Effects of Spiroconjugation on the Calculated Singlet-Triplet Energy Gap in 2,2-Dialkoxycyclopentane-1,3-diyls and on the Experimental Electronic Absorption Spectra of Singlet 1,3-Diphenyl Derivatives. Assignment of the Lowest-Energy Electronic Transition

Article abstract of DOI:10.1021/ja038305b

The effect of a 2,2-ethylene-ketal functionality on the singlet-triplet energy gap (ΔE_ST) and on the first electronic transition in singlet cyclopentane-1,3-diyls (1) has been investigated. UDFT calculations predict a significant increase in th

Full text of DOI:10.1021/ja038305b

Published on Web 12/24/2003
Effects of Spiroconjugation on the Calculated Singlet-Triplet
Energy Gap in 2,2-Dialkoxycyclopentane-1,3-diyls and on the
Experimental Electronic Absorption Spectra of Singlet
1,3-Diphenyl Derivatives. Assignment of the Lowest-Energy
Electronic Transition of Singlet Cyclopentane-1,3-diyls
Manabu Abe,*,† Waldemar Adam,‡ Weston Thatcher Borden,*,§ Masanori Hattori,†
David A. Hrovat,§ Masatomo Nojima,† Koichi Nozaki, and Jakob Wirz*,|
Contribution from the Department of Materials Chemistry and Frontier Research Center
(HANDAI FRC), Graduate School of Engineering, Osaka UniVersity (HANDAI), Suita,
Osaka 565-0871, Japan, Institut fu¨r Organische Chemie der UniVersita¨t Wu¨rzburg,
Am Hubland, D-97074 Wu¨rzburg, Germany, Department of Chemistry, Box 351700,
UniVersity of Washington, Seattle, Washington 98195-1700, Department of Chemistry,
Graduate School of Science, Osaka UniVersity (HANDAI), 1-16 Machikaneyama,
Toyonaka, Osaka 560-0043, Japan, and Departement Chemie der UniVersita¨t Basel,
Klingelbergstrasse 80, CH-4056 Basel, Switzerland
Received September 3, 2003; E-mail: abe@chem.eng.osaka-u.ac.jp, borden@chem.washington.edu, j.wirz@unibas.ch
Abstract: The effect of a 2,2-ethylene-ketal functionality on the singlet-triplet energy gap (∆EST) and on
the first electronic transition in singlet cyclopentane-1,3-diyls (1) has been investigated. UDFT calculations
predict a significant increase in the preference for a singlet ground state in the diradical with the cyclic
ketal at C2 (1g; ∆EST ) -6.6 kcal/mol in C2 symmetry and -7.6 kcal/mol in C2v symmetry), compared to
the 2,2-dihydroxy- and 2,2-dimethoxy-disubstituted diradicals (1d, ∆EST ) -3.6 kcal/mol in C2 symmetry,
and 1e, ∆EST ) -3.4 kcal/mol in C2 symmetry). Spiroconjugation is shown to be responsible for the larger
calculated value of |∆EST| in 1g, relative to 1d and 1e. A strong correlation between the calculated values
of ∆EST and the computed electronic excitation energies of the singlet diradicals is found for diradicals 1d,
1e, and 1g and for 2,2-difluorocyclopentane-1,3-diyl (1c). A similar correlation between ∆EST and λcalcd is
predicted for the corresponding 1,3-diphenylcyclopentane-1,3-diyls 3, and the predicted blue shift in the
spectrum of 3g, relative to 3e, has been confirmed by experimental comparisons of the electronic absorption
spectra of the annelated derivatives 2c, 2e, and 2g in a glass at 77 K. The wavelength of the first absorption
band in the singlet diradicals decreases in the order 2e (λonset ) 650 nm) > 2g (λonset ) 590 nm) > 2c
(λonset ) 580 nm). The combination of these computational and experimental results provides a sound
basis for reassignment of the first electronic absorption band in singlet diradicals 2c, 2e, and 2g to the
excitation of an electron from the HOMO to the LUMO of these 2,2-disubstituted derivatives of cyclopentane-
1,3-diyl.
Introduction
jugative electron acceptors, and geminal disubstitution of
cyclopentane-1,3-diyls (1) at C2 with such substituents has been
Singlet diradicals are key intermediates in processes involving
bond breaking and formation. In contrast to triplet diradicals
such as 1a1 and 2b2, which are EPR-active and relatively long-
lived, singlet diradicals are EPR-silent, and their short lifetimes
make them difficult to observe and characterize.3 Electronegative
substituents (e.g., X ) F or OR) make C-X bonds hypercon-
predicted to result in singlet ground states for 1c4 and 1d5. These
theoretical predictions have been confirmed experimentally by
studies of the singlet diradicals 2c, 2e, and 2f,6 which were found
(3) (a) Pederson, S.; Herek, J. L.; Zewail, A. H. Science 1994, 266, 294-309.
(b) De Feyter, S.; Diau, E. W.-G.; Zewail, A. H. Angew. Chem., Int. Ed.
2000, 39, 260-263.
(4) (a) Xu, J. D.; Hrovat, D. A.; Borden, W. T. J. Am. Chem. Soc. 1994, 116,
5425-5427. (b) Getty, S. J.; Hrovat, D. A.; Borden, W. T. J. Am. Chem.
Soc. 1994, 116, 1521-1527. (c) Borden, W. T. Chem. Commun. 1998,
1919-1925
† Graduate School of Engineering, Osaka University.
‡ Universita¨t Wu¨rzburg.
§ University of Washington.
(5) Abe, M.; Adam, W.; Nau, W. M. J. Am. Chem. Soc. 1998, 120, 11304-
11310.
Graduate School of Science, Osaka University.
| Universita¨t Basel.
(6) (a) Adam, W.; Borden, W. T.; Burda, C.; Foster, H.; Heidenfelder, T.;
Jeubes, M.; Hrovat, D. A.; Kita, F.; Lewis, S. B.; Scheutzow, D.; Wirz, J.
J. Am. Chem. Soc. 1998, 120, 593-594. (b) Abe, M.; Adam, W.;
Heidenfelder, T.; Nau, W. M.; Zhang, X. J. Am. Chem. Soc. 2000, 122,
2019-2026. (c) Abe, M.; Adam, W.; Hara, M.; Hattori, M.; Majima, T.
Nojima, M.; Tachibana, K.; Tojo, S. J. Am. Chem. Soc. 2002, 124, 6540-
6541.
(1) (a) Buchwalter, S. L.; Closs, G. L. J. Am. Chem. Soc. 1975, 97, 3857-
3858. (b) Buchwalter, S. L.; Closs, G. L. J. Am. Chem. Soc. 1979, 101,
4688-4694.
(2) (a) Adam, W.; Harrer, H. M.; Kita, F.; Nau, W. M. AdV. Photochem. 1998,
24, 205-254. (b) Kita, F.; Adam. W.; Jordan, P.; Nau, W. M.; Wirz, J. J.
