treatment
of the de data always afford good linear plots (vide infra), a
single diastereodifferentiation mechanism should be operative
under the reaction conditions employed. We may conclude,
therefore, that 1,3-pentadiene behaves not only as a triplet
quencher but also acts as a chemical quencher. It seems that
the singlet manifold of the cyanobenzoate is also quenched by
this diene presumably via energy or electron transfer process.
Accordingly, the intervention of triplet species in the overall
mechanism and its effect on de in the reaction of 1 with 2 are
insignificant under our experimental conditions.
(c) Temperature and Solvent Effects on the Photoreaction.
For better comparison of the direct versus CT excitation and
also the diastereodifferentiation mechanisms operative in
both excitation modes, we performed the photoreaction of
1a and 2 at various temperatures ranging from þ50 to -50
°C to obtain the results shown in Table 2. Attempts to further
lower the reaction temperature down to -70 °C in toluene
resulted in extremely low conversion and yield (3a in <2%)
after 6 h irradiation at 330 nm, for which the insufficient
solubility of 1 or 2 seems to be responsible as the sample
solution was not completely transparent at this temperature.
As can be seen from Table 2, the direct, rather than CT,
excitation affords much higher de’s particularly in nonpolar
solvents, and the temperature-dependence profiles of de are
just opposite in the direct and CT excitation.
The solvent effects on the photoreaction of 1a with 2 were
also studied in methylcyclohexane, toluene, tetrahydrofur-
an, and acetonitrile (Table 2). By increasing the solvent
polarity, the yield of oxetane 3a was reduced in the direct
excitation but improved in the CT excitation. The conversion
and the yield of 3a obtained were remarkably insensitive to
the reaction temperature as far as the same excitation mode
and solvent were used. In methylcyclohexane, (10S,2R)-3a
was obtained as the major product irrespective of the excita-
tion mode and the reaction temperature, while the epimer,
(10S,2S)-3a, was favored particularly upon CT excitation in
polar solvents [the absolute configurations of diastereomers
of 3a were already established by X-ray crystallography,10
and the negative de value in Table 2 indicates the predomi-
nant formation of (10S,2S)-3a]. Thus, the (10S,2S)-isomer
was formed as the major product in the CT excitation at the
6
toluene
direct
CT
12
13
9
10
10
4
4
5
4
5
9
7
4
4
5
4
4
4
24
f
6
f
f
f
f
tetrahydrofuran
acetonitrile
direct
CT
3
5
3
12
18
8
13
17
24
20
25
22
23
23
27
26
26
14
-12
-8
12
direct
-13
-12
-11
-11
-3
-8
-10
-11
1
25
-20
-40
50
5
6
CT
7
3
11
15
15
14
3
25
-20
-40
50
2% acetonitrile in
methylcyclohexane
direct
25
-20
-40
50
25
-20
-40
22
22
22
10
6
3
4
4
4
3
3
4
-2
-4
-2
16
13
0
CT
8
21
20
3
a[1a] = 0.2 M, [2] = 1.0 M. For data in methylcyclohexane, see ref. 10.
bIrradiation was performed at 290 nm (direct excitation) or at 330 nm
(CT excitation) using an appropriate band-path filter fitted to the light
source LAX-101 (100 W xenon lamp). cConsumption of 1a determined by
HPLC. dHPLC yield of 3a. eThe absolute configurations of diastereomeric
3a were established by X-ray crystallography (ref 10), and the positive and
negative de values indicate the predominant formation of (10S,2R)- and
f
(10S,2S)-3a, respectively. Not determined.
for direct and selective CT excitation, respectively, through-
out the study. Although the possible contribution of hidden
CT bands at shorter wavelengths is difficult to strictly
exclude, the same de values obtained at 254 and 290 nm
indicate the insignificant contribution of the excited CT
complex in the direct excitation.
(b) Effect of Triplet Quencher on the Diastereoselectivity.
ꢀ
€
While typical Paterno-Buchi reactions of ketones with ole-
fins undergo via triplet biradical intermediates,21 the photo-
addition of cyanobenzoate 1 with diphenylethene 2, in both
(22) Kellogg, R. E.; Simpson, W. T. J. Am. Chem. Soc. 1965, 87, 4230–
4234.
(23) (a) Gorman, A. A.; Rodgers, M. A. J. J. Am. Chem. Soc. 1986, 108,
5074–5078. (b) Sajimon, M. C.; Ramaiah, D.; Thomas, K. G.; George, M. V.
