Photochemical generation and methanol trapping of localized 1,3 and 1,4 singlet diradicals derived from a spiroepoxy-substituted cylcopentane 1,3- diyl

Article abstract of DOI:10.1021/ja9822761

The direct and benzophenone-sensitized photodenitrogenation of the azoalkane 2,3-diazabicyclo-[2.2.1]hept-2-ene-7,3'-spiro-2',2'-diphenyloxirane 1 at ambient temperature (ca. 15 °C) in benzene afforded exclusively the 6- oxabicyclo[3.2.0]hept-1-ene 3. The

Full text of DOI:10.1021/ja9822761

11304
J. Am. Chem. Soc. 1998, 120, 11304-11310
Photochemical Generation and Methanol Trapping of Localized 1,3
and 1,4 Singlet Diradicals Derived from a Spiroepoxy-Substituted
Cyclopentane-1,3-diyl
Manabu Abe,*,† Waldemar Adam,† and Werner M. Nau‡
Contribution from the Institut fu¨r Organische Chemie der UniVersita¨t Wu¨rzburg,
Am Hubland, D-97074 Wu¨rzburg, Germany, and Institut fu¨r Physikalische Chemie der UniVersita¨t Basel,
Klingelbergstrasse 80, CH-4056 Basel, Switzerland
ReceiVed June 29, 1998
Abstract: The direct and benzophenone-sensitized photodenitrogenation of the azoalkane 2,3-diazabicyclo-
[2.2.1]hept-2-ene-7,3′-spiro-2′,2′-diphenyloxirane 1 at ambient temperature (ca. 15 °C) in benzene afforded
exclusively the 6-oxabicyclo[3.2.0]hept-1-ene 3. The labile oxetane 3 is proposed to be formed by the selective
cleavage of the C-O bond in the epoxy ring of the initially photogenerated, spiroepoxy-substituted, localized
cyclopentane-1,3-diyl diradical 1,3-DR to the 1,4-DR diradical and subsequent cyclization of the latter. Even
at -100 °C, the highly strained (strain energy estimated to be ca. 94 kcal/mol) spiroepoxy housane 2 could
not be observed by 1H NMR (600 MHz) spectroscopy. In MeOH, instead of the oxetane 3, the two regioisomeric
MeOH adducts 6 (trapping product of the diradicals 1,3-DR and 1,4-DR) and 6′ (methanolysis product of 3)
were obtained in high yields. For the first time methanol trapping of a cyclopentane-1,3-diyl singlet diradical
was achieved; the required dipolar character is presumably induced in this symmetric species by the polar
methanol medium. Computational work (UB3LYP/6-31G*) reveals that the S-1,3-DR singlet diradical possesses
a singlet ground state (∆EST ca. 1 kcal/mol), for which ring closure (EA ca. 2.6 kcal/mol) to the spiroepoxy
housane 2 requires more activation than epoxide-ring opening (EA ca. 0.7 kcal/mol); the T-1,3-DR triplet
species ring-opens spontaneously (EA ca. 0 kcal/mol) to the corresponding T-1,4-DR. Stereoelectronic effects
dictate the appropriate conformational alignment of the C-O bond with the radical p orbital in the 1,3-DR for
the selective cleavage of the C-O rather than C-C bond in the epoxy ring.
Introduction
For localized cyclopentane-1,3-diyl diradicals, the balance
between through-space and through-bond coupling of the
nonbonding molecular orbitals (NBMO’s, ΨS and ΨA) affects
the singlet-triplet energy gap (∆EST ) ES - ET) and thus
determines the spin multiplicity.1 When the two are of
approximately equal strength, a triplet ground-state applies;
otherwise the singlet is preferred. In such localized 1,3
diradicals (structures A-D) the through-bond interaction is a
sensitive function of the substituents X at the carbon atom
between the two radical sites (Figure 1).2 For X ) H (A) the
through-bond interaction of the symmetric NBMO (ΨS) with
the σ C-H (pseudo-π) orbital dominates, while for X ) F (B)
it is the interaction with the σ* C-F orbital due to the greater
electronegativity of fluorine. These interactions either increase
or decrease the energy of ΨS, and thus the small or large energy
gap between the NBMOs results in a triplet or singlet ground
state for the diradicals A and B.2d
For the parent cyclopentane-1,3-diyl A (X ) H), the triplet
state is favored by 1.10-1.35 kcal/mol according to high-level
ab initio calculation (Table 1).3 Indeed, the triplet ground state
was confirmed by EPR spectroscopy, the first fully characterized
localized triplet diradical.4 Geminal fluorine substitution in the
derivative B favors the singlet state by 6.1 to 9.7 kcal/mol
according to calculations (Table 1).5 The singlet ground state
in the case of the fluorine-substituted derivative B was recently
substantiated by time-resolved spectroscopy for the diphenyl
derivative E.6 This constitutes the first sufficiently persistent
(ca. 80 ns lifetime in n-pentane at room temperature) localized
singlet 1,3 diradical that has been spectroscopically character-
* Corresponding author. Fax: 0049-931-888-4756, e-mail: adam@
chemie.uni-wuerzburg.de.