Am. Chem. Soc. 1999, 121, 9265-9275.
9
574
J. AM. CHEM. SOC. 2004, 126, 574-582
10.1021/ja038305b CCC: $27.50 © 2004 American Chemical Society

Electronic Transitions of Singlet Cyclopentane-1,3-diyls
A R T I C L E S
Table 1. Calculated Singlet-Triplet Energy Gaps (∆EST ) ES - ET) and Excitation Wavelengths and Energies of Cyclopentane-1,3-diyls 1
2
corr
∆EST (kcal/mol)a ( S S , ∆EST
)
λ
calcd (nm)b
excitation energies (kcal/mol)b
1
X
C
2
C
2v
lit.
C (f)c
C (f)c
C
2
C
2v
2
2v
1a
1c
1d
1e
1g
H
+0.64 (1.01, +1.3)
-7.1 (0.77, -11.4)
+1.2d
-9.8e
1148 (0.14)
368.4 (0.39)
24.9
77.6
F
OH
OMe
-O(CH2)2O-
-3.6f (0.87, -6.4)
-3.4f (0.88, -6.1)
-6.6 (0.77, -10.7)
424.1 (0.32)
397.6 (0.33)
68.8
71.9
-7.6 (0.75, -12.2)
390.7 (0.34)
73.2
a Calculations were performed at the UB3LYP/6-31G* level of theory. The values in parentheses indicate the S2 values for the singlet state and the
S
corr
energy gaps computed after correction for spin contamination in the singlet (∆ESTcorr), by using Yamaguchi’s formula (ref 19); ∆EST ) S2 /( S2
-
T
T
S2 ) × ∆EST. The S2 values for the triplet state were calculated to be ca. 2.0; see Supporting Information. b Absorption wavelengths and excitation
S
T
energies were calculated at the RCIS/6-31+G*//UB3LYP/6-31G* level of theory. c Calculated oscillator strengths. d In C2 symmetry at the CISD level of
theory; ref 20. e In C2V symmetry at the CASPT2N level of theory; ref 4a. f Substituents (H or Me) on both oxygen atoms are exo to the cyclopentane ring.
to be relatively long-lived, with lifetimes of up to 1 µs at room
temperature. Each of these three singlet diradicals was also
discovered to exhibit an intense absorption band in the visible
region.
density functional theory (TD-RB3LYP)16 with either the
6-31G*12 or the 6-31+G*17 basis set. All of the calculations
were carried out with the Gaussian 98 suite of programs.18
Optimized geometries and energies for all the molecules
discussed in this paper are available as Supporting Information.
Results and Discussion
Calculated Effect on ∆EST of the Ethylene Ketal Func-
tionality at C2 in 1g. The singlet-triplet energy gaps (∆EST)
of cyclopentane-1,3-diyls 1c-g were calculated at the UB3LYP/
6-31G* level of theory (Table 1). UB3LYP calculations
generally give values of ∆EST that are too small in magnitude
when they are compared to the values obtained from high-level
ab initio calculations. For example, the UB3LYP/6-31G*
calculations on 1c give a value of ∆EST ) -7.1 kcal/mol, which
is 2.7 kcal/mol smaller in magnitude than that obtained by
CASPT2/6-31G* calculations.4a
The kinetic stability of 2c, 2e, and 2f has provided opportuni-
ties to investigate the chemical behavior and electronic character
of these localized singlet 1,3-diradicals.7-9 The possibility of
performing experiments on 2c, 2e, and 2f has, in turn, motivated
additional calculations.10,11 In the context of these ongoing
experimental and computational studies, we have investigated
the effects of different C26a,b and p-phenyl6c substituents on the
chemistry and spectroscopy of the singlet diradicals 1 and 2.
For the seemingly inconsequential replacement of either the
hydroxy substituents in 1d or the methoxy substituents in 1e
with ethylene-ketal in 1g, our calculations have predicted
unexpectedly large effects on both the singlet-triplet energy
gap (∆EST ) ES - ET) and on the wavelength (λ) of the first
electronic absorption band of the singlet diradicals. The surpris-
ing substituent effect, calculated for 1g, prompted us to generate
the diradical 2g to test these predictions.
Here we report the results of this combined theoretical and
experimental study. Our calculations and experiments not only
reveal a new type of substituent effect on both ∆EST and λ in
singlet 2,2-disubstituted cyclopentane-1,3-diyls, but also provide
a definitive assignment of the lowest-energy electronic transition
in the singlet diradicals 2c and 2e-g. These findings have led
us to correct our earlier tentative assignment of the long-
wavelength absorption band of 2c, from being due to a parity-
forbidden absorption of its benzyl chromophores6a to being
caused by the excitation of an electron from the HOMO to the
LUMO of this singlet 2,2-difluorocyclopentane-1,3-diyl deriva-
tive.
The tendency for UB3LYP calculations to underestimate the
magnitudes of ∆EST values is due to the fact that unrestricted
“singlet” wave functions are actually mixtures of the wave
functions for pure singlet and triplet states. For example, the
(7) For studies of boron-centered singlet cyclobutane-1,3-diyls, see: (a)
Scheschkewitz, D.; Amii, H.; Gornitzka, H.; Schoeller, W. W.; Bourissou,
D.; Bertrand, G. Science 2002, 295, 1880-1881. (b) Wentrup, C. Science
2002, 295, 1846-1847.
(8) For studies of singlet 1,3-diphosphacyclobutane-2,4-diyls, see: (a) Grutz-
macher, H.; Breher, F. Angew. Chem., Int. Ed. 2002, 41, 4006-4011. (b)
Schoeller, W. W.; Begemann, C.; Niecke, E.; Gudat, D. J. Phys. Chem. A
2001, 105, 10731-10738. (c) Niecke, E.; Fuchs, A.; Baumeister, F.; Martin
Nieger, Schoeller, W. W. Angew. Chem., Int. Ed. Engl. 1995, 34, 555-
557.
(9) For reviews of non-Kekule´ diradicals with singlet ground states, see: (a)
Borden, W. T.; Iwamura, H.; Berson, J. A. Acc. Chem. Res. 1994, 27, 109-
116. (b) Berson, J. A. Acc. Chem. Res. 1997, 30, 238-244.
(10) Abe, M.; Ishihara, C.; Nojima, M. J. Org. Chem. 2003, 68, 1618-1621.
(11) Zhang, D. Y.; Hrovat, D. A.; Abe, M.; Borden, W. T. J. Am. Chem. Soc.
2003, 125, 12823-12828.
(12) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213-222.
(13) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652.
(14) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(15) Foresman, J. B.; Head-Gordon, M.; Pople, J. A.; Frisch, M. J. J. Phys.
Chem. 1992, 96, 135-149.