J. Org. Chem. 2001, 66, 3182–3187.
(21) (a) Freilich, S. C.; Peters, K. S. J. Am. Chem. Soc. 1981, 103, 6257–
6259. (b) Freilich, S. C.; Peters, K. S. J. Am. Chem. Soc. 1985, 107, 3819–
3822. (c) Yang, N. C.; Loeschen, R. L.; Mitchell, D. J. Am. Chem. Soc. 1967,
89, 5465–5466.
5464 J. Org. Chem. Vol. 75, No. 16, 2010

Matsumura et al.
JOCFeatured Article
TABLE 3. Activation Parameters Determined by the Eyring Treatment of the Diastereomeric Excess (de) of 3a (or 3b) Obtained in Diastereodiffer-
entiating Photocycloaddition of 1a (or 1b) to 2 in Various Solventsa
excitation mode
direct
solvent
methylcyclohexane
ΔΔH‡/kJ mol-1
ΔΔS‡/J mol-1 K-1
TΔΔS‡a/kJ mol-1
T0/°Cb
þ6.4
(þ1.6
þ6.3
þ3.9
-0.2
þ0.4
-3.3
(-3.8
-3.6
-2.8
þ1.0
þ2.1
þ39
þ21
þ35
þ23
-2.7
þ1.2
-10
-9
-13
-11
þ2.4
þ9.1
þ12
-109
-197)c
-93
þ6.3
þ11
toluene
tetrahydrofurane
acetonitrile
2% acetonitrile in methylcyclohexane
methylcyclohexane
þ6.8
-0.8
þ0.3
-3.1
-2.7
-3.8
-3.2
þ0.7
þ2.7
-104
-199
þ60
CT
þ57
þ149)c
toluene
þ4
tetrahydrofurane
acetonitrile
2% acetonitrile in methylcyclohexane
-19
þ144
-42
aT = 298 K. For data in methylcyclohexane, see ref. 10. bEquipodal temperature. cFor the reactions of 1b with 2 to give 3b.
higher temperatures in toluene and tetrahydrofuran, while
the (10S,2R)-isomer was predominant at any (examined)
temperature in the direct excitation. Interestingly, in acet-
onitrile, the (10S,2S)-isomer was always the major product
under the conditions employed. It is also important to note
that the stereoselectivities in acetonitrile became low in both
excitation modes, presumably due to the less tightly bound
radical ionic excited species, but the both excitation modes
still gave the distinctly different linear Eyring plots, implying
that the same diastereodifferentiation mechanism operativ-
ing in this solvent.
(d) Differential Activation Parameters for the Diastereo-
differentiation. To more quantitatively discuss the solvent effect
on the two excited-state complex species, the Eyring analyses24
were performed for the de’s obtained at different temperatures in
various solvents. By plotting the ln{[(10S,2S)-3a]/[(10S,2R)-3a]}
values against the reciprocal temperature (1/T), we obtained the
differential activation enthalpy (ΔΔH‡) and entropy (ΔΔS‡)
values (Table 3). As each Eyring plot gave a nice straight line
of different slope and intercept (Figure 1), operation of a single,
but distinct, diastereodifferentiation mechanism is confirmed for
each excitation mode in all the solvents examined. In this
FIGURE 1. Temperature dependence of the de of 3a obtained upon di-
rect (blue) and CT (green) excitation in methylcyclohexane (filled circle),
toluene (cross), tetrahydrofuran (triangle), and acetonitrile (open circle).
The regression lines were obtained by the least-squares fitting.
exemplified in Figure 1. As shown in Table 3, the equipodal
temperature T0 is a critical function of the irradiation condi-
tions, i.e., the excitation mode and solvent. This analysis
reveals that the observed preference for a specific diastereo-
mer is just an incidental result obtained in a given solvent and
at a given temperature, and therefore, a comparison of the
differential activation parameters for both excitation modes is
more crucial for elucidating the nature of the two distinct
excited species.
(3) The direct and CT excitation constantly led to the
oppositely signed ΔΔH‡ and ΔΔS‡ in all the solvents em-
ployed. This clearly designates that the two excited-state
complexes formed via direct and CT excitation are comple-
tely distinct species in structure and reactivity and, more
importantly, that despite the same chirogen introduced to
the substrate, the diastereoface selectivity is inverted for the
two complex species in terms of the activation parameters.
