Mechanism of ene reactions of singlet oxygen. A two-step no-intermediate mechanism

Article abstract of DOI:10.1021/ja027225p

The mechanism of the ene reaction of singlet (¹Δg) oxygen with simple alkenes is investigated by a combination of experimental isotope effects and several levels of theoretical calculations. For the reaction of 2,4-dimethyl-3-isopropyl-2-pentene, the olefinic carbons exhibit small and nearly equal ¹³C isotope effects of 1.005-1.007, while the reacting methyl groups exhibit ¹³C isotope effects near unity. In a novel experiment, the ¹³C composition of the product is analyzed to determine the intramolecular ¹³C isotope effects in the ene reaction of tetramethylethylene. The new ¹³C and literature ²H isotope effects are then used to evaluate the accuracy of theoretical calculations. RHF, CASSCF(10e, 8o), and restricted and unrestricted B3LYP calculations are each applied to the ene reaction with tetramethylethylene. Each predicts a different mechanism, but none leads to reasonable predictions of the experimental isotope effects. It is concluded that none of these calculations accurately describe the reaction. A more successful approach was to use high-level, up to CCSD(T), single-point energy calculations on a grid of B3LYP geometries. The resulting energy surface is supported by its accurate predictions of the intermolecular ¹³C and ²H isotope effects and a very good prediction of the reaction barrier. This CCSD(T)//B3LYP surface features two adjacent transition states without an intervening intermediate. This is the first experimentally supported example of such a surface and the first example of a valley-ridge inflection with significant chemical consequences.

Full text of DOI:10.1021/ja027225p

Published on Web 01/14/2003
Mechanism of Ene Reactions of Singlet Oxygen. A Two-Step
No-Intermediate Mechanism
Daniel A. Singleton,*,† Chao Hang,† Michael J. Szymanski,† Matthew P. Meyer,†
Andrew G. Leach,‡ Keith T. Kuwata,§ Jenny S. Chen,‡ Alexander Greer,‡
Christopher S. Foote,*,‡ and K. N. Houk*,‡
Contribution from the Department of Chemistry, Texas A&M UniVersity,
P.O. Box 30012, College Station, Texas 77842, Department of Chemistry and Biochemistry,
UniVersity of California, Los Angeles, California 90095-1569, and Department of Chemistry,
Macalester College, Saint Paul, Minnesota 55105-1899
Received June 7, 2002 ; E-mail: singleton@mail.chem.tamu.edu
Abstract: The mechanism of the ene reaction of singlet (1∆g) oxygen with simple alkenes is investigated
by a combination of experimental isotope effects and several levels of theoretical calculations. For the
reaction of 2,4-dimethyl-3-isopropyl-2-pentene, the olefinic carbons exhibit small and nearly equal 13C isotope
effects of 1.005-1.007, while the reacting methyl groups exhibit 13C isotope effects near unity. In a novel
experiment, the 13C composition of the product is analyzed to determine the intramolecular 13C isotope
2
effects in the ene reaction of tetramethylethylene. The new 13C and literature H isotope effects are then
used to evaluate the accuracy of theoretical calculations. RHF, CASSCF(10e, 8o), and restricted and
unrestricted B3LYP calculations are each applied to the ene reaction with tetramethylethylene. Each predicts
a different mechanism, but none leads to reasonable predictions of the experimental isotope effects. It is
concluded that none of these calculations accurately describe the reaction. A more successful approach
was to use high-level, up to CCSD(T), single-point energy calculations on a grid of B3LYP geometries.
The resulting energy surface is supported by its accurate predictions of the intermolecular 13C and 2H
isotope effects and a very good prediction of the reaction barrier. This CCSD(T)//B3LYP surface features
two adjacent transition states without an intervening intermediate. This is the first experimentally supported
example of such a surface and the first example of a valley-ridge inflection with significant chemical
consequences.
Introduction
number of simple reactions, including the ring opening of
cyclopropylidene to form allene,2 pseudorotations in SiH4F- and
For a continuous function of a single variable, there must be
a minimum between any two maxima. When this idea is applied
to reaction coordinate diagrams drawn on paper, it might be
concluded that there must be an intermediate between two
transition states (Figure 1a). Indeed, this idea lies behind rules
for the interpretation of mechanistic data as evidence for
intermediates. In higher dimensions, however, a transition state
is a saddle point, not a maximum, and an energy surface can
readily have two adjacent saddle points with no intervening
minimum (Figure 1b).1 On such surfaces, the reaction path
bifurcates, and a selection between possible products occurs
after the initial transition state. This selection occurs in an area
of the surface referred to as the Valley-ridge inflection.1-5
Theoretical studies have predicted surfaces of this type for a
PH4F,3 1,2-hydrogen migration in H3CO•,4 the photodissociation
of thioformaldehyde,5 reactions of ketyl anion radicals with
methyl halides,6 and bond shifting in cyclooctatetraene.7 Recent
theoretical calculations by Caramella and co-workers have
intriguingly predicted a symmetrical bifurcating surface for
the Diels-Alder dimerization of cyclopentadiene.8 Whenever
symmetry is broken after an initial symmetrical transition
state, the reaction path must bifurcate. This makes numerous
other reactions, such as the dimerizations of cyclobutadiene
and cyclopropene,9 strong candidates for the type of surface in
Figure 1b.
(4) (a) Yanai, T.; Taketsugu, T.; Hirao, K. J. Chem. Phys. 1997, 107, 1137.
(b) Kumeda, Y.; Taketsugu, T. J. Chem. Phys. 2000, 113, 477. (c)
Taketsugu, T.; Kumeda, Y. J. Chem. Phys. 2001, 114, 6973.