† Institut fu¨r Organische Chemie der Universita¨t Wu¨rzburg.
‡ Institut fu¨r Physikalische Chemie der Universita¨t Basel.
(1) (a) Diradicals; Borden, W. T. Ed.; Wiley-Interscience: New York,
1982. (b) Dougherty, D. A. Acc. Chem. Res. 1991, 24, 88.
(2) (a) Hoffmann, R. J. Am. Chem. Soc. 1968, 90, 1475. (b) Doubleday,
C., Jr.; McIver, J. W., Jr.; Page, M. J. Am. Chem. Soc. 1982, 104, 6533. (c)
Goldberg, A. H.; Dougherty, D. A. J. Am. Chem. Soc. 1983, 105, 284. (d)
Getty, S. J.; Hrovat, D. A.; Borden, W. T. J. Am. Chem. Soc. 1994, 116,
1521.
(3) (a) Conrad, M. P.; Pitzer, R. M.; Schaefer, H. F., III J. Am. Chem.
Soc. 1979, 101, 2245. (b) Sherrill, C. D.; Seidl, E. T.; Schaefer, H. F., III.
J. Phys. Chem. 1992, 96, 3712.
(4) (a) Buchwalter, S. L.; Closs, G. L. J. Am. Chem. Soc. 1975, 97, 3857.
(b) Buchwalter, S. L.; Closs, G. L. J. Am. Chem. Soc. 1979, 101, 4688.
(5) Xu, J. D.; Hrovat, D. A.; Borden, W. T. J. Am. Chem. Soc. 1994,
116, 5425.
10.1021/ja9822761 CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/27/1998

Singlet Diradicals from a Cyclopentane-1,3-diyl
J. Am. Chem. Soc., Vol. 120, No. 44, 1998 11305
Figure 1. The effect of through-bond interaction of the methylene bridge in cyclopentane-1,3-diyl diradicals on the nonbonding molecular orbitals
(NBMOs, ΨS and ΨA); a triplet ground state (X ) H) or a singlet ground state (X ) F) is preferred as a consequence of the through-bond
interaction by the σ and σ* C-X orbitals with the symmetric NBMO (ΨS).
Table 1. Singlet-Triplet Energy Gaps (∆EST) of Localized
Scheme 1. Possible Transformation of the
Spiroepoxy-Substituted Cyclopentane-1,3-diyl Diradical
1,3-DR
Cyclopentane-1,3-diyl Diradicals
∆EST ) ES - ET
diradical
UB3LYP/6-31G*a
calcd, lit.
exp
A
B
C
D
1.12 [1.20]
-7.07 [-6.48]
1.85 [1.80]
1.10 to 1.35b
Tc
Se
Tg
Si
-(6.1 to 9.7)d
1.5f
-
-1.00 [-0.71]h,i
a C2V-symmetric structures A-C, CS-symmetric structure D from this
work; values in brackets are corrected for zero-point vibrational energy
differences. b Reference 3b. c Reference 4. d Reference 5. e 1,3-Diphenyl
derivative E from ref 6. f Reference 10, footnote 21. g Reference 10,
inferred from the triplet diradical formed by cyclopropane ring opening.
h Vertical energy gap at the singlet geometry (the triplet diradical is a
dissociative state). i This work.
initio calculations and EPR-spectroscopic data10 suggest a triplet
ground state (Table 1), while for D, a singlet ground state has
been established in the present work.
The diphenyl derivative 1,3-DR has been examined in the
present study. As displayed in Scheme 1, a variety of
competitive transformations may be anticipated. Besides cy-
clization of the singlet cyclopentane-1,3-diyl to the highly
strained housane, attractive alternatives are epoxide-ring opening
by cleavage of the C-O or C-C bond to afford the respective
diradicals 1,4-DR and 1,4-DR′. The following pertinent ques-
ized.7,8b However, attempts to trap it by methanol or diylophiles
failed, because fast fluoride ion elimination competed especially
in polar solvents. Nevertheless, alcohol trapping of singlet
diradicals has been reported, e.g., for the delocalized non-
Kekule´-type trimethylenemethanes F8 and the localized donor/
acceptor-substituted trimethylene G.9
Figure 1 illustrates that through the proper selection of
substituents X, the through-bond coupling may be manipulated
such that either a triplet or a singlet ground-state prevails. An
attractive pair is given by the spirocyclopropyl- and spiroepoxy-
substituted cyclopentane-1,3-diyls C and D. For C, both ab
(9) (a) Cram, D. J.; Ratajczak, A. J. Am. Chem. Soc. 1968, 90, 2198. (b)
Yankee, E. W.; Badea, F. D.; Howe, N. E.; Cram, D. J. J. Am. Chem. Soc.
1973, 95, 4210. (c) Yankee, E. W.; Spencer, B.; Howe, N. E.; Cram, D. J.