(4) The ΔΔH‡ and ΔΔS‡ values are sensitive to solvent
polarity and gradually decrease with increasing polarity to
eventually be inverted in sign in acetonitrile, irrespective of
the excitation mode. This suggests the role of specific solva-
tion of polar solvents to the excited species in the critical dias-
tereodifferentiating step. More intriguingly, the direct and
CT excitation in methylcyclohexane containing 2% acetoni-
trile led to contrasting behavior of the activation parameters
(Table 3), showing smaller decreases (than in acetonitrile)
ꢀ
€
Paterno-Buchi reaction occurring in the singlet manifold, the
photocyclization is likely to proceed thorough the concerted or
short-lived 1,4-biradical mechanism, and hence, the diastereo-
face-differentiating complexation of chiral alkyl cyanobenzoate
with olefin donor is solely responsible for the observed de values,
as the subsequent ring-closure process is very fast.
A carefulcomparisonoftheactivation parametersobtained
in both excitation modes reveals the contrasting nature of the
two excited-state complex species as indicated below.
(1) Consistently larger |ΔΔH‡| and |ΔΔS‡| values were
obtained in nonpolar methylcyclohexane and toluene than in
polar THF and acetonitrile, irrespective of the excitation mode.
This implies that the two excited-state complexes, generated by
direct and CT excitation, share a common feature to certain
degree, being more or less polarized than the relevant ground-
state species and therefore better solvated in polar solvents to
make the critical diastereodifferentiation less efficient.
(2) The changes in ΔΔH‡ and ΔΔS‡ do not behave inde-
pendently but compensate each other(videinfra). Asa natural
consequence of the ΔΔH‡ and ΔΔS‡ values with the same sign
and the differential Gibbs-Helmholtz equation, ΔΔG‡ =
ΔΔH‡ - TΔΔS‡, the dominant diastereomer is switched at
a critical temperature (i.e., equipodal temperature, T0), as
(24) Eyring, H. J. Chem. Phys. 1935, 3, 107–115.
J. Org. Chem. Vol. 75, No. 16, 2010 5465

Matsumura et al.
JOCFeatured Article
FIGURE 3. Enthalpy-entropy compensation plots of the activa-
tion parameters obtained upon direct (blue closed circle) and CT
(green open circle) excitation in a variety of solvents shown in
Table 3.
FIGURE 2. Effect of acetonitrile content on the de of 3a produced
upon CT excitation of 1a and 2 in methylcyclohexane at 25 °C (solid
line) and at 50 °C (dashed line).
upon direct excitation versus further augmentation upon CT
excitation. These results indicate that the degree and impact
of selective solvation are significantly different for the two
excited-state complex species, probably due to the difference
in dipole moment and polarizability.
from 4% to 13%, but further addition lowered the de and
eventually switched the chiral sense of product to give 3a in
-5% de, which is comparable to that (-8% de) obtained in
pure acetonitrile. A parallel trend, including the de values opti-
mized in 2% acetonitrile and the inverted chiral sense of
product in 100% acetonitrile, was observed for the CT excita-
tion at 50 °C (Figure 2). These results are accounted for in terms
of the specific solvation to the excited CT complex, as proposed
previously for a polarized exciplex species involved in the polar
photoaddition of alcohols to aromatic olefins.25 The different
solvent dependence behavior and activation parameters ob-
served upon direct versus CT excitation confirm that the con-
ventional exciplex and the excited CT complex produced in
each excitation mode do not interconvert or share common
geometry, polarity/polarizability, or reactivity.
(f) Enthalpy-entropy Compensation. The compensatory
enthalpy-entropy relationship26 is a widely observed phenom-
enon in chemistry and biochemistry, and the origin and physical
meanings of the extrathermodynamic relationship have been a
target of much debate.27 Figure 3 shows the enthalpy-entropy
compensation plots for the differential activation parameters
obtained upon direct and CT excitation of a mixture of 1a and 2
in various solvents, including methylcyclohexane containing
2% acetonitrile. In all of the cases examined, a good linear
regression line that passes through the origin was obtained for
each of the two excitation modes, indicating that a single mecha-
nism is operative in the critical diastereodifferentiating step. The
slope of the ΔΔS‡-ΔΔH‡ plot, having the dimension of
temperature, is called the “isokinetic,” or “isodiastereodiffer-
entiating,” temperature (β), since the de value becomes identical
at this specific temperature, regardless of the solvent used.28 The
smaller slope (β) forformationof3a means heavier contribution
of the entropy factor than the enthalpy factor to the diastereo-
differentiation in the direct rather than CT excitation (Table 4).