(16) (a) Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1998,
109, 8218-8224. (b) Casida, M. E.; Jamorksi, C.; Casida, K. C.; Salahub,
D. R. J Chem. Phys. 1998, 108, 4439-4449.
(17) Frisch, M. J.; Pople, J. A.; Binkley, J. S. J. Chem. Phys. 1984, 80, 3265-
3269.
(18) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A., Jr.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.;
Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,
C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.;
Johnson, B. G.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon,
M.; Replogle, E. S.; Pople, J. A. Gaussian 98, revision A 11.3; Gaussian,
Inc.: Pittsburgh, PA, 1998.
Computational Methodology
Geometry optimizations and calculations of the energy gap
(∆EST) were performed with the 6-31G* basis set,12 using
Becke’s hybrid, three-parameter functional13 and the nonlocal
correlation functional of Lee, Yang, and Parr (B3LYP).14
Excitation energies were computed by RCIS15 or time-dependent
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 575

A R T I C L E S
Abe et al.
stronger the HC between ψS and the antibonding C2-X σ*
orbitals, the greater the stabilization of ψS relative to ψA and,
hence, of the singlet relative to the triplet.
As depicted graphically in Figure 1, the greater electrone-
gativity of fluorine, compared to oxygen, results in σC-F
*
orbitals being lower in energy than σC-O* orbitals. The lower
the energy of the σC-X* pseudo π orbitals, the stronger their
HC with ψS and the greater the stabilization of the singlet
relative to the triplet. Thus, the larger calculated magnitude of
|∆EST| in 1c (X ) F) than that in 1d (X ) OH) or 1e (X )
OMe) is easily understood. However, the unexpected finding,
that the ethylene-ketal functionality (X,X ) -O(CH2)2O-) at
C2 in 1g stabilizes the singlet state relative to the triplet, as
much or more than the pair of fluorine atoms in 1c, indicates
that, in addition to HC, another factor plays a role in determining
∆EST. As shown in Figure 1 and as discussed below, we propose
that this additional factor is spiroconjugation (SC)21 of the
pseudo-π combination of 2p nonbonding AOs on the oxygen
atoms (nO) with the antisymmetric combination of 2p-π AOs
at C1 and C3 (ψA). SC stabilizes the lone pair of electrons nO
but destabilizes ψA. Consequently, like HC, SC increases the
energy difference between ψS and ψA and, thus, increases the
magnitude of ∆EST.
Figure 1. Orbital interaction diagram for a cyclopentane-1,3-diyl, disub-
stituted at C2 with X ) F or X ) OR, showing (a) hyperconjugation (HC)
of the symmetric combination of the 2p-π AOs at C1 and C3 (ψS) with the
pseudo π combination of the antibonding C-X σ* orbitals at C2. (b)
Spiroconjugation (SC) of the pseudo-π combination of 2p nonbonding AOs
on X (nX) with the antisymmetric combination of 2p-π AOs at C1 and C3
(ψA). The greater electronegativity of fluorine relative to oxygen should
make HC more important for X ) F than for X ) OR in determining the
energy of the HOMO. However, at a C2V geometry, where nO is properly
aligned for maximal interaction with ψA, the greater electronegativity of
fluorine relative to oxygen should make SC more important for X ) OR
than for X ) F in determining the energy of the LUMO.
Since SC in 1g involves the 2p nonbonding AOs on oxygen,
SC should be sensitive to the spatial orientation of these AOs.
SC of nO with ψA should be strongest at a geometry where the
nodal planes of these two orbitals are coincident. The expected
conformational dependence of SC would explain why ∆EST is
larger at an idealized C2V geometry of 1g than at the C2
equilibrium geometries of the singlet and triplet states and why
∆EST is larger in 1g than in 1d and 1e. Diol 1d and dimethyl
ketal 1e both lack the ethano bridge that is present in the
ethylene ketal 1g. Without this constraint, the nO orbitals at the
equilibrium geometries of 1d and 1e are less well aligned than
the nO orbital in 1g for SC with ψA (vide infra).
To confirm the dependence of ∆EST on the conformations
about the C2-O bonds in 1d, 1e, and 1g, the effect of
synchronous changes in both O-C2-O-H dihedral angles (θ)
on ∆EST was calculated for diradical 1d (X ) OH). The
geometry optimizations at θ ) 0 and 180° were performed in
value of S2 ) 0.77 for the UB3LYP “singlet” wave function
S
for 1c is intermediate between the values of S2 ) 0.00 for a
S
pure singlet and S2 ) 2.00 for a pure triplet.
T
Houk and Yamaguchi have proposed a simple formula to cor-
rect values of ∆EST, obtained using unrestricted wave functions,
for the effects of spin contamination.19 As shown in Table 1,
the corrected UB3LYP value of ∆EST ) -11.4 kcal/mol for 1c
is 1.6 kcal/mol larger in magnitude than the CASPT2 value but
closer to it than is the uncorrected UB3LYP value of ∆EST
)
-7.1 kcal/mol. The same type of bracketing of the CASPT2
value of ∆EST has also been found in the results of UB3LYP
and CASPT2 calculations on the 1,3-diphenyl derivative of 1c.11
Since the results in Table 1 show the same trends whether
the corrected or uncorrected UB3LYP values of ∆EST are used,
for simplicity, the following discussion is based on the uncor-
rected values. No matter which set of values is employed, |∆EST
(1c)| . |∆EST (1d)|∼|∆EST (1e)|, and |∆EST (1c)| is midway
between the C2 and C2V values of |∆EST (1g)|. Since a pair of
oxygen substituents is attached to C2 in 1d, 1e, and 1g, the
finding that |∆EST (1g)| . |∆EST (1d)| and |∆EST (1e)| is quite
surprising.
The singlet ground states that have been predicted and found
for 2,2-difluoro- and 2,2-dialkoxycyclopentane-1,3-diyls 1 and
24-6,10,11 have been attributed to hyperconjugation (HC), involv-
ing the C-X bonds of the electronegative substituents at C2.
The symmetric combination of the 2p-π AOs at C1 and C3
(ψS) can donate electron density into the pseudo π combination
of the low-lying antibonding C2-X σ* orbitals. As shown
schematically in Figure 1a, this interaction stabilizes ψS relative
to the antisymmetric combination of the 2p-π AOs at C1 and
C3 (ψA), because ψA has the wrong symmetry for interaction
with either the σ or σ* orbitals of the C-X bonds at C2. The
C2V symmetry. For the other C2-O conformations, C2 symmetry
was maintained in the optimizations. A plot of ∆EST versus θ
in 1d is given in Figure 2a.