J. Am. Chem. Soc. 1973, 95, 4220. (d) Howe, N. E.; Yankee, E. W.; Cram,
D. J. J. Am. Chem. Soc. 1973, 95, 4230. (e) Chmurny, A. B.; Cram, D. J.
J. Am. Chem. Soc. 1973, 95, 4237. (f) Danishefsky, S. Acc. Chem. Res.
1979, 12, 66. (g) Griffiths, G.; Hughes, S.; Stirling, C. J. M. J. Chem. Soc.,
Chem. Commun. 1982, 236.
(6) Adam, W.; Borden, W. T.; Burda, C.; Foster, H.; Heidenfelder, T.;
Heubes, M.; Hrovat, D. A.; Kita, F.; Lewis, S. B.; Scheutzow, D.; Wirz, J.
J. Am. Chem. Soc. 1998, 120, 593.
(7) For delocalized singlet diradicals see Bushby, R. J. J. Chem. Soc.,
Perkin Trans. 2, 1985, 1211 and also see ref 8b.
(8) (a) Little, R. D.; Brown, L. M.; Masjedizadeh, M. R. J. Am. Chem.
Soc. 1992, 114, 3071. (b) Berson, J. A. Acc. Chem. Res. 1997, 30, 238.
(10) Jain, R.; McElwee-White, L.; Dougherty, D. A. J. Am. Chem. Soc.
1988, 110, 552.

11306 J. Am. Chem. Soc., Vol. 120, No. 44, 1998
Abe et al.
Scheme 2. Synthesis of Diazene 1 and Its Photolysis Products
tions shall be addressed: Which reaction channel in Scheme 1
dominates? What differences in the chemical reactivity manifest
the singlet and triplet spin states of the various diradicals? Do
the singlet species possess dipolar character? Can such localized
singlet diradicals be trapped by methanol?
(NMR, IR) and elemental analysis (see Supporting Information).
This photoproduct did not persist under acidic conditions or at
temperatures above 40 °C. Thus, the thermal decomposition
of the oxetane 3 (60 °C in benzene, 45 h) led to the allene 4
(78%, conversion 100%), while prolonged irradiation gave the
symmetrical adduct 5 (65%, conversion 100%)15 by photo-
chemical cycloaddition with Ph2CO. These two transformations
clearly establish the 6-oxabicyclo[3.2.0]heptene structure for the
labile oxetane 3. Not even traces of the regioisomeric 7-oxa-
bicycloheptene 3′ were formed.
Results
Preparation of the Diazene 1 and Its Photodenitrogena-
tion. The spiroepoxy diazene 1 was prepared from the
hydrazide11,12 as depicted in Scheme 2. The configurations of
the diastereomers syn-1 (λmax ) 344 nm, ꢀ 230) and anti-1 (λmax
) 352 nm, ꢀ 90), which could be separated by silica gel
chromatography, were determined by NOE experiments. The
stereoisomer with appreciable (2.5%) NOE enhancement be-
tween the exo hydrogen atoms and the aromatic ones was
assigned to the syn configuration.
The azoalkane 1 was photochemically denitrogenated at
15 °C by means of an argon-ion laser (Scheme 2). The
6-oxabicyclo[3.2.0]hept-1-ene 3 was formed exclusively (>95%)
in benzene (5.2 × 10-2 M), independently of the diazene 1
configuration. Also the irradiation conditions had no influence,
since the direct irradiation (333 and 351 nm, 15 min, conversion
100%) or benzophenone (5.2 × 10-2 M) sensitization (364 nm,
5 min, conversion ca. 30%) gave exclusively the labile oxetane
3. A control experiment without Ph2CO (364 nm, 60 min,
conversion only 5%) established that the photodenitrogenation
of the diazene 1 at 364 nm is triplet-sensitized in the presence
of Ph2CO.13
To check for the intervention of the housane 2 (Scheme 2),
the photoreaction (333 and 351 nm, 30 min, conversion 40%)
was performed at -98 °C in a 10:1 mixture of d8-toluene/d6-
benzene, and the reaction mixture was directly analyzed by 600-
MHz 1H NMR and 150-MHz 13C NMR spectroscopy. Even at
this low temperature, only the additional signals of the oxetane
3 were observed.16 These results suggest that the housane 2, if
formed at all, does not persist even at -98 °C to allow NMR-
spectral detection. The formation of the oxetane 3 establishes
that the C-O bond of the epoxy ring rather than the C-C bond
(Scheme 1) suffers ring-opening in the diradical 1,3-DR, which
is assisted through heterolytic cleavage to the 1,4 dipole 1,4-
DP.19 Had homolysis taken place, C-C bond fragmentation
to the regioisomeric 1,4-diradical (1,4-DR′) should have been
(14) Engel, P. S.; Horsey, D. W.; Scholz, J. N.; Karatsu, T.; Kitamura,
A. J. Phys. Chem. 1992, 96, 7524.
(15) The regioisomer was observed as a minor product by 1H NMR
spectroscopy; the product ratio was ca. 91:9.