(5) The ΔΔH‡ and TΔΔS‡ values obtained for the CT
excitation balance to each other at T = 298 K, while the en-
tropic contribution (TΔΔS‡) overwhelms the enthalpic con-
tribution (ΔΔH‡) upon direct excitation, as can be seen from
Table 3. This means that the diastereoface selectivity upon
photocycloaddition is controlled almost equally by the en-
thalpic and entropic factors in the excited CT complex gene-
rated by CT excitation but is more critically controlled by the
entropic factor in the conventional exciplex formed via direct
excitation. Furthermore, both the absolute ΔΔH‡ and ΔΔS‡
values are larger for the direct excitation than for the CT
excitation, indicating that the conventional exciplex is more
susceptible to the temperature and solvent than the excited CT
complex. This seems reasonable because the exciplex is a
speciesformeddynamicallyinthe excitedstate withsignificant
desolvation and solvent reorganization, while the CT complex
which is originally solvated in the ground state suffers only
modest solvent reorganization after excitation.
(e) Specific Solvation by Acetonitrile. In both excitation
modes, the use of polar acetonitrile solvent led to the sign
inversion of the ΔΔH‡ and ΔΔS‡ values (Table 3). We there-
fore examined the effects of acetonitrile (1-10%) added to
methylcyclohexane on the photocycloaddition of 1a to 2. In
the presence of acetonitrile, the photoreaction was more or
less decelerated but gave 3a in modest to good yields (Table
S2, Supporting Information). However, the de of 3a obtained
upon direct irradiation decreased abruptly from 64% to -3%
upon addition of only a small portion (1%) of acetonitrile and
stayed at low values, ranging from þ2 to -8%, in methylcy-
clohexane solutionscontaining2-10% acetonitrile, which are
comparable to the -12% de obtained in 100% acetonitrile.
This result indicates the strong preferential solvation of
acetonitrile molecules to the conventional exciplex.
(25) Asaoka, S.; Wada, T.; Inoue, Y. J. Am. Chem. Soc. 2003, 125, 3008–
3027.
(26) Leffler, J. E. J. Org. Chem. 1955, 20, 1202–1231.
(27) (a) Starikov, E. B.; Norden, B. J. Phys. Chem. B 2007, 111, 14431–
14435. (b) Kirk, W. R. J. Theo. Comp. Chem. 2004, 34, 511–520. (c) Liu, L.;
Guo, Q.-X. Chem. Rev. 2001, 101, 673–695. (d) Inoue, Y.; Wada, T. Adv.
Supramol. Chem. 1997, 4, 55–96. (e) Dunitz, J. D. Chem. Biol. 1995, 2, 709–
712.
In the CT excitation, the effect of added acetonitrile was
more modest and not uniform. As shown in Figure 2, a small
amount of acetonitrile (up to 2%) added to the methylcyclo-
hexane solution at 25 °C appreciably enhanced the de of 3a
(28) Giese, B. Acc. Chem. Res. 1984, 17, 438–442.
5466 J. Org. Chem. Vol. 75, No. 16, 2010

Matsumura et al.
JOCFeatured Article
TABLE 4. Comparison of Isodiastereodifferentiating Temperature β
Obtained from the Entropy-Enthalpy Compensation Plot for the Direct
versus Charge Transfer Band Photoexcitation of 1a or 1b with 2a
substrate
βdirect (K)
βCT (K)
Δβb (K)
1a
1b
177
79
275
436
98
357
aThe values were obtained by the slope of plots in Figures 3 and S4
(Supporting Information). bΔβ = βCT - βdirect
.
The β values for the formation of 3b are also tentatively
evaluated using the data in methylcyclohexane for a comparison
purpose. These observations reinforce the above-mentioned
conclusions that the exciplex and the excited CT complex differ
in structure and reactivity and such a difference in slope is
consistent with a greater flexibility in the conventional exciplex
than the excited CT complex (vide infra).