As shown in Figure 2a, ∆EST is calculated to depend strongly
on the dihedral angle θ. The orientation of the 2p AOs on the
oxygen atoms (nO) affects ∆EST in the manner expected on the
basis of SC. The most negative values of ∆EST occur at dihedral
angles of θ ) 0 and 180°, where the spiroconjugative interaction
between nO and ψA is the strongest. The size of |∆EST| is
smallest at a dihedral angle of θ ) 90°, where nO lies in the
nodal plane of ψA, so that the spiroconjugative interaction
between them vanishes. The degree to which nO mixes with
ψA is apparent from the plots in Figure 3 of the lowest unfilled
(LU)MO in 1d at these three dihedral angles.
The same type of dependence of ∆EST on the O-C2-O-
CH3 dihedral angle was also calculated for 1e (X ) OCH3).
The uncorrected singlet-triplet energy gaps were found to be
∆EST ) -7.2 kcal/mol at θ ) 0°, ∆EST ) -3.4 kcal/mol at θ
) 90°, and ∆EST ) -5.1 kcal/mol at θ ) 180°.
(19) (a) Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. N. Chem. Phys. Lett.
1988, 149, 537. (b) Soda, T.; Kitagawa, Y.; Onishi, H.; Takano, Y.; Shigeta,
Y.; Nagao, H.; Yoshioka, Y.; Yamaguchi, K. Chem. Phys. Lett. 2000, 319,
223-230.
(20) Sherrill, C. D.; Seidl, E. T.; Schaefer, H. F., III. J. Phys. Chem. 1992, 96,
3712-3716.
(21) Simmons, H. E.; Fukunaga, T. J. Am. Chem. Soc. 1967, 89, 5208-5215.
9
576 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Electronic Transitions of Singlet Cyclopentane-1,3-diyls
A R T I C L E S
electronegativity of fluorine, compared to oxygen, presumably
makes the effect of HC on lowering the energy of the HOMO
larger for X ) F than for X ) OR. However, at a C2V geometry,
where nO is aligned for maximal interaction with ψA, the greater
electronegativity of fluorine relative to oxygen should make the
effect of SC on raising the energy of the LUMO more important
for X ) OR than for X ) F.
Table 1 shows that, when the geometry of 1g is constrained
to have C2V symmetry, the calculated value of |∆EST| is larger
than that computed for 1c. Therefore, it appears that the larger
effect of oxygen on SC in the LUMO dominates the larger effect
of fluorine on HC in the HOMO, when the nO orbital is properly
oriented for maximal interaction with ψA.
It should be noted that only at θ ) 0 and 180° is nO or-
thogonal to ψS. At θ ) 90°, where nO is orthogonal to ψA, nO
is best oriented to mix with and destabilize ψS. Therefore, the
O-C2-O-R dihedral angles in 1d, 1e, and 1g affect |∆EST|
by modulating the energy of not only the LUMO but also of
the HOMO. To assess quantitatively the dihedral angle depen-
dence of the HOMO and LUMO energies in 1d, the energies
of these Kohn-Sham orbitals were calculated at different values
of θ for the singlet state at the RB3LYP/6-31G*//UB3LYP/6-
31G* level of theory. The results of these calculations are
displayed in Figure 4.23
Figure 2. Effect of synchronous changes in both O-C2-O-H dihedral
angles (θ) in 2,2-dihydroxycyclopentane-1,3-diyl (1d) on (a) the calculated
singlet-triplet energy gap (∆EST) (9) and (b) the calculated absorption
wavelength (λcalcd) (blue b).
As shown in Figure 4a, the LUMO energy of 1d depends
strongly on the angle θ. The maximum energy is calculated to
occur at an angle of 0° (∆ELUMO ≡ 0 kcal/mol); but, surpris-
ingly, the energy minimum for the LUMO is found not at θ )
90° but at a dihedral angle of θ ) 120° (∆ELUMO ) -12.0
kcal/mol). The LUMO energy is rather insensitive to θ for θ >
90°, where the protons of both hydroxy groups are endo to the
cyclopentane ring, but the energy of the LUMO increases
sharply with decreasing values of θ for θ < 90°.
The asymmetry about θ ) 90° of the plot of ∆ELUMO versus
θ in Figure 4a indicates that ψA is stabilized when the protons
of both hydroxy groups are endo to the cyclopentane ring. This
suggests the existence of an electrostatic interaction between
the electrons in the 2p-π AOs on C1 and C3 and the protons of
the hydroxy groups in the endo conformers (θ > 90°). Such an
interaction may be the reason why the LUMO energy shows a
minimum at 120°. At this angle, each proton is close to one of
the 2p-π AOs. Without this type of interaction, ∆ELUMO should
be the same at θ ) 120° as at θ ) 60°, rather than being 4.6
kcal/mol lower at θ ) 120°.
As expected, the variation of the HOMO energy of 1d with
the dihedral angle θ (Figure 4b) is opposite to the LUMO (Fig-
ure 4a). In going from θ ) 0° to θ ) 90°, the HOMO energy
increases by 4.3 kcal/mol. As already noted, at θ ) 90°, nO is
optimally oriented to interact with ψS via SC, and this mixing
destabilizes the HOMO of 1d. The interaction between nO and
ψS at θ ) 90° is illustrated schematically in Figure 5a, and a
boundary surface plot of the HOMO at θ ) 90° is shown in
Figure 5b.
Figure 3. Boundary-surface plots of the LUMO of 1d (RB3LYP/6-31G*//
UB3LYP/6-31G*) at θ ) 0, 90, and 180°. The plots show that nO mixes
with ψA at θ ) 0° and 180° but not at 90°. The small contribution of oxygen
to the LUMO at θ ) 90° comes largely from the lone pairs on oxygen that
occupy hybrid AOs, rather than the pure 2p AOs that comprise nO.
The uncorrected values of ∆EST in 1d, plotted as a function
of the O-C2-O-H dihedral angles (θ) in Figure 2a, mirror
the uncorrected values of ∆EST in Table 1 at the optimized
geometries of the singlet ground states22 of 1d (θ ) 51.2°), 1e
(θ ) 61.8°), 1g (θ ) 12.0°), and 1g when this diradical is
constrained to have C2V symmetry (θ ) 0.0°). This cor-
respondence shows that the differences between the values of
∆EST for 1d, 1e, and 1g in Table 1 are largely due to the
different values of θ at the equilibrium geometry of each of
these oxygen-substituted singlet diradicals. The O-C2-O-H
dihedral angles control the degree of SC between nO and ψA,
and the mixing between these two orbitals modulates the energy
of the LUMO, which, in turn, affects the size of ∆EST.
Since the fluorine atoms in 1c also have a lone pair of
electrons in an orbital (nF) that can mix with ψA, SC, as well
as HC, presumably plays a role in determining the size of ∆EST
in 1c. As shown schematically in Figure 1, the greater
From θ ) 90°, the HOMO energy of 1d decreases by 8.4
kcal/mol to a minimum at θ ) 180° (∆EΗÃΜÃ ) -4.1 kcal/
mol). The asymmetry about θ ) 90° in the plot of the HOMO
(23) A similar dihedral angle dependence of the energies of the two singly
occupied orbitals was found in UB3LYP/6-31G* calculations on the triplet
state of 1d. Since the energies plotted in Figure 4 are those of Kohn-
Sham orbitals, the changes in the energies of these orbitals are only
qualitatively related to the change in ∆EST that are plotted in Figure 2.