(5) Tachibana, A.; Okazaki, I.; Koizumi, M.; Hori, K.; Yamabe, T. J. Am.
Chem. Soc. 1985, 107, 1190.
(6) (a) Bakken, V.; Danovich, D.; Shaik, S.; Schlegel, H. B. J. Am. Chem.
Soc. 2001, 123, 130. (b) Yamataka, H.; Aida, M.; Dupuis, M. Chem. Phys.
Lett. 1999, 300, 583.
(7) Castan˜o, O.; Palmeiro, R.; Frutos, L. M.; Luisandre´s, J. J. Comput. Chem.
2002, 23, 732. See also Wenthold, P. G.; Hrovat, D. A.; Borden, W. T.;
Lineberger, W. C. Science 1996, 272, 1456. Hrovat, D. A.; Borden, W. T.
J. Am. Chem. Soc. 1992, 114, 5879.
† Texas A&M University.
‡ University of California.
§ Macalester College.
(1) Metiu, H.; Ross, J.; Silbey, R.; George, T. F. J. Chem. Phys. 1974, 61,
3200.
(2) Valtazanos, P.; Ruedenberg, K. Theor. Chim. Acta 1986, 69, 281.
Valtazanos, P.; Elbert, S. T.; Ruedenberg, K. J. Am. Chem. Soc. 1986,
108, 3147.
(3) Windus, T. L.; Gordon, M. S.; Burggraf, L. W.; Davis, L. P. J. Am. Chem.
Soc. 1991, 113, 4356. Windus, T. L.; Gordon, M. S. Theor. Chim. Acta
1992, 83, 21.
(8) Caramella, P.; Quadrelli, P.; Toma, L. J. Am. Chem. Soc. 2002, 124, 1130.
(9) Li, Y.; Houk, K. N. J. Am. Chem. Soc. 1996, 118, 880. Deng, Q.; Thomas,
B. E.; Houk, K. N.; Dowd, P. J. Am. Chem. Soc. 1997, 119, 6902.
9
10.1021/ja027225p CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 1319-1328
1319

A R T I C L E S
Singleton et al.
have each been suggested. Trapping of intermediates is some-
times observed,13-15 but not for acyclic alkenes capable of
undergoing the ene reaction. The reaction is stereospecifically
suprafacial.16
The central experimental evidence characterizing the inter-
mediate has come from KIE studies. While the intermolecular
isotope effect observed for d0- versus d12-2,3-dimethyl-2-butene
is small (1.11),11 larger intramolecular isotope effects were
observed on the product distribution in the reactions of the
stereoisomerically d6-labeled 1 and 2.10,17 However, isomer 3
exhibited only a small intramolecular isotope effect on the
product distribution. Stephenson and Orfanopoulos interpreted
these results as strongly implicating an intermediate with the
symmetry of a perepoxide.17 After rate-limiting formation of
the perepoxide (and thus no significant intermolecular isotope
effect), perepoxides derived from 1 (see 5) and 2 can select
between methyl and deuteriomethyl groups in a product
determining step, leading to the observation of an intramolecular
isotope effect. A similar intramolecular isotope effect on the
product distribution was observed by Foote for the reaction of
4.18 The perepoxide derived from 3 would have no such
selection available. A diradical intermediate would be expected
to exhibit intramolecular isotope effects with 1 and 3 but not 2
or 4, in contrast to observations.
Unlike experimental studies, theory has usually failed to
support the perepoxide. Using a combination of GVB calcula-
tions and thermochemical estimates of substituent effects,
Harding and Goddard predicted that perepoxide would be 10-
16 kcal/mol above the energy of simple starting alkenes + 1O2
and that a diradical intermediate would be preferred by 6-8
kcal/mol.19 Tonachini and Robb found that initial formation of
a diradical was favored in an MC-SCF study of the attack of
1O2 on ethylene.20 Yamaguchi and Houk had found that
Figure 1. (a) In two dimensions, transition states must be separated by an
intermediate. (b) In three dimensions, transition states may be adjacent on
an energy surface.
Experimental evidence for the importance of surfaces involv-
ing two adjacent transition states has been lacking, and the
possibility has been ignored in interpreting experimental
observations. We describe here evidence for two sequential
transition states in a complex organic reaction, and we discuss
how this finding resolves conflicting experimental and theoreti-
cal observations in an enigmatic mechanism.
A clear picture of the mechanism of ene reactions of singlet
(1∆g) oxygen (1O2) with alkenes has proved elusive despite
extensive experimental and theoretical study. Although the
concerted ene reaction (a) is a formally allowed pericyclic
process, the absence of significant intermolecular H/D kinetic
isotope effects (KIEs) suggests the transition state occurs prior
to the product determining step and has focused attention on
stepwise mechanisms.10-12 The question has then become the
(10) Kopecky, K. R.; van de Sande, J. H. Can. J. Chem. 1972, 50, 4034.
(11) Gollnick, K.; Hartmann, H.; Paur, H. In Oxygen And Oxy-Radicals In
Chemistry and Biology; Rodgers, M. A. J., Powers, E. L., Eds.; Academic
Press: New York, 1981; pp 379-95.
(12) Stratakis, M.; Orfanopoulos, M.; Foote, C. S. J. Org. Chem. 1998, 63, 1315.
(13) Jefford, C. W.; Rimbault, C. G. J. Am. Chem. Soc. 1978, 100, 295. Jefford,
C. W.; Rimbault, C. G. J. Am. Chem. Soc. 1978, 100, 6437. Manring, L.
E.; Foote, C. S. J. Am. Chem. Soc. 1983, 105, 4710. Wilson, S. L.; Schuster,
G. B. J. Org. Chem. 1986, 51, 2056. Fenical, W.; Kearns, D. R.; Radlick,
P. J. Am. Chem. Soc. 1969, 91, 3396.
(14) Frimer, A. A.; Bartlett, P. D.; Boschung, A. F.; Jewett, J. G. J. Am. Chem.