(g) Substituent Effects. Photocycloaddition of cyano-
benzoate 1a to more electron-rich 1,1-bis(p-methoxyphenyl)-
ethene under comparable conditions was unsuccessful, although
the diarylethene was completely consumed to give the corre-
sponding dimer and other unidentified products, which is pro-
bably attributable to an efficient electron-transfer quenching by
this donor. The free energy changes (ΔG) of the electron transfer
from diarylethenes to excited 1a in toluene and acetonitrile were
estimated by the Rehm-Weller equation.29 Using the oxidation
potentials of the ethenes (Eox = 1.32 and 1.88 V for methoxy-
and parent diphenylethene, respectively),30 the reduction poten-
tial and the singlet energy of acceptor (Ered = -1.85 V for
methyl p-cyanobenzoate31 and ΔE0,0 = 411 kJ mol-1 of 1a from
its absorption spectrum), intensely negative ΔG values were
obtained for 1,1-bis(p-methoxyphenyl)ethene; i.e., ΔG = -58
and -112 kJ mol-1 in toluene and acetonitrile, respectively. It is
FIGURE 4. Optimized geometries of (10S,2R)- and (10S,2S)-3a at
the B-LYP/TZVP level and their precursors (i.e., complex between
(S)-1a with 2) at the DFT-D-B-LYP/TZVP level. The most abun-
dant Tgþ conformers are shown.
the direct excitation. In the direct excitation, both parameters
became smaller, but the effect was more pronounced on the
enthalpic term. The relatively smaller effects were found for
the CT excitation, where the ΔΔH‡ was slightly decreased
while the ΔΔS‡ was practically unaffected (Table 3). This
contrasting behavior of the activation parameters upon direct
versus CT excitation indicates that the peripheral part of the
chiral group is more closely located to the reaction partner 2
and hence more influential in the exciplex than in the excited
CT complex. The smaller changes found for the parameters in
the CT excitation may be ascribed to the diastereodifferentia-
tion that principally occurred at the stage of the complexation
in the ground state.
ꢀ
€
to be noted, however, that the Paterno-Buchi reaction of 1a
with 2 underwent at comparative efficiencies in acetonitrile and
in toluene, where the |ΔG| value is much smaller (ΔG=-30 and
þ25 kJ mol-1, respectively).
A further mechanistic support for the distinction between
the conventional exciplex and the excited CT complex came
from the differential activation parameters obtained upon
photocycloaddition of 1b, which possesses a slightly bulkier
chiral group.10 Note that two additional methyl groups are
introduced at a peripheral position of the chiral group in 1b.
Thus, the irradiation of a mixture of 1b and 2 at 290 or 330
nm afforded the corresponding oxetane 3b as a sole product
in a good yield. The absolute configurations of diastereo-
meric 3b isomers were tentatively assigned by the compar-
ison of their retention times with those of 3a upon HPLC
analysis on the same reversed-phase column, assuming that
the elution order is preserved for both 3a and 3b.
The temperature-dependence behavior of de of 3b was
similarly investigated and analyzed by using the Eyring
analysis,10 and the differential activation parameters obtained
in various solvents were examined to give excellent straight
lines of significantly different slopes or β for direct and CT
excitation (Figure S4, Supporting Information). The modifi-
cation of the chiral ester moiety at a remote position caused
the dramatic changes in differential activation parameters for
Quantum Chemical Investigation of the Diastereomeric
Complexes. To theoretically estimate the structural and ener-
getic (enthalpic) differences between the diastereomeric com-
plexes of 1a with 2, all possible conformations of the complex,
taking into account the nine rotamers about the C*-O and
C*-C bond in 1a (see Chart S1 in the Supporting Information),
were calculated by the recently developed dispersion-corrected
DFT method.32 A pair of diastereomeric [1a 2] complexes,
3
which are precursors to (10S,2R)- and (10S,2S)-3a, were opti-
mized at the DFT-D-B-LYP/TZVP level,33 starting from the
hypothetical complexation of 1a and 2 with a separation of 3.4
˚
A. As can be seen from Figure 4, the optimized structures
obtained sustain the π-πstacking between the aromatic rings of
the donor and acceptor. In general, the π overlap between the
donor and acceptor molecules is maximized in the CT complex
to give a conformationally rigid stacked structure.34 Since the
second aromatic ring in diphenylethene is almost perpendicular
€
(32) (a) Grimme, S.; Antony, J.; Schwabe, T.; Muuck-Lichtenfeld, C.
Org. Biomol. Chem. 2007, 5, 741–758. (b) Grimme, S. J. Comput. Chem. 2004,
25, 1463–1473.
(33) (a) Becke, A. D. Phys. Rev. A 1988, 38, 3098–3100. (b) Lee, C.; Yang,
W.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789. (c) Schaefer, A.; Huber, C.;
Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829–5835.
(34) (a) Rosokha, S. V.; Kochi, J. K. J. Org. Chem. 2002, 67, 1727–1737.