(22) At the optimized geometries of the triplet states, the dihedral angles θ were
calculated to be θ ) 58.2° (1d), 63.2° (1e), and 12.7° (1g). The smaller
dihedral angle at the optimized geometry of each singlet state is consistent
with SC stabilization of the singlet.
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 577

A R T I C L E S
Abe et al.
Figure 4. Effect of synchronous changes in both O-C2-O-H dihedral angles (θ) on the energies (a) of the LUMO (blue 2) and (b) of the HOMO (1)
in 1d. The energy scales in the two plots differ, because the change in ∆ELUMO is 50% larger than that in ∆EHOMO
.
1e with θ ) 0° have optimized O-C-O bond angles of φ )
110.1° and φ ) 119.1°, both of which are considerably larger
than the value of φ ) 105.1° at the optimized C2V geometry of
1g. The values of ∆EST ) -7.4 and -7.2 kcal/mol calculated
for 1d and 1e, respectively, at these geometries are quite close
to the value of ∆EST ) -7.6 kcal/mol for 1g at its optimized
C2V geometry.
Predicted Effect of the Ethylene-Ketal Functionality at
C2 on the Electronic Spectra of Singlet Cyclopentane-1,3-
diyls. Since singlet diradicals 2c, 2d, and 2f are EPR-silent
species at 77 K, an experimental test of the predicted effect of
the ethylene ketal in 2g on the size of the singlet-triplet energy
gap is difficult. However, an experimentally accessible expres-
sion of the predicted effect of the ethylene ketal in 2g on
increasing the difference between the HOMO and LUMO
energies should be found in the electronic absorption spectrum
of 2g when compared to the spectra of 2c, 2e, and 2f.
As expected, when the electronic excitation energies were
computed for the singlet diradicals 1, we found a strong
correlation between the calculated excitation energies and the
calculated magnitudes of ∆EST. As shown in Table 1 and Figure
2, the computed excitation energies increase with decreasing
O-C2-O-R dihedral angle (θ) in the same fashion as |∆EST|.
The first strong absorption bands in the spectra (f > 0.3) are
predicted to correspond to electronic excitation from the HOMO
to the LUMO, and the absorption wavelengths (λcalcd), calculated
at the RCIS/6-31+G*//UB3LYP/6-31G* level of theory, de-
crease in the order 1e (X ) OCH3) > 1g (X,X ) -O(CH2)2O-,
C2) > 1g (C2V) > 1c (X ) F).
Similar substituent effects on ∆EST and on the absorption
bands are also predicted for the 1,3-diphenylcyclopentane-1,3-
diyls 3 (Table 2). Both the RCIS and TD-RB3LYP/6-31G*
methods predict that the λcalcd decrease in the order of 3e (X )
OCH3) > 3g (X,X ) -O(CH2)2O-) > 3c (X ) F).24 This
corresponds to the same type of blue shift that is predicted for
the diradicals without phenyl substituents at C1 and C3.
We have already measured and reported the absorption spectra
of 2c and 2e.6b,c Therefore, to test the prediction that the singlet
diradical 2g would absorb at longer wavelengths than 2c but at
Figure 5. (a) Spiroconjugation (SC) between nO and ψS at θ ) 90° and
(b) a boundary-surface plot of the HOMO at this dihedral angle, showing
the effect of the mixing between nO and ψS by spiroconjugation.
energy versus the O-C2-O-H dihedral angle suggests that
an electrostatic interaction between the electrons in the carbon
2p-π AOs and the protons of the endo hydroxy groups tends to
stabilize not only the LUMO but also the HOMO of 1d.
The dihedral-angle (θ) dependence of the spiroconjugation
(SC) effect on the HOMO and LUMO energies of the exo,exo
conformers of 1d (0° < θ < 90°) (Figure 4) accounts
qualitatively for the increase of 3.0 kcal/mol in the uncorrected
magnitudes of ∆EST (Table 1) on replacing the dimethyl ketal
in 1e (θ ) 61.8°) by the ethylene ketal in 1g (θ ) 12.0°).23
Figure 4 also rationalizes the 1.0 kcal/mol increase in the
uncorrected magnitudes ∆EST between the optimized C2 (θ )
12.0°) and C2V (θ ) 0.0°) geometries of 1g.
The actual changes in the HOMO and LUMO energies of
1g between θ ) 12.0° and θ ) 0.0°, calculated at the RB3LYP/
6-31G* level, are close to those expected from Figure 4. The
energy difference between the HOMO energies at the optimized
C2 and C2V geometries of 1g was computed to be small,
amounting to only 0.2 kcal/mol. In contrast, a significant
difference of 1.6 kcal/mol was calculated between the LUMO
energies at these two geometries. The sizable destabilization of
the LUMO on decreasing θ by just 12.0° is consistent with the
expected effect of increasing SC between nO and ψA by
increasing the overlap between these two orbitals.
The O-C2-O bond angle (φ) might also affect the size of
∆EST in 2,2-dialkoxycyclopentane-1,3-diyls, since this bond
angle is calculated to be slightly smaller at the optimized
geometry of singlet 1g (φ ) 104.4°) than at those of 1d (φ )
108.1°) or 1e (φ ) 108.7°). However, the effect of φ on ∆EST
appears to be minimal. For example, the conformers of 1d and
(24) For more accurate predictions of the UV-vis spectra of 3c, 3e, and 3g, we
would have liked to have performed calculations based on multireference
CI or multireference perturbation theory. Unfortunately, such calculations,
with appropriately large active spaces in the reference wave functions, were
beyond the computational resources available to us.