Soc. 1977, 99, 7977.
(15) (a) Poon, T. H. W.; Pringle, K.; Foote, C. S. J. Am. Chem. Soc. 1995, 117,
7611-7618. (b) Stratakis, M.; Orfanopoulos, M.; Foote, C. S. J. Am. Chem.
Soc. 1991, 32, 863.
(16) Orfanopoulos, M.; Stephenson, L. M. J. Am. Chem. Soc. 1980, 102, 1417.
(17) Grdina, B.; Orfanopoulos, M.; Stephenson, L. M. J. Am. Chem. Soc. 1979,
101, 3111.
(18) Orfanopoulos, M.; Smonou, I.; Foote, C. S. J. Am. Chem. Soc. 1990, 112,
3607-3614.
(19) Harding, L. B.; Goddard, W. A., III J. Am. Chem. Soc. 1980, 102, 439.
(20) Tonachini, G.; Schlegel, H. B.; Bernardi, F.; Robb, M. A. J. Am. Chem.
Soc. 1990, 112, 483.
identity of the intermediate. Diradical or zwitterionic intermedi-
ates (b), perepoxides (c), and π-complexes or exciplexes (d)
9
1320 J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003

Mechanism of Ene Reactions of Singlet Oxygen
A R T I C L E S
unrestricted Hartree-Fock calculations also favor the diradical,
though this study also found that the diradical intermediate
appeared inconsistent with experimentally observed regioselec-
tivities and stereospecificity.21 The “cis effect”, the preference
for hydrogen abstraction from cis-substituted groups in trisub-
stituted alkenes, was interpreted as a consequence of a concerted
rather than a stepwise diradical mechanism.22 These papers and
most later theoretical studies have favored a single ene transition
state with concerted formation of the C-O and O-H bonds.23,24
To make this mechanism consistent with the intramolecular
isotope effect observations, rate-limiting formation of a π-com-
plex or exciplex with the same symmetry as a perepoxide is a
necessary postulate. However, Gorman has argued convincingly
on a kinetic basis that exciplex formation cannot be rate
limiting.25
To summarize, both the diradical and concerted mechanisms
have appeared viable theoretically but cannot account for isotope
effect observations. The perepoxide, while seeming to provide
a clear explanation of the intramolecular isotope effects, has
not been a viable intermediate theoretically. These two observa-
tions add up to a mechanistic paradox. The combination of
experimental and theoretical studies described here suggests a
resolution to this paradox involving a reaction surface with two
sequential transition states.
Figure 2. 13C Kinetic isotope effects (k12C/k13C, 10 °C) for the singlet
oxygen ene reaction of 6.
NMR methods.26 Ene reactions of 6 in 1:1 methanol/2-propanol
at 10 °C were taken to 76 ( 2% and 77 ( 2% conversion by
irradiation with a sunlamp under a flow of oxygen in the
presence of Rose Bengal as sensitizer. Under these conditions,
1H NMR analysis of the crude reaction product exhibited 7 and
traces of 8 as the only observable products (see Supporting
Information). The reaction mixture was treated with excess
trimethyl phosphite at 25 °C to convert 7 to 8, and the unreacted
6 was then recovered by codistillation with the methanol/2-
propanol and partitioning between pentane and water phases to
remove the alcohols, followed by fractional distillation. The
recovered 6 was then analyzed by 13C NMR compared to a
standard sample of 6 from the same synthetic lot that had not
been subjected to the reaction conditions. The changes in 13C
isotopic composition in 6 were calculated by using the isopropyl
methyl carbons as an “internal standard” with the assumption
that their isotopic composition does not change. From the
changes in isotopic composition, the 13C KIEs were calculated
in the previously reported fashion.26 The results are shown in
Figure 2, and there is very good agreement between the two
experiments. Similar small isotope effects were observed at each
of the olefinic carbons, while the isotope effects at the adjacent
methyl and methine carbons were within experimental error of
unity.
Results and Discussion
Intermolecular 13C Isotope Effects. The absence of a
significant intermolecular H/D isotope effect in previous studies
established that there was no hydrogen abstraction in the rate-
limiting step but otherwise provided no information about the
structure or symmetry of the rate-limiting transition state. We
turned to 13C isotope effects to provide a basic isotope effect
characterization of the rate-limiting step. Of particular interest
was the symmetry of the rate-limiting transition state; does it
involve a symmetrical attack on the olefinic carbons, as might
be expected for formation of a perepoxide, or does the 1O2 attack
at the termini of the π-bond to form a diradical? A difficulty in
experimentally probing the transition state symmetry is that
symmetrical alkenes inescapably exhibit symmetrical isotope
effects because of averaging, while unsymmetrical alkenes
necessarily react via unsymmetrical transition states. We
therefore chose the “nearly symmetrical” alkene 6 for study.
While the olefinic termini of 6 are electronically very similar,
differing only in substitution by methyl versus isopropyl groups,
the ene reaction of 6 affords the single product 7 from selective
hydrogen abstraction from the methyls.
1
Qualitatively, these results support an attack of the O2 on
the center of the π-bond, with nearly synchronous bond forma-
tion to the olefinic carbons. In principle, the similar olefinic
isotope effects could also have resulted from a mixture of two
highly unsymmetrical additions to the termini of the π-bond.
However, this would require that the rates of attack on the two
olefinic carbons be comparable, despite the differing sterics,
and each mode of unsymmetrical attack would have to lead to
the same product 7. This seems unlikely. A synchronous attack
1
of O2 in the rate-limiting transition state would appear to be
the most economical explanation for the observed 13C KIEs.