The strained oxetane 3, which could be recrystallized from
n-hexane at -15 °C, was fully characterized by spectroscopy
(16) Although for the corresponding spirocyclopropane-substituted 1,3
diradical the housane persists (ref 17), the failure to observe the housane 2
derives presumably from the fact that epoxide ring is cleaved ca. 100 times
faster than the cyclopropane ring (ref 18).
(17) (a) Roth, W. R.; Enderer, K. Justus Liebigs Ann. Chem. 1969, 730,
82. (b) Roth, W. R.; Enderer, K. Justus Liebigs Ann. Chem. 1970, 733, 44.
(c) Adam, W.; Oppenla¨nder, T.; Zang, G. J. Org. Chem. 1985, 50, 3303.
(18) Krosley, K. W.; Gleicher, G. J. J. Phys. Org. Chem. 1993, 6, 228.
(11) Marullo, N. P.; Alford, J. A. J. Org. Chem. 1968, 33, 2368.
(12) Platz, M. S.; McBride, J. M.; Little, R. D.; Harrison, J. J.; Shaw, S.
R.; Potter, S. E.; Berson, J. A. J. Am. Chem. Soc. 1976, 98, 5725.
(13) The triplet energies of diazabicyclo[2.2.1]hept-2-ene (DBH) deriva-
tives are around 60 kcal/mol (ref 14) and that of Ph2CO is 68 kcal/mol;
thus, the sensitized process is energywise feasible.

Singlet Diradicals from a Cyclopentane-1,3-diyl
J. Am. Chem. Soc., Vol. 120, No. 44, 1998 11307
Table 2. Photoreactions of the Spiroepoxy Azoalkane 1 in
in both the direct and sensitized photolysis, while in 10% MeOH
(entry 8), a significant drop in the amount of the MeOH adduct
6 was found in the direct photolysis.
Methanola
product distribution (%)b
In the presence of 1,3-cyclohexadiene (entry 5)22 or oxygen
(up to 2 atm) as triplet quenchers almost the same product ratio
as in the direct photolysis (entry 1) was observed. The
insensitivity toward addition of triplet quenchers rules out the
intervention of a long-lived azoalkane triplet state. In fact, since
2,3-diazabicyclohept-2-ene and its derivatives, except some
special cases,23 do not undergo intersystem crossing (ISC),24
we assume that the chemical reactivity derives from the singlet-
entry
1
conditions
6′ () 3)
6
1
2
3
syn 333/351 nm
anti 333/351 nm
syn Ph2CO (0.05 M), 364 nm
anti Ph2CO (0.05 M), 364 nm
syn 1,3-cyclohexadiene (1 M),
333/351 nm
syn 333/351 nm
syn Ph2CO (0.05 M), 364 nm
syn 333/351 nm
78
79
91
91
80
22
21
9
9
20
4
5c
6d
7d
8e
73
87
90
27
13
10
S
excited azoalkane 1*.
Quantum-Chemical Calculations. The assessment of the
ground-state multiplicity of diradicals has remained a compu-
tational challenge over the years.3,5 High-level, ab initio
calculations with configuration interaction for the proper
description of such open-shell species and electron correlation,
e.g., the CISD and CASPT2 methods with large basis sets, are
expensive and inapplicable for the large systems as examined
herein.3,5 Also, experimental studies on absolute singlet-triplet
energy gaps may show a considerable scatter and have often
borne out surprising effects,25 which render the selection of a
reliable computational procedure, i.e., one which reproduces
experimental findings, quite difficult.
a At 15 °C, 36.0 µmol 1 in 0.7 mL of degassed C6H6/MeOH (1:1),
unless otherwise noted; azoalkane consumption was >95% and the mass
balance >88%, except for entry 5 (75%). b Relative yields (normalized
1
to 100%) are based on H NMR (200 MHz) peak areas of the methyl
singlets (δ 3.22 ppm for 6 and 3.05 ppm for 6′); the reproducibility
was (0.8% for independent experiments. c O2 (up to 2 atm) showed
no effect on the product ratios. d The photoreactions (1, 36.0 µmol)
were run at 15 °C in 0.7 mL of degassed MeOH. e The photoreaction
(1, 36.0 µmol) was run at 15 °C in 0.7 mL of degassed C6H6/MeOH
(9:1).