(b) See also: Niimura, N.; Ohashi, Y.; Saito, Y. Bull. Chem. Soc. Jpn. 1968,
41, 1815–1820. (c) Kobayashi, T.; Yoshihara, K.; Nagakura, S. Bull. Chem.
Soc. Jpn. 1971, 44, 2603–2610.
(29) (a) Rehm, D.; Weller, A. Isr. J. Chem. 1970, 8, 259–270. (b) Weller, A.
Z. Phys. Chem. Neue Folge 1982, 133, 93–98.
(30) Gollnick, K.; Schnatterer, A.; Utschick, G. J. Org. Chem. 1993, 58,
6049–6056.
(31) Arnold, D. R.; Maroulis, A. J. J. Am. Chem. Soc. 1976, 98, 5931–
5937.
J. Org. Chem. Vol. 75, No. 16, 2010 5467

Matsumura et al.
JOCFeatured Article
SCHEME 2. Mechanisms of Diastereodifferentiating [2 þ 2] Photocycloaddition of Chiral Alkyl Benzoate 1 to 1,1-Diphenylethene 2 via
Conventional Exciplex (Blue) and Excited CT Complex (Green) Formed, Respectively, upon Direct and Charge-Transfer Band Excitation
(Black Indicates Ground-State Events)a
aThe ester substituent R contains a chiral group.
to the first π-stacked ring, the whole molecule is twisted in the
CT complex. The calculated structures are for the ground-state
complexes, but we assume that these structures are more or less
relevant to the excited CT complexes. Due to the strong π-π
stacking, the excited CT complex will not be able to gain addi-
tional conformational freedoms upon excitation, and the en-
tropic difference between the diastereomer pair should be much
smaller than that for the exciplex, in nice agreement with the
experimental results. The subsequent (COSMO)-SCS-MP2 en-
ergy calculations35 revealed that the conformers are populated
in a ratio of 55:45 in the gas phase or 54:46 in acetonitrile (Table
S3, Supporting Information). This preference for the (10S,2R)-
isomer predicted by theory is in nice agreement with the
experimental result obtained at temperatures lower than the
equipodal temperature (T0) in the CT excitation of 1a with 2.
The solvent effect was not readily approximated by the current
solvation model (conductor-like screening model) employed in
our calculation.
(10S,2S)-3a. In the direct excitation, the excited state of
cyanobenzoate interacts with diphenylethene to form an
exciplex, and the interaction is expected to occur near the
carbonyl group. The conformation of the chiral group is
expected to affect more dynamically on the more flexible
exciplex structure but less effectively on the originally
π-stacked, rigid CT complex. However, the present calcula-
tions do not take into account the entropic contributions,
despite their critical roles in differentiating the two excited
complex species. Since the reliable structural estimation of
such excited species by theoretical calculation is not feasible
at the current stage, our experimental procedures to obtain
and analyze the activation parameters under a variety of
conditions provide us with a wealth of information to
elucidate the nature of the excited-state complex species
generated via the two distinct excitation modes.
Summary and Conclusions
The diastereomeric pairs of nine rotamers of photopro-
duct 3a were computed by the standard DFT method at the
B-LYP/TZVP level (Figure 4 and Table S3, Supporting
Information). By weight-averaging the relative (COSMO)-
SCS-MP2 energies of all possible conformers, (10S,2R)-3a
was found to be energetically more favored than (10S,2S)-3a
in a ratio of 58:42 (in gas phase) or 56:44 (in acetonitrile). The
most favored conformation of the chiral group was calcu-
lated to be Tgþ in (10S,2R)-3a, which is consistent with the
X-ray structure of (10S,2R)-3a, and this conformation is also
the predominant one in the parent ester 1a.10 It is to be noted
that this conformer is slightly disfavored in diastereomeric
From the comprehensive analysis of the temperature and
ꢀ
€
solvent effects on the diastereodifferentiating Paterno-Buchi
reaction of chiral alkyl cyanobenzoates with 1,1-diphenylethene,
we can contrast the two diastereodifferentiation mechanisms
operativeindirect versusCTexcitationasillustratedinScheme2.