Intramolecular 13C Isotope Effects. Intermolecular KIEs
probe only the rate-limiting step and provide no direct informa-
tion about the rest of a mechanism. The elegant use of
intramolecular isotope effects to overcome this limitation was
pioneered by Dolbier27 and elaborated by Stephenson and
Orfanopoulos.17 A problem, however, in determining intramo-
lecular isotope effects has been the required regio- or stereospe-
cific synthesis of partially labeled materials. A natural abundance
approach to the measurement of intramolecular isotope effects
has been recently reported.28 We describe here the elaboration
of this technique to provide a simple 13C analogue of the
Stephenson experiment.
The intermolecular 13C isotope effects for the 1O2 ene reaction
of 6 were determined combinatorially at natural abundance by
(21) (a) Yamaguchi, K.; Yabushita, S.; Fueno, T.; Houk K. N. J. Am. Chem.
Soc. 1981, 103, 5043. (b) Unlike the seemingly similar triazolinedione and
nitrosoarene dienophiles, ene reactions with 1O2 are notably unselective
regiochemically. See: Stratakis, M.; Orfanopoulos, M. Tetrahedron 2000,
56, 1595.
(22) Houk, K. N.; Williams, J. C.; Mitchell, P. A.; Yamaguchi, K. J. Am. Chem.
Soc. 1981, 103, 949.
(23) Yoshioka, Y.; Yamada, S.; Kawakami, T.; Nishino, M.; Yamaguchi, K.;
Saito, I. Bull. Chem. Soc. Jpn. 1996, 69, 2683.
(24) Okajima, T. Nippon Kagaku Kaishi 1998, 2, 107.
(25) Gorman, A. A.; Hamblett, I.; Lambert, C.; Spencer, B.; Standen, M. C. J.
Am. Chem. Soc. 1988, 110, 8053.
(26) Singleton, D. A.; Thomas, A. A. J. Am. Chem. Soc. 1995, 117, 9357.
(27) Dai, S.-H.; Dolbier, W. R., Jr. J. Am. Chem. Soc. 1972, 94, 3946.
(28) Singleton, D. A.; Szymanski, M. J. J. Am. Chem. Soc. 1999, 121, 9455.
9
J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003 1321

A R T I C L E S
Singleton et al.
balanced by a relative increase in 13C at a nonreacting carbon.
From Figure 3, the experimental pattern of 13C integrations
appears inconsistent with the concerted and diradical mecha-
nisms but is as expected for the perepoxide model. This strongly
supports the conclusion of Stephenson and Orfanopoulos from
intramolecular 2H KIEs that the mechanism involves a species
after the rate-limiting step with the symmetry of a perepoxide.
Assuming the perepoxide model allows the actual intramo-
lecular 13C KIEs to be calculated. For the perepoxide model,
carbons e and one of the geminal methyl groups (say f) should
have identical 13C compositions. Subtracting the integration of
carbon e from that for carbons d + f allows calculation of the
ratio of 13C in carbons a versus d. From this ratio and the ratio
of 13C in carbons b versus c, the relative intramolecular isotope
effects for these pairs of carbons can be calculated (Figure 3).
Theoretical Calculations. (A) First Iteration. Reactions of
1O2 are intrinsically difficult cases for theoretical study, so our
initial efforts were focused on determining what accessible level
of theory would provide a reasonably accurate picture of the
transition structures and energy surface. Toward that end, the
1
ene reaction of O2 with tetramethylethylene was studied in
RHF, CASSCF, and both restricted and unrestricted Becke3LYP
calculations, all employing a 6-31G* basis set.29 By using
tetramethylethylene as the ene component, comparisons of the
experimental KIEs with predicted values from differing calcu-
lational levels could be used to help gauge the accuracy of the
calculated structures. The predicted reaction pathways and
energies are summarized in Figure 4.30 Each calculation makes
a significantly different prediction, and it will be seen that none
satisfactorily accounts for experimental observations.
In contrast to the concerted mechanism deduced from
previous ab initio RHF calculations,23,24 the exploration of the
RHF/6-31G* energy surface for the reaction of tetramethyleth-
ylene here predicts a two-step mechanism.31 In this mechanism,
the rate-limiting Cs-symmetric transition structure 10 leads to
(by an IRC analysis) the perepoxide 11 at 5.9 kcal/mol below
the energy of the starting materials. The perepoxide then
abstracts a hydrogen atom in a second transition structure (12)
to form the ene product.
Figure 3. Relative 13C integrations for 9, the pattern of 13C integrations
expected for three mechanistic models, and the relative intramolecular 13
C
KIEs for formation of 9 assuming the perepoxide model.
Using natural abundance tetramethylethylene, 2,3-dimethyl-
1
1-buten-3-ol (9) was formed by the ene reaction with O2 at 0
°C followed by reduction with NaBH4. 13C NMR analysis of 9
was carried out using the recently developed methodology for
the accurate and precise measurement of the relative integrations
of pairs of peaks within a spectrum.28 The results are shown in
Figure 3. Of the carbons that were originally olefinic in
tetramethylethylene, the carbinol carbon (c) of 9 was found to
contain a greater abundance of 13C. For the carbons that started
as methyl groups in tetramethylethylene, a relatively greater
abundance of 13C was found in the geminal methyl groups d
and f of 9.
The interpretation of the integration results is complicated
because their translation into KIEs depends on an initial choice
of mechanistic model. The approach that must be taken, then,
is to assume a mechanistic model and then see if the observed
13C integrations are consistent with the model. Three models
are considered here: (A) concerted, in which intermolecular
and intramolecular selectivity are determined simultaneously;
(B) diradical, in which rate-limiting unsymmetrical addition of
1O2 to the termini of the π-bond is followed by a selection
between the two distal methyl groups in the product-determining
step; and (C) perepoxide, in which a rate-limiting symmetrical
For CASSCF calculations, an 8 electron/6 orbital active space
was used to initially explore the energy surface. Important
(29) 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. J.; Millam, 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.; Rega, N.; Salvador,
P.; Dannenberg, J. J.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; 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.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M.