favored due to the stabilization of the unpaired electron at the
benzylic site by phenyl conjugation (Scheme 1).20
Time-resolved (nanosecond time scale) laser-flash spectros-
copy was performed to detect intermediates (λexc ) 351 nm,
25-ns pulse); however, no transient with significant absorption
above 300 nm was observed. This result rules out that the
regioisomeric 1,4-diradical 1,4-DR′, derived from C-C-bond
cleavage of the epoxide ring (Scheme 1), is formed as an
intermediate since such diradicals should be sufficiently long-
lived and should show sufficiently strong absorption in the UV
region (benzyl chromophore) to allow detection by laser flash
photolysis.5,6 Furthermore, no paramagnetic species could be
detected by EPR spectroscopy under matrix isolation (MTHF,
at 77 K), which indicates that a diradical with a triplet ground
state does not intervene under these conditions.4,10
Methanol-Trapping Experiments. The 1,4-diradical (1,4-
DR, Scheme 1) is expected to possess pronounced dipolar
(zwitterionic) character due to the localized negative charge on
the electronegative oxygen atom and allylic stabilization of the
positive charge.19a,21 Consequently, to provide evidence for
dipolar character in the 1,4-diradical (1,4-DR), the photoreac-
tions of azoalkane 1 were performed in the presence of MeOH
to trap such intermediates. As shown in Table 2, in 1:1 MeOH/
C6H6 the methanol adducts 6 and 6′ (see Scheme 2) were
obtained instead of the oxetane 3. The structure of adduct 6′
was determined by oxidation to the ketone 7 (56%) with
pyridium chlorochromate (PCC); the regioisomer 6 expectedly
resisted oxidation (see Supporting Information). A control
experiment showed that the oxetane 3 was exclusively (96%)
converted to the MeOH adduct 6′. Even in the triplet-sensitized
denitrogenation (entries 3, 4, 7), the MeOH adduct 6 was
observed but by 10% (absolute) less than in the direct photolysis
(cf. the 6:6′ product ratios in Table 2). In pure MeOH (entries
6 and 7), a small increase in the 6 trapping product was observed
Density functional theory (DFT) is well-known to handle
properly both open-shell and closed-shell structures,26a and we
have employed this method for the diradical intermediates A-D.
Although the question has been raised whether spin-projected
rather than uncorrected energies should be employed in the
calculation of diradicals by density functional theory,26 we find
that the uncorrected DFT energies reproduce adequately the
experimental data. Hence, the UB3LYP/6-31G* level of theory
(in Gaussian 94)27 provides excellent results on the singlet-
triplet gaps of diradicals for some experimentally examined 1,3-
cyclopentanediyl systems (Table 1).
The calculations confirm a triplet ground state for the parent
1,3-cyclopentanediyl and a singlet ground state for the 2,2-
difluoro derivative. The calculated singlet-triplet energy gaps
for these model species agree well with the data from the
highest-level computations (Table 1). For the spiroepoxy
diradical D, the singlet ground state is favored by 1.0 kcal/mol
(0.7 kcal/mol when corrected for the zero-point vibrational
energies). The preference for singlet-spin multiplicity is much
less pronounced than for the 2,2-difluoro case, as would be
expected from the reduced through-bond interaction imposed
by one alkoxy versus two fluoro groups.
(22) The triplet energy of 1,3-cyclohexadiene is 52 kcal/mol (Kellogg,
R. E.; Schwenker, R. P. J. Chem. Phys. 1964, 41, 2860); that of
2,3-diazabicyclo[2.2.1]hept-2-ene (DBH) derivatives is ca. 60 kcal/mol (ref
14), and thus the triplet quenching is feasible.
(23) (a) Adam, W.; Nau, W. M.; Sendelbach, J.; Wirz, J. J. Am. Chem.
Soc. 1993, 115, 12571. (b) Adam, W.; Fragale, G.; Klapstein, D.; Nau, W.
M.; Wirz, J. J. Am. Chem. Soc. 1995, 117, 12578.
(24) (a) Clark, W. D. K.; Steel, C. J. Am. Chem. Soc. 1971, 93, 6347.
(b) Engel, P. S.; Culotta, A. M. J. Am. Chem. Soc. 1991, 113, 2686.
(25) See, for example, Matsuda, K.; Iwamura, H. J. Am. Chem. Soc.
1997, 119, 7412.
(19) For dipolar character in singlet diradicals see (a) Salem, L.; Rowland,
C. Angew. Chem., Int. Ed. Engl. 1972, 11, 92. (b) Jean, Y.; Salem, L. J.
Chem. Soc., Chem. Commun. 1971, 382. (c) Horsley, J. A.; Jean, Y.; Moser,
C.; Salem, L.; Sevens, R. M.; Wright, J. S. J. Am. Chem. Soc. 1972, 94,
279.
(20) (a) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317. (b)
Bowry, V. W.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1991, 113,
5687. (c) Stogryn, E. L.; Gianni, M. H. Tetrahedron Lett. 1970, 3025. (d)
Ayral-Kaloustian, S.; Agosta, W. C. J. Am. Chem. Soc. 1980, 102, 314.
(21) Platz, M. S. In Diradicals; Borden, W. T., Ed.; Wiley-Interscience:
New York, 1982; pp 195-258.
(26) (a) Goldstein, E.; Beno, B.; Houk, K. N. J. Am. Chem. Soc. 1996,
118, 6036. (b) Cramer, C. J.; Smith, B. A. J. Phys. Chem. 1996, 9664.
(27) Gaussian 94. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P.
M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson,
G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A., Gaussian Inc.: Pittsburgh, PA,
1995 Revision A.1.

11308 J. Am. Chem. Soc., Vol. 120, No. 44, 1998
Abe et al.
Figure 2. Calculated energy profile (B3LYP/6-31G*) for the reactions of the spiroepoxy-substituted cyclopentane-1,3-diyl diradical D; the energies
are relative to the housane.