9
578 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Electronic Transitions of Singlet Cyclopentane-1,3-diyls
A R T I C L E S
Table 2. Calculated Substituent Effects on the Singlet-Triplet Energy Gaps (∆EST) and on the Absorption Spectra of Singlet
1,3-Diphenylcyclopentane-1,3-diyls 3 and Absorption Wavelengths λcalcd and Oscillator Strengths (f) Computed for the HOMO f LUMO
Transition by the RCIS and TD-RB3LYP Methods. Absorption Wavelengths Calculated Are Compared to Those Observed (λonset and λmax
for Singlet Diradicals 2, Which Have a C4-C5 Propano Bridge That Is Absent in 3
)
diradical 3
diradical 2
λcalcd/nm
RCIS (f)b
TD-RB3LYP (f)b
∆EST in kcal/mola
2
corr
X
( S S , ∆EST
)
C
2
C
2v
C
2
C
2v
λonset (λmax) innm
F
3c
3e
3g
-4.7 (0.79, -7.7)
-2.9 (0.90, -5.1)
-4.6 (0.81, -7.6)
563 (1.27)
657 (1.02)
596 (1.16)
636 (0.55)
754 (0.38)
638 (0.53)
2c 580 (525)
2e 650 (571)
2g 590 (514)
OMec
720 (0.93)
612 (1.03)
859 (0.29)
672 (0.44)
-O(CH2)2O-
a Calculations were performed at the UB3LYP/6-31G* level of theory. The values in parentheses indicate the S2 values for the singlet state and the
S
energy gaps computed after correction for spin contamination in the singlet (∆ESTcorr) by using Yamaguchi’s formula (ref 19); ∆EST ) S2 /( S2
-
corr
T
T
S2 ) × ∆EST. The S2 values for the triplet state were calculated to be ca. 2.0 (see Supporting Information). b Calculated oscillator strengths. c At the C2
S
T
minimum for singlet 3e, the methyl groups are exo to the cyclopentane ring. However, at the C2V global minimum, the methyl groups are endo to the
cyclopentane ring, and this conformer was calculated to be 6.2 kcal/mol lower in energy than the C2 geometry (see Supporting Information).
Scheme 1. Synthesis of Azoalkane AZ
Scheme 2. Generation and Reaction of Singlet Diradical 2g.
Figure 6. Experimental electronic absorption spectra of the singlet
diradicals 2, generated in the photodenitrogenation of the corresponding
a significantly shorter wavelength than 2e, we sought to generate
azoalkanes in MCH/MTHF glass at 77 K: (a) 2g (X,X ) -O(CH2)2O-),
2g and to obtain the electronic spectrum of this singlet diradical.
(b) 2e (X ) OMe), and (c) 2c (X ) F). The absorbance scale for spectra
Generation of Singlet Diradical 2g. The precursor azoalkane
AZ (λmax ) 370 nm, ꢀ ) 90 M-1 cm-1 in benzene) was prepared
by the method shown in Scheme 1. The photodenitrogenation
of AZ was performed by irradiation with a 500W Xenon lamp
(λirr ) 370 ( 10 nm), using a monochromator for wavelength
selection. Photolysis of AZ in a degassed benzene solution at
ca. 20 °C for 1 h, afforded a mixture of housane 7 and the two
stereoisomers of the oxygen migration product 8 (Scheme 2).
After allowing the photolysate to stand for 1 h at ca. 20 °C, the
housane 7 was completely converted to 8.
The photodenitrogenation was also performed at -50 °C in
toluene-d8, and direct analysis of the photolysate at -50 °C by
1H and 13C NMR spectroscopy revealed quantitative formation
of the housane 7 (>95%). The housane 7 persisted below 0 °C,
but at higher temperatures it was cleanly converted to rear-
rangement product 8. For example, at 30 °C, the housane 7
completely disappeared within 45 min, with concomitant
formation of 8 (Supporting Information, Figure S1). A similar
rearrangement was observed in the experimental study of the
singlet diradical 2c.6a
a and b is on the left side of the figure; that for spectrum c is on the right.
a methylcyclohexane (MCH)/2-methyltetrahydrofuran (MTHF)
glass at 77 K (λirr ) 370 ( 10 nm). The resulting electronic
absorption spectrum is shown in Figure 6a.
As predicted, the first band in the absorption spectrum of
singlet 2g (λmax ) 514 nm, λonset ) 590 nm) exhibits a sizable
blue shift when compared to that in the spectrum of singlet 2e
(λmax ) 571 nm, λonset ) 650 nm, Figure 6b). The species that
absorbs at λmax ) 514 nm was EPR-silent, as is the case for the
singlet diradicals 2c and e. This finding is consistent with the
prediction that 2g, like 2c and 2e, should have a singlet ground
state.
The difluoro-substituted singlet diradical 2c (X ) F) was also
generated. Its electronic absorption spectrum, obtained under
conditions similar to those used to measure the spectrum of 2g,
showed λmax ) 525 nm and λonset ) 580 nm (Figure 6c). We
have previously reported the spectrum of 2c in an EPA glass to
have λmax ) 530 nm at 77 K.6a
The observed substituent effect on the band onsets (λonset
λ0-0) for the diradicals 2 agrees well with the λcalcd of the model
≈
Formation of housane 7 strongly suggests the intervention
of diradical 2g in the photodenitrogenation of AZ. To measure
the electronic absorption spectrum of this putative singlet
diradical intermediate, the photolysis of AZ was conducted in
diradicals 3 (Table 2). The RCIS computational method
reproduces the experimental data especially well. The RCIS
value of λcalcd ) 657 nm at the global energy minimum for 3e
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 579

A R T I C L E S
Abe et al.
Figure 8. Transient electronic absorption spectra for the singlet diradical
2g at the picosecond time scale (λexc ) 355 nm, 20 ps pulse width).
(as a computationally tractable model for 2c) argued against
this assignment. The CASPT2 calculations predicted the HOMO
f LUMO excitation in the divinyl model compound to occur
at a significantly shorter wavelength than the absorption actually
found in 2c. These computational results, combined with the
long-wavelength absorption that we found in the UV-vis
spectrum of the ketyl radical, derived from 1,1,1-trifluoroac-
etophenone, led us tentatively to assign the strong absorption
in 2c to a parity-forbidden transition, largely localized in the
benzyl chromophores.6a
To assess the effect of the ethylene-ketal unit on the lifetime
of the singlet diradical, we also measured the lifetime of the
transient absorption of 2g at room temperature. No transient
for 2g was detected when a Q-switched Nd:YAG laser (λexc
)
355 nm, 10 ns pulse width) was used for excitation of AZ.
However, using a mode-locked laser (λexc ) 355 nm, 20 ps
pulse width) for excitation of AZ in acetonitrile led to the
observation of a transient absorption. As shown in Figure 8,
singlet diradical 2g (λmax ) 510 nm) was formed within the
duration of the laser pulse. The transient at 510 nm decayed
with a lifetime of ca. 2 ns, leaving a broad absorption (415-
650 nm), which decayed more slowly (6∼10 ns; see the 6000
ps spectrum in Figure 8). The latter species was not detected
by nanosecond laser flash photolysis.
Figure 7. Electronic absorption spectra (solid line) and emission spectra
(dashed line) of (a) the singlet diradical 2c and (b) the singlet diradical 2e
in MCH/MTHF glass at 77 K.
(a C2V geometry, in which both the methyl groups are endo to
the cyclopentane ring) is quite close to the experimental value
of λonset ) 650 nm for 2e.
Fluorescence emission, which mirrored the first electronic
absorption band, was observed for 2c (X ) F, Figure 7a, λexc
) 480 nm) and 2e (X ) OMe, Figure 7b, λexc ) 540 nm) at 77
K. These emission bands clearly indicate that the observed
absorption bands in the visible region are the lowest-energy
electronic transitions of these two singlet diradicals. Fluores-
cence emission from 2g (X,X ) -O(CH2)2O-) was not detected
under similar conditions, which suggests that the singlet diradical
2g may be more photolabile than the singlet diradicals 2c and
2e.