W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople,
J. A. Gaussian 98, revision A.11.2; Gaussian, Inc., Pittsburgh, PA, 2001.
(30) The UB3LYP energies of open-shell singlets were corrected by spin
projection. See: Yamaguchi, K.; Jensen, F.; Dorigo, A.; Houk, K. N. Chem.
Phys. Lett. 1988, 149, 537. Yamanaka, S.; Kawakami, T.; Nagao, H.;
Yamaguchi, K. Chem. Phys. Lett. 1994, 231, 25. For a caveat on this
procedure, see: Wittbrodt, J. M.; Schlegel, H. B. J. Chem. Phys. 1996,
105, 6574. All of the open-shell singlet species have S2 values of
approximately 1. Projection lowered the energies by 1-4 kcal/mol. The
projection procedure was necessary for UB3LYP to obtain a reasonable
value for the singlet-triplet gap in O2.
1
attack of O2 is followed by a product-determining selection
between methyl groups that started cis in tetramethylethylene.
To predict qualitatively the pattern of 13C relative integrations
expected for these models, we start by assuming that k12C/k13C
would be >1 if a carbon is undergoing loss or gain of a bonded
atom in a step but that otherwise k12C/k13C would be unity. When
k
12C/k13C is >1 at a position, the relative amount of 13C in that
position in the product would be decreased to <1. This leads
to the predicted pattern of 13C relative integrations for the three
model mechanisms shown in Figure 3. It should be noted for
these predictions that when a selection occurs between two
methyl groups in a product-determining step, the decreased rate
of reaction of a 13C at a reacting center will necessarily be
(31) The RHF calculations predict that a very loose π-complex is formed without
a barrier on the approach of singlet oxygen to the pi bond of tetrameth-
ylethylene. The energy of the π-complex was -1.1 kcal/mol relative to
those of the separate alkene and singlet oxygen. After allowance for basis
set superposition error and entropy, this “complex” is unlikely to correspond
to a free-energy minimum at 25 °C.
9
1322 J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003

Mechanism of Ene Reactions of Singlet Oxygen
A R T I C L E S
Figure 4. Mechanisms predicted at differing theoretical levels for the reaction of 1O2 with tetramethylethylene. Numbers in parentheses are energies of
transition structures relative to the starting materials in kcal/mol, including zpe.
structures were then reoptimized using a 10 electron/8 orbital
active space to include σ/σ* orbitals for C-H. The CASSCF
pathway differs from the RHF pathway by involving a nonleast-
motion attack of the 1O2 on the alkene. The transition structure
13 is highly unsymmetrical, and there is asynchronous formation
of the C-O bonds of the perepoxide 14. However, the
perepoxide 14 and product-forming transition structure 15 are
similar to the corresponding RHF structures. The product-
forming step is predicted to be rate-limiting at the CASSCF-
(10e,8o)/6-31G* level. However, single-point QCISD/6-31G*
energies for the CASSCF structures predict the first step to be
rate limiting, placing 13, 14, and 15 at 2.0, -8.6, and -8.3
kcal/mol, respectively, relative to the starting materials.
The unrestricted Becke3LYP (UB3LYP) pathway also in-
carbons and the attacking oxygen atom. The resulting energy
surface (see Supporting Information) exhibits a Cs-symmetric
channel for attack of the oxygen on the olefinic carbons. This
channel leads toward the Cs perepoxide, though the reaction
path would bifurcate shortly before the perepoxide to form the
equivalent unsymmetrical products. A similar energy surface
will be discussed in detail in the next section. Attack of the 1O2
on tetramethylethylene would presumably face a free-energy
barrier at some relatively large separation of the oxygen from
the alkene, though no free-energy maximum was found in a
limited exploration of the surface out to a 3.8 Å separation
1
between O2 and tetramethylethylene. There is no point in an
exhaustive search for the free-energy saddle point, as entropy
1
calculations on the flat surface at large O2/alkene separations
1
volves a highly unsymmetrical attack of the O2, but in this
are questionable.
case, the transition structure 16 leads to the formation of the
diradical 17. In the UB3LYP calculation, the perepoxide 19 is
a transition structure for interconversion of the regioisomeric
diradicals 17 and 20.32
In restricted Becke3LYP (RB3LYP) calculations, there is no
potential energy barrier for attack of 1O2 on tetramethylethylene
and the only transition structure that could be located on the
potential energy surface was the Cs-symmetric perepoxide 21.
This transition structure connects regioisomeric product hydro-
peroxides. The RB3LYP surface was explored systematically
in a grid search by fixing the distance between the olefinic
Inter- and intramolecular KIEs based on structures 10, 12,
13, 15, and 16 were predicted from the scaled theoretical
vibrational frequencies33 by the method of Bigeleisen and
Mayer.34 Tunneling corrections were applied for steps involving
oxygen attack on the alkene using the one-dimensional infinite
parabolic barrier model.35 Because of the questionable accuracy
of simplistic tunneling corrections for hydrogen-abstraction
steps, no tunneling correction was applied in these cases. A more
(33) For the calculations, the program QUIVER was used (Saunders, M.; Laidig,
K. E.; Wolfsberg, M. J. Am. Chem. Soc. 1989, 111, 8989) with Becke3LYP
and RHF frequencies scaled by 0.9614 and 0.91, respectively (Scott, A.
P.; Radom, L. J. Phys. Chem. 1996, 100, 16502).