The reaction pathways (Figure 2) for the singlet and triplet
states of the spiroepoxy diradical D (without the phenyl
substituents) were also examined computationally at the same
level of theory. All geometries were restricted to CS symmetry
for reasons of computational economy. Vibrational frequencies
and zero-point vibrational energy (ZPE) contributions were also
calculated (cf. Supporting Information). The ZPE corrections,
which varied up to 3.5 kcal/mol, are included in the relative
energy values given in Figure 2. Minima and transition state
structures were identified by the absence or presence of one
imaginary vibrational frequency. Exceptions were the singlet
and triplet 1,4 diradicals 1A′′ and 3A′, which indicated an out-
of-plane movement (one small imaginary frequency with A′′
symmetry) as a consequence of the imposed symmetry con-
straints.
Discussion
The mechanism in Scheme 3 is proposed to account for all
the pertinent experimental and computational results: (a) The
DFT calculations suggest that the epoxy substitution at the C-2
position in the cyclopentane-1,3-diyl diradical places the singlet
state below the triplet by 1.0 kcal/mol (0.7 kcal/mol with ZPE
correction); the singlet state S-1,3-DR possesses a small, but
significant, activation energy (0.7 kcal/mol) for epoxide-ring
opening and a larger one (2.6 kcal/mol) for the formation of
the housane 2; for the triplet state T-1,3-DR, no barrier was
found. (b) C-O bond cleavage of the epoxy ring to the 1,4-
DR was observed to give exclusively the oxetane 3 in the direct
and triplet-sensitized photolyses; the regioisomeric 1,4-DR′
diradical derived from C-C bond cleavage of the epoxy ring
was not observed by time-resolved (nanosecond) spectroscopy.
(c) The housane 2 was not detected even at -98 °C. (d) The
dipolar intermediate 1,4-DP was trapped by MeOH to afford
the methoxy alcohol 6; its yield was appreciably higher (10%
absolute) in the direct photolysis compared to the triplet-
sensitized irradiation.
The denitrogenation of the singlet-excited spiroepoxy diazene
S1* in the direct irradiation leads first to the singlet 1,3-diradical
S-1,3-DR (Scheme 1). With very few exceptions, intersystem
crossing (ISC) in such diazenes does not occur;23,24 a long-lived
diazene triplet state is also ruled out since no triplet quenching
was observed (Table 2, entry 5). ISC from the singlet diradical
S-1,3-DR to the triplet species T-1,3-DR is expected to be
inefficient since the singlet state of the parent epoxy-substituted
diradical lies below the triplet (Figure 2). The fate of the singlet
1,3 diradical S-1,3-DR is ring opening to the 1,4 diradical S-1,4-
DR by C-O bond breakage, a process that requires some
activation (0.7 kcal/mol, Figure 2). This is considerably more
attractive than ring closure to the highly strained housane 2 (a
strain energy as high as ca. 94 kcal/mol has been estimated;28
the bond dissociation energy was calculated to be only 15.2
kcal/mol), for which an activation energy of ca. 2.6 kcal/mol
was computed (DFT). Thus, the difference in these activation
The singlet cyclopentane-1,3-diyl diradical D is found to be
an intermediate with a finite lifetime, which in principle should
be trappable. For ring closure to the housane, the singlet
diradical requires an activation energy (EA) of 2.6 kcal/mol,
while for ring opening to the 1,4-allylic diradical, only 0.7 kcal/
mol is necessary. The bond dissociation energy of the spiro-
epoxide housane to the singlet diradical D amounts only to 15.2
kcal/mol (corrected for ZPE contributions, Figure 2), which is
much below the value calculated here for the parent housane
(ca. 38 kcal/mol) and even the spirocyclopropane housane (ca.
29 kcal/mol). The low bond dissociation energy may account
for the fact that the singlet diradical is a real minimum rather
than a transition state.3b,5
In contrast to the singlet diradical, the triplet diradical, when
calculated at the geometry of the singlet diradical, is not a
stationary state, but undergoes spontaneous (no activation) ring
opening to the 1,4 diradical (A′′), since the triplet diradical lies
at the same energy (or slightly above without ZPE corrections)
as the transition state for epoxide ring opening of the singlet
species (Figure 2). Hence, the calculated singlet-triplet energy
gap (Table 1) refers to the geometry of the singlet diradical
(vertical singlet-triplet energy). Note that the singlet and triplet
1,4 diradicals may adopt two different electronic states with
the unpaired electron on the oxygen atom located within the
plane of symmetry (A′′) or orthogonal to it (A′); all states are
close in energy (Figure 2). The A′ singlet state was calculated
as the most stable electronic configuration due to the admixture
of dipolar character (cf. Discussion).
(28) The strain energy of spiroepoxybicyclo[2.1.0]pentane (ca. 94 kcal/
mol) was estimated as the sum of its calculated (AM1 method, ref 29) heat
of formation (syn: 42.5 kcal/mol, anti: 42.3 kcal/mol) and that of an
unstrained C6H8O analogue (-51.6 kcal/mol, ref 30) by using group-
additivity values (ref 31). The strain energy of the housane 9 (ca. 89 kcal/
mol) was estimated similarly by adding its heat of formation (-65.6 kcal/
mol) to that of an unstrained C7H10 analogue (-23.0 kcal/mol, ref 31).