ꢀ
€
Upon direct excitation of cyanobenzoate, the Paterno-Buchi
reaction with diphenylethene proceeds in the singlet manifold to
give a diastereomeric pair of excited-state complexes (i.e., “con-
ventional” exciplex), the diastereomer ratio of which determines
the product’s de through the subsequent cyclodimerization
reaction. The good linear Eyring plots obtained indicate that
no switching of diastereodifferentiation mechanism, such as
different cyclization rates for the diastereomeric exciplex pair,
is involved in the present photocycloaddition reaction. The first
(35) (a) Grimme, S. J. Phys. Chem. A 2005, 109, 3067–3077. (b) Grimme,
S. J. Chem. Phys. 2003, 118, 9095–9102.
5468 J. Org. Chem. Vol. 75, No. 16, 2010

Matsumura et al.
JOCFeatured Article
excited singlet state of cyanobenzoates is mostly π-π* in nature
but borrows some intensity from the n-π* state,10 and thus, the
interaction is expected to occur near the carbonyl group in the
exciplex formation. For this reason, the chiral substituent can
more critically control the product’s de upon direct, rather than
Photolysis. All irradiations were performed under argon
atmosphere at a given temperature in a temperature controller.
The solution was irradiated in a 1-cm square cuvette with a 100-
W xenon light source. For the band-selective excitations at
direct (290 nm) and CT bands (330 nm), a UVB mirror module
fitted with HQBP290-UVB filter and a UVA mirror module
with HQBP330-UVA filter were employed, respectively. The
product distribution and conversion of photolyzed samples
were determined by HPLC fitted with a UV detector (254 nm)
by monitoring each peak (relative to 3,5-dimethoxymethyl-
benzoate added as an external standard) on an ODS column
(4.6 ꢀ 250 mm) eluted with a 65:35 mixture of acetonitrile and
water, containing 0.5% (v/v) triethylamine, at a flow rate of
1 mL/min. The retention times of 1a, 2, (10S,2S)-3a, (10S,2R)-3a,
and the standard were 10.1, 23.2, 47.3, 49.8, and 6.0 min,
respectively. The photoproducts from the reaction of 1b with 2
were not isolated but identified by GC-MS, and the diastereos-
electivities were similarly determined by HPLC (monitor at 254
nm) under conditions identical to those employed for the
reaction of 1a with 2.
ꢀ
€
CT, excitation in the Paterno-Buchi reaction. Exciplex forma-
tion is a dynamic process that involves the initial desolvation
upon association and the subsequent structural relaxation of the
collision complex with accompanying extensive solvent reorga-
nization around it. In contrast, the ground-state CT complexa-
tion is mostly driven by the π-π stacking of donor and acceptor
molecules to maximize the orbital overlap. Originally stacked to
each other in the ground state, the excited CT complex is consi-
dered to be less flexible in structure and less extensive in desolva-
tion and/or solvent reorganization. This makes the exciplex
more susceptible to environmental factors such as temperature
and solvent, relative to the excited CT complex. Indeed, the de of
the oxetane produced upon direct excitation displayed larger
(and oppositely signed) slope and intercept in the Eyring plot
compared to those for CT excitation, supporting this view.
Quantum Chemical Computations. All calculations were per-
formed on Linux PCs using the TURBOMOLE 5.8 program
suite.36 Geometry optimizations were performed at the B-LYP/
TZVP or dispersion-corrected DFT-D-B-LYP/TZVP level
within the density functional theorem without any symmetry
constraint (C1) with a numerical quadrature multiple grid of m4.
Subsequent single-point energy calculations were performed by
the SCS-MP2 method with a TZVPP basis set. Further details of
the calculation were described previously.20,37 The effect of solvent
on the conformer distribution was also evaluated by using the
PCM model. Thus, single-point energy calculations were per-
formed with the conductor-like screening model (COSMO)38 as
implemented in the TURBOMOLE program suite. The di-
electric constant (ε) of 36.64 in acetonitrile and optimized
ꢀ
€
In this detailed mechanistic study on the Paterno-Buchi
reaction of chiral cyanobenzoates with diphenylethene, we
have demonstrated that the conventional exciplex (formed
upon direct excitation) and the excited CT complex (formed
upon CT-band excitation) are independent and noninterconver-
tible excited species that possess distinctly different structures
and reactivities, which eventually afford the same diastereomeric
oxetane pair in different diastreoselectivities. The two excited-
state complex species were comparatively and quantitatively
discussed for the first time in terms of the differential activation
parameters and isokinetic temperatures. It is interesting and
potentially useful that the excitation mode, manipulated simply
by changing the irradiation wavelength, can switch the diastereo-
facial selectivities upon complex formation and the subsequent
photocycloaddition process (at a very low or high temperature,
where the major controlling factor is the enthalpy or entropy,
respectively). This sort of stereochemical control by irradiation
wavelength can be extended to other photoreactions if the
excitation at the CT band affords the excited species different
from the conventional exciplex. Further research along this line is
currently underway in other photochemical systems.