(34) (a) Bigeleisen, J.; Mayer, M. G. J. Chem. Phys. 1947, 15, 261. (b)
Wolfsberg, M. Acc. Chem. Res. 1972, 5, 225.
(32) Spin-projected UB3LYP/6-311++G** puts 18 1.2 kcal/mol above the
energy of 17 and 19 5.2 kcal/mol above the energy of 17, virtually identical
to the spin-projected UB3LYP/6-31G* energetics.
(35) Bell, R. P. The Tunnel Effect in Chemistry; Chapman & Hall: London,
1980; pp 60-63.
9
J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003 1323

A R T I C L E S
Singleton et al.
Table 1. Predicted versus Experimental Intermolecular and
Intramolecular Isotope Effects for the Ene Reaction of Singlet
Oxygen with Tetramethylethylene
equilibration (5.4 kcal/mol; see Figure 4) is higher than that
for product formation (1.3 kcal/mol). The same is true in CCSD-
(T)/6-31G* single-point energy calculations, where the 17/20
equilibration barrier increases to 8.2 kcal/mol, so the equilibra-
tion of diradicals would not be expected. Despite this, the isotope
effects for the equilibration mechanism were calculated for
comparison with experiment. For this calculation, it was
assumed that the equilibration of 17 and 20 is fast but that
rotation about the C-O bond in 17 and 20 is slow; these “best-
case” assumptions give the equilibrating 17/20 the same
effective symmetry as that of a perepoxide. This leads to
predictions of intramolecular KIEs (Table 1) that qualitatively
fit the pattern of the experimental values, with overall better
predictions than the other calculations. However, the quantitative
disagreement of the calculated KIEs versus experiment is still
substantial, and the lack of theoretical support for the equilibra-
tion of 17 and 20 makes this mechanism unlikely.
The predicted intermolecular 13C KIEs for all of the calcula-
tions were very far from the experimental values. Because there
is a significant history of accurate 13C KIEs predictions for steps
not involving hydrogen transfer,38 this observation is most
dissuasive of the ability of the calculations to depict accurately
the reaction surface. Our conclusion is that none of these
calculations adequately represent the reaction surface.
(B) Second Iteration. The previous observations indicate that
an adequate description of the energy surface for these reactions
requires theoretical methodology that includes substantial
dynamic configuration interaction. Because the required levels
of theory are not readily practical for full geometry optimization,
we turned to a composite procedure. A 15 × 15 grid of the
reaction surface for attack of 1O2 on cis-2-butene and on
tetramethylethylene was formed by fixing the distance of the
attacking oxygen from the olefinic carbons, distances C1-O1
(X) and C2-O1 (Y) in 22, at a series of values and then
optimizing all other variables in restricted B3LYP/6-31G*
calculations. High-level single-point energy calculations were
then carried out for each geometry on the grid (after eliminating
points that are redundant by symmetry).
UB3LYP UB3LYP CCSD(T)//
c
RHF CASSCF no equila equilb
B3LYP
experimental
intermolecular
olefinic 13
d0/d12
C
1.021 1.039d 1.034d 1.034d 1.007 1.005-1.007e
0.92 0.97
1.10
intramolecular
1.026 1.018 0.968 1.016
1.10
1.113 1.11f
olefinic 13
reacting
C
g
g
1.011(4),
1.008(1)
1.009(6),
1.013 1.015
h
1.011
methyl 13
2H KIE in 1
2H KIE in 2
2H KIE in 3
C
1.009(3)
1.38-1.41i
1.4j
3.13 5.14
3.40 5.90
1.04 1.29
1.68
1.21
1.69
1.69
2.11
1.03
g
g
1.045 1.04-1.09i
a Assuming no equilibration of 17 and 20 and that C-O rotation in 17
allows both distal methyl groups to react. b Assuming rapid equilibration
of 17 and 20 and that C-O rotation in 17 and 20 is slow. c See text for
description. d On the basis of an average of the predicted olefinic KIEs.
e On the basis of the KIEs observed for 6. f See ref 11. g Not predictable
2
from transition state theory. See ref 45 for a dynamical analysis of the H
KIEs. h The prediction cannot be compared with the experimental KIE listed
because the experimental number is based on a conflicting mechanistic
assumption. i See ref 17. j See ref 10.
complete treatment would normally raise the isotope effect, so
the calculated KIEs in the absence of tunneling are useful as
lower-bound predictions. The results are summarized in Table
1. The absence of an intermediate for the RB3LYP pathway
precludes calculation of intramolecular isotope effects by
transition state theory.
Qualitatively, the mechanisms predicted by the RHF and
CASSCF results lead to predicted KIEs that are largely in the
same direction as the experimental observations. Quantitatively,
however, the KIE predictions are poor. In contrast to the small
observed 13C KIEs for the olefinic carbons of 6, both calcula-
tions predict large intermolecular 13C KIEs. The pattern of
intramolecular KIEs predicted from the RHF and CASSCF
pathways fits with the experimental values, but in every case,
the predicted KIE is far too large. These results suggest that
the calculated geometries for both pathways are at least
quantitatively inaccurate.
In evaluating the UB3LYP diradical mechanism, we consid-
ered two distinct mechanistic possibilities. The first possibility
is that the diradical 17 acts in accord with the classical
expectations that Stephenson assumed. That is, it is assumed
that 17 does not equilibrate with 20 and that in 17 there is an
equal chance of hydrogen abstraction at the two distal methyl
groups. As shown in Table 1, the predicted isotope effects for
this mechanism disagree with experiment in several ways. Most
notably, the predicted intramolecular olefinic 13C KIE is in the
wrong direction and the predicted intramolecular 2H KIEs have
the wrong pattern versus experiment. This mechanism is
excluded.