Singlet Diradicals from a Cyclopentane-1,3-diyl
J. Am. Chem. Soc., Vol. 120, No. 44, 1998 11309
Scheme 3. Mechanism of the Direct and Benzophenone-Sensitized Photolysis of the Spiroepoxy-Substituted Diazene 1
energies establishes that epoxide-ring opening is preferred over
housane formation. Entropic factors also favor the epoxide-
ring opening, as the calculated entropies of the transition states
for housane formation (77.9 cal mol-1K-1) and epoxide ring
opening (79.7 cal mol-1K-1) confirm.
The failure to detect the housane 2 even down to -98 °C
(cf. NMR experiments) provides circumstantial evidence for this
mechanistic interpretation. For comparison, the related spiro-
cyclopropane azoalkane 8 cleanly affords the corresponding
housane 9 in the direct photolysis (eq 1),17 which persists even
at +100 °C.
epoxide-ring opening in S-1,3-DR. The final product of the
1,4 diradical in benzene is the oxetane 3 from cyclization.
The triplet diradical T-1,3-DR, produced in the benzo-
phenone-sensitized photolysis, exhibits analogous chemistry, i.e.,
epoxide-ring opening. ISC in T-1,3-DR should be energywise
a feasible process, since the resulting singlet diradical S-1,3-
DR is the ground state; however, the calculations predict a
spontaneous (no activation energy) epoxide-ring opening of the
triplet diradical (Figure 2), and the spin-forbidden ISC process
is too slow to compete. Hence, we assume that all oxetane 3
product in the sensitized photolysis derives from the sequence
T-1,3-DR fT-1,4-DR fS-1,4-DR f3.
What appears at first sight puzzling is why the T-1,3-DR
triplet diradical prefers cleavage of the C-O instead of C-C
bond in the epoxide-ring so that the regioismeric oxetane 3′ is
not observed. Also, time-resolved spectroscopy failed to detect
its precursor, the T-1,4-DR′ species, an attractive structure in
view of stabilization of the spins by allyl and benzyl conjugation.
Since both the C-C and the C-O ring-opening reactions are
symmetry-allowed for the A′′ 1,3-DR species, other factors must
be responsible for this preference. The UB3LYP/6-31G*
computations for the parent system D and, more definitively,
the AM1 calculations29 for the diphenyl derivative provide a
clue in that the energy-favored conformation (Scheme 4) of the
diradical 1,3-DR aligns the C-O bond better (more axially)
for cleavage than the C-C one. Additionally, the bond strength
for a C-O bond is ca. 6 kcal/mol lower than for a C-C bond.31
(1)
The allylic diradical S-1,4-DR also possesses a singlet ground
state (A′). In this case, admixture of dipolar character is held
responsible for the singlet preference. The dipolar character is
most pronounced in the A′ state of the diradical, which displays
an increase of the charge density on the oxygen atom (-0.34)
compared to the triplet A′ (-0.29) state. This charge separation
is accompanied by a concomitant decrease in oxygen spin
density, i.e., singlet A′ (0.82) has a lower oxygen spin density
than the triplet A′ (0.90) state. We consider this trend
significant, since the computational results refer to a gas-phase
model; in solution, solvent polarity should promote a more
pronounced charge separation and hence increase the preference
for a singlet ground state of the diradical 1,4-DR. Such dipolar
stabilization of the 1,4 singlet diradical provides an additional
incentive (besides loss of strain energy, Figure 2) for the
(29) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J.
Am. Chem. Soc. 1985, 107, 3902.
(30) (a) Greenberg, A.; Liebman, J. F. in Strained Organic Molecule;
Wasserman, H., Ed.; Academic Press: New York, 1978. (b) Baird, N. C.
Tetrahedron 1970, 26, 2185. (c) Newton, M. D.; Schulman, J. M. J. Am.
Chem. Soc. 1972, 94, 773.

11310 J. Am. Chem. Soc., Vol. 120, No. 44, 1998
Abe et al.
Scheme 4. AM1-Optimized Conformation of the
Spiroepoxy-Substituted Diradical 1,3-DR
The additional methanol product 6 is proposed to come from
direct trapping of S-1,3-DR. While such a reaction constitutes
an unprecedented event for a localized cyclopentane-1,3-diyl
diradical, we reiterate that the calculations suggest that the S-1,3-
DR species should be
a
minimum-energy structure
(0.7 kcal/mol activation energy for ring opening) with a finite
lifetime to allow trapping. Whether the polar methanol medium
induces zwitterionic character in the S-1,3-DR singlet diradical
and stabilizes it is open to debate,8a but certainly in such a
symmetrical species charge separation is necessary for trapping
by methanol. The structural and electronic differences between
the 1,3 and 1,4 diradicals S-1,3-DR and S-1,4-DR must be
clearly recognized, since only the latter one may exist as a bona
fide 1,4 dipole with the negative charge stabilized by the
electronegative oxygen atom and the positive charge by allyl
resonance.