The consistency between the theoretical predictions for the
wavelengths of the HOMO f LUMO excitations in 1 (Table
1) and 3 (Table 2) and the experimental spectra of 2c, 2e, and
2g (Figures 6 and 7 and Table 2) provides a sound basis for the
reassignment of the visible absorption band of the singlet
diradical 2c. Although there were good arguments for assigning
this strong, long-wavelength absorption to the excitation of an
electron from the HOMO to the LUMO of the singlet diradical,
the results of CASPT2 calculations of the excitation energy for
this transition in 1,3-divinyl-2,2-difluorocyclopentane-1,3-diyl
We cannot identify the second transient, but it might arise
from a two-photon process, induced by the 20 ps laser pulse.
As already noted, a high chemical yield (>95%) of the housane
7 was obtained upon preparative photolysis of AZ in acetonitrile
at -20 °C (cf. Scheme 2). Therefore, it seems unlikely that the
longer-lived transient is formed in significant yield by the same
process that leads to the housane 7 in the preparative photolysis
of AZ, unless, like diradical 2g, the longer-lived transient also
gives 7 in very high yield.
The lifetime of the singlet diradical 2g (τ ca. 2 ns) is much
shorter than those of 2e (τ ) 320 ns at 293 K in benzene) or 2f
(τ ) 880 ns in benzene at 293 K and τ ) 1.01 µs in acetonitrile
at 293 K). The dramatic substituent effects on the lifetime of
these singlet diradicals may be attributed to steric repulsion
between the two phenyls and the alkoxy group that is cis to
them in the transition state for cyclization. The effective size
of the alkoxyl groups in 2f, 2e, and 2g decreases in the order
9
580 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004

Electronic Transitions of Singlet Cyclopentane-1,3-diyls
A R T I C L E S
1
1695 cm-1; H NMR (270 MHz, CDCl3) δ 4.24 (s, 4 H), 7.36-8.09
(m, 10 H); 13C NMR (67.8 MHz, CDCl3) δ 66.5 (2 x t), 108.6 (s),
128.5 (4 x d), 130.2 (4 x d), 133.3 (2 x d), 133.9 (2 x d), 192.6 (2 x
s). Anal. Calcd for C17H14O4: C, 72.33; H, 5.00. Found: C, 72.10; H,
4.98.
EtO > MeO . -O(CH2)2O-, and this is, in fact, the order in
which the lifetimes of these three diradicals decrease.
Conclusion
Our calculations predict a strong correlation between the
calculated singlet-triplet energy gaps in 2,2-disubstituted cy-
clopentane-1,3-diyls and the electronic excitation energies of
these singlet diradicals. The combined theoretical and experi-
mental data show that the ethylene-ketal functionality in 2g (X,X
) -O(CH2)2O-) exerts a powerful effect on both the ∆EST
and the longest wavelength absorption in this singlet diradical.
Our calculations indicate that spiroconjugation (SC) is respon-
sible for the remarkable influence of the ketal group in 2g on
increasing |∆EST| and on shifting the first absorption in the
electronic absorption spectrum of this singlet diradical to shorter
wavelengths. The present study provides a sound basis for
reassignment of the strong absorption band of the singlet 2,2-
disubstituted 1,3-diphenylcyclopentane-1,3-diyl derivatives 2 to
excitation of an electron from the HOMO to the LUMO in each
of these diradicals.
The HOMOs of these diradicals, which are formed from the
symmetric combinations (ψS) of the π-NBMOs of the two
benzylic moieties, are stabilized by hyperconjugation (HC) with
the low-lying σ* orbitals of the electronegative substituents, X,
at C2. The LUMOs, which are formed from the antisymmetric
combinations (ψA) of the benzylic NBMOs, are destabilized by
spiroconjugation (SC) with the combination of lone pair orbitals
on X that have the same symmetry as ψA. Both HC and SC
contribute to making the singlet the theoretically predicted and
experimentally found ground state of the diradicals 2c and 2e-
g.
4-Ethylenedioxy-3,5-diphenyl-4H-pyrazol (6). To a solution of 5
(6.6 g, 23.3 mmol) in CHCl3 was added hydrazine monohydrate (1.13
mL, 23.3 mmol) under an Ar atmosphere. The mixture was refluxed
for 24 h. After removal of the solvent (150 mmHg at 30 °C), the residue
was subjected to silica gel chromatography (EtOAc/n-hexane ) 20/
80). Pyrazole 6 (1.50 g) was obtained together with some impurities:
1H NMR (270 MHz, CDCl3) δ 4.29 (s, 4 H), 7.44-7.88 (m, 10 H).
endo-10,10-Ethylenedioxy-1,7-diphenyl-8,9-diazatricyclo-
[5.2.1.02,6]dec-8-ene (AZ). The azo compound AZ was prepared from
the pyrazole (1.50 g, 5.2 mmol) according to the Hu¨nig route.27 AZ
(611 mg, 1.80 mmol, 35%) was obtained as colorless needles after
recrystallization from methanol, mp 123-124 °C: IR (KBr) ν
2865-3090, 1498, 1473, 1445, 1120, 1035 cm-1; 1H NMR (270 MHz,
CDCl3) δ 1.35-1.74 (m, 6 H), 3.17 (t, J ) 6.4 Hz, 2 H), 4.46 (t, J )
6.4 Hz, 2 H), 3.54-3.65 (m, 2 H), 7.35-7.83 (m, 10 H); 13C NMR
(67.8 MHz, CDCl3) δ 25.7 (2 x t), 27.9 (t), 47.8 (2 x d), 65.1 (t), 65.7
(t), 93.7 (2 x s), 127.3 (s), 127.7 (4 x d), 128.0 (2 x d), 128.3 (4 x d),
134.8 (2 x s); UV (benzene) λmax 370 nm (ꢀ 90.2). Anal. Calcd for
C22H22N2O2: C, 76.28; H, 6.40; N, 8.09. Found: C, 76.08; H, 6.31; N,
8.18.
Photodenitrogenation of Azoalkane AZ in Toluene-d8. A solution
of AZ (10.0 mg, 0.030 mmol) in toluene-d8 (1 mL) was irradiated (λexc
370 ( 10 nm) for 3 h at -50 °C. The photolysate was directly analyzed
1
by H and 13C NMR spectroscopy at -50 °C. Only housane 7 was
detected under these conditions (>95%). After warming up to 30 °C,
housane 7 was cleanly converted to the oxygen migration product 8.