The second possibility to consider for the UB3LYP diradical
mechanism is that the diradicals 17 and 20 could equilibrate
via the perepoxide transition structure 19. Theoretical and
experimental support for a similar equilibration of diradicals
has been presented for ene reactions of triazolinediones36 and
nitrosoalkenes.37 However, the UB3LYP calculations do not
predict such a mechanism here; the UB3LYP barrier for
Composite grid searches of this type are subject to some
dangers, and these are considered here briefly before discussing
the results. In any high-level single-point energy calculation on
a low-level geometry, an intrinsic assumption is that the error
in the low-level geometry is systematic, relative to the hypo-
thetical high-level optimized geometry. This process can fail if
the low-level and high-level surfaces differ qualitatively. The
1
B3LYP surface overestimates the energy of O2 but otherwise
closely resembles surfaces calculated at higher levels, particu-
larly as the 1O2 draws close (X or Y < 2.3 Å) to the alkene (see
Supporting Information). This supports the validity of using the
B3LYP grid geometries. Other dangers arise from the limited
two-dimensional exploration of a 3N-6 dimension surface.39 The
(36) Singleton, D. A.; Hang, C. J. Am. Chem. Soc. 1999, 121, 11885.
(37) Leach, A. G.; Houk, K. N. Chem. Commun. 2002, 1243.
(38) Meyer, M. P.; DelMonte, A. J.; Singleton, D. A. J. Am. Chem. Soc. 1999,
121, 10865.
9
1324 J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003

Mechanism of Ene Reactions of Singlet Oxygen
A R T I C L E S
The CCSD(T)//B3LYP (+ zpe) surface in Figure 5 predicts
a symmetrical attack of the 1O2 through a relatively early
transition state 24 [X ) Y ) 2.15 Å for cis-2-butene and 2.38
Å for tetramethylethylene (29)]. At this first transition state,
hydrogen abstraction has not yet started; in fact, the reaction
has not yet “decided” which hydrogen to abstract. From 24,
the reaction path appears to lead toward perepoxide-like structure
25. However, 25 is not an intermediate! There is no enthalpic
or entropic barrier for its decomposition.40 This “perepoxide”
is a transition state, formally connecting the two product valleys
27 and 28. These products are formed from 25 by hydrogen
abstraction, but as described above, C-H bond breaking does
not become significant until there is substantial desymmetriza-
tion of the Cs-symmetric 25. The minimum energy path
connecting reactants with products becomes unstable between
rate-limiting transition state 24 and perepoxide transition state
25 at the valley-ridge inflection.1-5,41 Intermolecular selectivity
between alkenes and product selectivity in 3 would be decided
in the rate-limiting transition states (24 or 29), while product
selectivity for reaction among groups that started cis on the
alkene would be decided in the area of the valley-ridge
inflection.
The use of CCSD(T) calculations here is subject to concern
over whether a single-reference electron-correlation procedure
is sufficient to represent the reaction surface. We examined this
question in two ways. First, the T1 diagnostic of Lee was
calculated for each of the grid points.42 For this diagnostic, T1
values > 0.02 suggest that nondynamical correlation effects are
large enough to make single-reference methods limited to single
and double excitations potentially unreliable. The inclusion of
a perturbative correction for triple excitations, as in CCSD(T)
here, will satisfactorily compensate for the deficiencies of a
single-determinant reference under certain circumstances,43 but
the T1 value remains a flag indicative of possible danger. The
T1 values (see Supporting Information) are unsurprisingly low
(0.012-0.017) in the product valleys, but they are also low
(0.013-0.020) as the 1O2 approaches the alkene up to the rate-
limiting transition state 24. However, the T1 values are increased
(0.021-0.031) in the area between 24 and 25 in Figure 5, and
T1 reaches a maximum of 0.036-0.037 at the edge of the “cliff”
above the product valley (from X ) 1.55 Å, Y ) 1.95 Å to X
) 1.75 Å, Y ) 2.15 Å in Figure 5). The second way in which
Figure 5. Two-dimensional grid of the CCSD(T)/6-31G*//RB3LYP/6-31G*
+ zpe energy surface for reaction of 1O2 with cis-2-butene. X and Y are
C1-O1 and C2-O1 distances, as defined in 22. The vertical energy scale is
in kcal/mol relative to the energy of the starting materials. The product
valleys for 27 and 28 are cut off at -20 kcal/mol but actually extend to
-41 kcal/mol.
most critical issue is the question of whether the grid geometries
are germane to the reaction path. To address this issue for the
B3LYP surface for 1O2/tetramethylethylene, the minimum-
energy path (MEP) was followed downhill, in mass-weighted
coordinates, starting from a Cs-symmetric structure with X ) Y
) 2.85 Å and proceeding toward 21. The resulting path pro-
ceeded through all of the relevant Cs-symmetric grid geometries.
Starting from 21 and proceeding toward products, the MEP
involves no significant hydrogen transfer until X or Y increases
from 1.67 Å to over 2.0 Å. Nearer to 21 than this, the MEP
very closely follows the grid geometries. These observations
support the relevance of the grid geometries to the reaction path.
The grid for the 1O2/cis-2-butene reaction was examined using
several levels of single-point calculations, including MP4(sdtq)/
6-31G*, MP4(sdtq)/6-311+G**, and CCSD(T)/6-31G*. All
three give very similar results (see Supporting Information). The
CCSD(T)/6-31G* results are shown pictorially in Figure 5. The
grid for attack of 1O2 on tetramethylethylene was studied only
partially at CCSD(T)/6-31G*, and the surface is analogous to
(40) As the structure departs from the saddle point along the MEP, a low-energy
rotational mode in the nonreacting methyl group is loosened, and overall,
the entropy is increasing (on the basis of B3LYP frequencies, 298 K). There
is thus no indication of a free-energy barrier to the decomposition of the
perepoxide.