We conclude that in agreement with theoretical predictions,
our experimental results confirm that the electronegative oxygen
atom in the spiroepoxy group at the C-2 position of the
cyclopentane-1,3-diyl diradical promotes a singlet ground state.
The fate of the singlet diradical S-1,3-DR is not to cyclize to
the housane 2, but to open up the epoxide ring by C-O bond
scission to the 1,4 diradical S-1,4-DR. For the cyclopentane-
1,3-diyl triplet diradical T-1,3-DR, which was generated by
benzophenone-sensitized photolysis, the selective C-O bond
cleavage also takes place to give 1,4-diradical T-1,4-DR, despite
the fact that the regioisomeric T-1,4-DR′ derived from the C-C
bond cleavage would be the thermodynamically better stabilized
triplet. Conformational effects align the C-O better than the
C-C bond and account for this ring-opening selectivity in the
1,3-DR diradical. The methanol-trapping experiments in the
direct and the triplet-sensitized photolyses (Table 2) provide
the first experimental evidence for dipolar character in the
localized 1,3-DR and 1,4-DR cyclopentanediyl diradicals, but
only for the latter is a genuine 1,4 dipole likely.
Moreover, the anomeric effect (in radicals the C-O bond prefers
an axial alignment with a radical center32 due to the stabilizing
interaction of the radical p orbital with σ* C-O bond) and steric
repulsion (geminal diphenyl substitution) are presumably re-
sponsible for the preferred conformation in Scheme 4. Con-
sequently, scission of the C-O bond is kinetically favored and
the 1,4-DR diradical is exclusively generated.33
The most convincing evidence for the dipolar character 1,4-
DP in the singlet 1,4 diradical S-1,4-DR (Scheme 3) comes
from methanol trapping experiments. As much as 10-30%
(Table 2) of the methoxy alcohol 6 was obtained as diradical-
trapped product; the remainder is the regioisomeric methoxy
alcohol 6′, which derives from methanolysis of the oxetane 3
(cyclization product). Thus, the product ratio 6:6′ serves as a
convenient monitor to assess the competition between trapping
(regioisomer 6) and cyclization (regioisomer 6′) of the 1,4
diradical S-1,4-DR. However, if the S-1,4-DR were the only
source of the trapping product 6, the 6:6′ ratio would have to
be the same in the direct and triplet-sensitized photolyses. Yet,
significantly more (ca. 12% absolute) methanol trapping is
observed in the direct process (Table 2). The differences in
the 6:6′ product ratio for the singlet and triplet processes confirm
that the ISC step T-1,3-DR fS-1,3-DR must be inefficient
because otherwise no differentiation between trapping and
cyclization would be observed for the direct and sensitized
photolyses. Indeed, the DFT calculations support this inter-
pretation, since the triplet diradical undergoes spontaneous (no
activation energy) epoxide-ring opening.
Acknowledgment. This paper is dedicated to Prof. M. V.
George (Trivandum) on the occasion of his 70th birthday. We
express our gratitude to the Deutsche Forschunggemeinschaft,
Volkswagen Foundation, the Fonds der Chemischen Industrie,
and the Swiss National Science Foundation for generous
financial support. M.A. thanks the Alexander-von-Humboldt
Foundation for a postdoctoral fellowship (1997-1998).
(31) (a) Benson, S. W., Thermochemical Kinetics. Methods for the
Estimation of Thermochemical Data and Rate Parameters, 2nd ed.; Wiley
& Sons: New York, 1976; p 309. (b) Benson, S. W.; Cruickshank, F. R.;
Golden, D. M.; Haugen, G. R.; O’Neal, H. E.; Rodgers, A. S.; Shaw, R.;
Walsh, R. Chem. ReV. 1969, 69, 279.
(32) (a) Giese, B.; Dupuis, J.; Gro¨ninger, K.; Hasskerl, T.; Nix, M.;
Witzel, T. In Substituent Effects in Radical Chemistry; Viehe, H. G.,
Janousek, Z., MerEÅ nyi, R., Eds.; NATO ASI Ser. C, Vol. 189: Dordrecht,
1986; pp 283-296. (b) Korth, H.-G.; Sustmann, R. Ibid., pp 297-300.
(33) For stereoelectronic effects on cyclopropane-ring cleavage, see (a)
Kochi, J. K.; Krusic, P. J.; Eaton, D. R. J. Am. Chem. Soc. 1969, 91, 1877.
(b) Friedrich, E. C. J. Org. Chem. 1969, 34, 528. (c) Friedrich, E. C.;
Holmstead, R. L. J. Org. Chem. 1971, 36, 971. (d) Friedrich, E. C.;
Holmstead, R. L. J. Org. Chem. 1972, 37, 2546. (e) Reference 20d.
Supporting Information Available: Experimental data and
computational details, i.e., absolute energies, Cartesian coordi-
nates, etc. (14 pages, print/PDF). See any current masthead
page for ordering information and Web access instructions.
JA9822761