3,3-Ethylenedioxy-2,4-diphenyltricyclo[3.3.0.02,4]octane (7): 1H NMR
(270 MHz, toluene-d8) δ 1.42-1.52 (m, 3 H), 1.92-1.95 (m, 3 H),
3.15 (t, J ) 6.5 Hz, 2 H), 3.20-3.22 (m, 2 H), 3.50 (t, J ) 6.5 Hz, 2
H), 7.06-7.19 (m, 6 H), 7.30-7.33 (m, 4 H); 13C NMR (67.8 MHz,
toluene-d8) δ 24.7 (t), 28.3 (2 x t), 41.1 (2 x d), 43.9 (2 x s), 64.3 (t),
64.4 (t), 103.1 (s), 125.2 (2 x d), 127.7 (4 x d), 129.6 (4 x d), 134.7 (2
x d).
Experimental Section
Low-Temperature Spectroscopy. The electronic absorption spectra
were measured at 77 K in a degassed MCH/MTHF glass by using a
BAS USB2000 multichannel spectrophotometer. The fluorescence
spectra were measured at 77 K on a HITACHI F2000 spectrophotom-
eter.
Photolysis of the Azoalkane AZ on a Preparative Scale. A solution
of AZ (180 mg, 0.52 mmol) in toluene (8 mL) was irradiated (λexc 370
( 10 nm) for 15 h at 0 °C. After warming up to ca. 20 °C, the
photolysate was stirred for 3 h in the dark. After removing the solvent
(0.1 mmHg, 0 °C), the stereoisomeric rearrangement products, endo-8
(46.0 mg, 32%) and exo-8 (92.0 mg, 66%), were isolated as oils by
using flash column chromatography on silica gel. The assignment of
Transient Absorption Spectroscopic Measurement. Nanosecond25
and picosecond26 time-resolved difference spectra were obtained by
using the third harmonic of a Nd3+:YAG laser (Continuum Surelite
I-10, λexc ) 355 nm) or of a mode-locked Nd3+:YAG laser (Continuum
PY61C-10, λexc ) 355 nm) for excitation. White light from a Xe-arc
lamp was used for acquisition of absorption spectra of species living
longer than 10 ns. The transient absorption spectra in the time range
from 20 ps to 6 ns were acquired by using continuum pulses with delays
of 20-6000 ps. The latter were generated by focusing the fundamental
laser pulse into a flowing H2O/D2O mixture (1:1 by volume).
Synthesis of Azoalkane AZ. 1,3-Diphenyl-2-ethlenedioxypropane-
1,3-dione (5). A suspension of 1,3-diphenylpropanetrione 4 (9.00 g,
37.9 mmol) and 2-bromoethanol (2.7 mL, 37.9 mmol) in 18 mL of
THF was mixed with K2CO3 (5.20 g, 37.9 mL) and 18 mL of DMSO.
After stirring at 40 °C under an Ar atmosphere for 24 h, the yellow-
brown solution was added to 100 mL of water. The organic layer was
extracted with diethyl ether, dried over MgSO4, and filtered. After the
solvent was removed under reduced pressure (200 mmHg at 20 °C) on
a rotary evaporator, the mixture was subjected to column chromato-
graphic separation on silica gel (EtOAc/n-hexane ) 10/90). The product
5 (7.00 g, 24.7 mmol, 65%) was obtained as a colorless powder after
recrystallization from pentane-ether, mp 119-120 °C: IR (KBr) ν
1
the configurations was performed by H NMR NOE measurements.
(1R*,2S*,6R*)-1,7-Diphenyl-9,12-dioxatricyclo[6.4.0.02.6]dodec-7-
ene (endo-8): IR ν 3083, 3056, 2952, 2866, 1657, 1599 cm-1; 1H NMR
(270 MHz, CDCl3) δ 1.45-2.00 (m, 6 H), 2.43-2.51 (m, 1 H), 3.23-
3.30 (m, 1 H), 3.68-4.03 (m, 4 H), 7.13-7.65 (m, 10 H); 13C NMR
(67.8 MHz, CDCl3) δ 26.4 (t), 27.1 (t), 31.3 (t), 43.5 (d), 53.2 (d),
61.7 (t), 67.6 (t), 84.8 (s), 119.7 (s), 125.9 (2 x d), 126.4 (d), 127.0
(d), 127.8 (2 x d), 128.1 (2 x d), 128.6 (2 x d), 134.2 (d), 137.6 (s),
145.4 (s); Anal. Calcd for C22H22O2: C, 82.99; H, 6.96. Found: C,
82.70; H, 7.04.
(1S*,2S*,6R*)-1,7-Diphenyl-9,12-dioxatricyclo[6.4.0.02.6]dodec-7-
ene (exo-8): IR ν 3088, 3056, 2954, 2866, 1658 cm-1; 1H NMR (270
MHz, CDCl3) δ 0.78-1.82 (m, 6 H), 2.80-2.93 (m, 1 H), 3.55-3.94
(m, 5 H), 7.16-7.40 (m, 10 H); 13C NMR (67.8 MHz, CDCl3) δ 26.9
(t), 28.1 (t), 30.8 (t), 46.4 (d), 52.3 (d), 61.8 (t), 69.3 (t), 88.7 (s), 122.6
(s), 125.9 (d), 126.7 (d), 127.1 (2 x d), 127.8 (2 x d), 128.1 (2 x d),
128.2 (2 x d), 134.4 (s), 141.4 (s), 149.8 (s). Anal. Calcd for
C22H22O2: C, 82.99; H, 6.96. Found: C, 82.75; H, 7.01.
(25) Yoshimura, A.; Nozaki, K.; Ikeda, N.; Ohno, T. J. Phys. Chem. 1996, 100,
1630-1637.
(26) Ohno, T.; Nozaki, K.; Haga, M. Inorg. Chem. 1992, 31, 548-555.
(27) (a) Beck, K.; Hu¨nig, S. Chem. Ber. 1987, 120, 477-485. (b) Nau, W. M.;
Harrer, H. M.; Adam, W. J. Am. Chem. Soc. 1994, 116, 10972-10982.
9
J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004 581

A R T I C L E S
Abe et al.
Acknowledgment. The work in Osaka was supported by a
grant-in-aid for Scientific Research from the Ministry of
Education, Science, Sports, and Culture of Japan. We thank the
Analytical Center of Faculty of Engineering, Osaka University.
The work at the University of Wu¨rzburg was generously
financed by the Deutsche Forschungsgemeinschaft, Volkswagen
Foundation, Alexander-von-Humboldt Foundation, and the
Fonds der Chemischen Industrie. The research at the University
of Washington was funded by grants from the U.S. National
Science Foundation. The work at Basel was supported by the
Swiss National Science Foundation. Dr. Sheila Buxton is
thanked for her help in the nomenclature for AZ, 7, and 8.
Supporting Information Available: Optimized geometries
and energies for the diradicals 1 and 3 and low-temperature
NMR spectra for product 7. This material is available free of
charge via the Internet at http://pubs.acs.org.
JA038305B
9
582 J. AM. CHEM. SOC. VOL. 126, NO. 2, 2004