˚
atomic radii (C, 2.00; N, 1.83; O, 1.72; H, 1.30 A) were used
for the construction of the molecular cavity.
Acknowledgment. Financial support by a Grant-in-Aid
for Scientific Research, JSPS, the Mitsubishi Chemical Cor-
poration Fund, and the Sumitomo Foundation is greatly
acknowledged.
Supporting Information Available: Details of the determi-
nation of the association constants in various solvents, CD
spectra of the complex, the temperature dependence of de upon
254 and 313 nm irradiation, the enthalpy-entropy compensa-
tion plot for the photoreaction of 1b with 2, the detailed results
of the photolysis in mixed solvents and in the presence of triplet
quencher, the NMR spectra for 1b, and the Cartesian coordi-
nate for optimized geometries. This material is available free of
charge via the Internet at http://pubs.acs.org.
Experimental Section
General Methods. Spectrophotometric-grade methylcyclo-
hexane, toluene, tetrahydrofuran, and acetonitrile were used
as received in the photoreactions and spectroscopic measure-
ments. 1,1-Diphenylethene was purified by column chromatog-
raphy over alumina under argon atmosphere before use, which
completely eliminated the impurities absorbing at around 340
nm (presumably benzophenone) in the commercial samples.
Preparation of (S)-1b. Reaction of (S)-4-methyl-2-pentanol
with 4-cyanobenzoyl chloride in pyridine gave (S)-1b as a colorless
oil in 93% yield. 1H NMR (400 MHz, CDCl3): δ 0.93 (d, J = 3.6
Hz, 3H), 0.95 (d, J = 3.6 Hz, 3H), 1.35 (d, J = 6.2 Hz, 3H),
€
(36) Ahlrichs, R.; Bar, M.; Baron, H. P.; Bauernschmitt, R.; Bocker, S.;
Crawford, N.; Deglmann, P.; Ehrig, M.; Eichkorn, K.; Elliott, S.; Furche, F.;
€
€
Haase, F.; Haser, M.; Horn, H.; Hattig, C.; Huber, C.; Huniar, U.;
Kattannek, M.; Kohn, A.; Kolmel, C.; Kollwitz, M.; May, K.; Nava, P.;
€
€
€
€
€
Ochsenfeld, C.; Ohm, H.; Patzelt, H.; Rappoport, D.; Rubner, O.; Schafer,
A.; Schneider, U.; Sierka, M.; Treutler, O.; Unterreiner, B.; Arnim, M. v.;
€
Weigend, F.; Weis, P.; Weiss, H. TURBOMOLE, version 5.8; Universitat
1.38-1.45 (m, 1H), 1.65-1.76 (m, 1H), 5.22-5.30 (m, 1H), 7.704
Karlsruhe: Karlsruhe, 2005. See also: http://www.cosmologic.de/Quantum-
Chemistry/main_turbomole.html.
0
(AA0XX0, JAX = 8.7 Hz, JAA = 1.8 Hz, 2H), and 8.13 (AA XX ,
0
JXA = 8.7 Hz, JXX = 1.8 Hz, 2H). 13C NMR (100 MHz, CDCl3):
(37) For recent examples, see: (a) Nishizaka, M.; Mori, T.; Inoue, Y.
J. Phys. Chem. Lett. 2010, 1, 1809–1812. (b) Nishizaka, M.; Mori, T.; Inoue, Y.
J. Phys. Chem. Lett. 2010, 1, 2402-2405.
(38) (a) Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027–2094. (b)
Sinnecker, S.; Rajendran, A.; Klamt, A.; Diedenhofen, M.; Neese, F. J. Phys.
Chem. A 2006, 110, 2235–2245.
0
δ 20.5, 22.4, 22.9, 24.9, 45.1, 71.4, 116.2, 118.1, 130.1, 132.2, 134.7,
and 164.6. CI-MS m/z (relative intensity): 232 (Mþ þ 1, 100), 85
(68). Anal. Calcd for C14H17NO2: C, 72.70; H, 7.41; N, 6.06.
Found: C, 72.47; H, 7.12; N, 6.22.
J. Org. Chem. Vol. 75, No. 16, 2010 5469