1
the O2/cis-2-butene surface (see Supporting Information).
(41) Kraus, W. A.; DePristo, A. E. Theor. Chim. Acta 1986, 69, 309.
(42) (a) Lee, T. J.; Rice, J. E.; Scuseria, G. E.; Schaefer, H. F., III. Theor. Chim.
Acta 1989, 75, 81. (b) Lee, T. J.; Taylow, P. R. Int. J. Quantum Chem.
Symp. 1989, 23, 199.
(43) (a) Kraka, E.; Cremer, D. J. Am. Chem. Soc. 1994, 116, 4929. (b)
Wesolowski, S. S.; Fermann, J. T.; Crawford, T. D.; Schaefer, H. F., III J.
Chem. Phys. 1997, 106, 7178. (c) Kraka, E.; Cremer, D. J. Am. Chem.
Soc. 2000, 122, 8245. (d) Lindh, R.; Lee, T. J.; Bernhardsson, A.; Persson,
B. J.; Karlstrom, G. J. Am. Chem. Soc. 1995, 117, 7186. (e) Cramer, C. J.
J. Am. Chem. Soc. 1998, 120, 6261.
(39) An obvious but surmountable problem is that any particular values of X
and Y do not by themselves uniquely define the molecular geometry. It is
therefore necessary to describe in detail the process by which the structures
were obtained. Starting from a Cs-symmetric structure with X ) Y ) 2.85
Å, we obtained a series of Cs-symmetric structures by decreasing the fixed
X and Y by 0.1 Å and reoptimizing the geometry at each step. Series of
nonsymmetrical structures were obtained from the Cs-symmetric structures
by successively decreasing X by 0.1 Å, leaving Y constant, and reoptimizing
the geometry at each step.
9
J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003 1325

A R T I C L E S
Singleton et al.
the multireference character of the wave function was examined
was by looking at the stability of the restricted singlet wave
function. The wave function was unsurprisingly UHF unstable
The approximate transition structure 34, leading to reaction of
one of the initially cis methyl groups, is a synchronous attack
on the olefinic carbons. In 34, there is a close contact between
O2 and hydrogens on the cis methyl groups, but there is no
C-H bond breaking yet at this stage. In contrast, approximate
transition structure 35, leading to reaction of the lone trans
methyl group, is much later than 34 and involves an asynchro-
nous attack on the olefinic carbons. Structure 35 is predicted to
be 2.8 kcal/mol higher in energy than structure 34 (CCSD(T)/
6-31G*//B3LYP/6-31G* + zpe), so the calculations correctly
predict a substantial cis effect.
1
when the O2 is distant from the alkene (the same is true for
1O2 by itself), but the wave function was uniformly both singlet
and triplet stable in the areas of the surface where T1 was high.
The two tests are therefore somewhat contradictory.44 Overall,
it is apparent that comparison with experimental observations
will be necessary for a decisive validation of the CCSD(T)
surface.
Experimental Support. Unfortunately, selectivity in the area
of the valley-ridge inflection of Figure 5 cannot be analyzed
using transition state theory. Without a barrier, transition state
theory simply does not apply, and a dynamics approach is
required. This is the subject of an accompanying paper, which
concludes that the intramolecular 2H KIEs in 1 and 2 are
consistent with dynamical selectivity on the type of surface in
Figure 5.45
Isotope effects allow a careful scrutiny here of the energy
surface in the area of the rate-limiting transition state. The
Significance. The mechanism supported here for reactions
of cis-2-butene and tetramethylethylene might be described as
concerted in normal parlance, in that there is no intermediate
between starting materials and product. In this respect, previous
theoretical calculations predicting a concerted reaction were
correct. However, the inability of previous calculations to
explain experimental observations highlights the problem with
the use of the word “concerted” as a synonym for “no
intermediate”. The meaning of the scientific word “concerted”
arises from its more catholic meaning of events happening in
unison, which is certainly not a satisfactory description here.
Dewar described reactions as “two stage” when the bonding
changes in a concerted reaction are unequal at the transition
state.50 The “two-stage” term may apply here, but it is quite an
2
intermolecular H and 13C KIEs for the tetramethylethylene
reaction as well as the intramolecular 2H KIE in 3 were predicted
from transition state theory using a numerically calculated
CCSD(T) frequency along the reaction coordinate and the
B3LYP frequencies for all other modes in the rate-limiting grid
saddle point 29. The results are given in Table 1. In contrast to
the other theoretical methods examined, the composite CCSD-
(T)/6-31G*//B3LYP/6-31G* grid results in excellent KIE
predictions. In each case, the calculated KIEs are within error
of the experimental values, and the level of agreement is quite
striking.
The predicted energy barrier for reaction of 1O2 with
tetramethylethylene at the rate-limiting grid saddle point is -2
kcal/mol (CCSD(T)/6-31G*//B3LYP/6-31G* + zpe), and the
predicted entropy of activation (B3LYP/6-31G*) is -34 eu. This
is in very good agreement with the experimental values -0.74
kcal/mol and -30.1 eu, respectively.25
Our conclusion from these results is that 29 is an accurate
depiction of the rate-limiting transition state for reaction of 1O2
with tetramethylethylene and that the CCSD(T) grid provides
an accurate representation of the energy surface in the area
around 29. This supports the accuracy of the CCSD(T) grid for
the rest of the surface.
Cis Effect. With trisubstituted alkenes and enol ethers, there
is a notable and general trend for singlet oxygen to abstract
allylic hydrogens from the disubstituted side of the alkene.22,46-49
The simplest example of this “cis effect” is the reaction of 30,
where 93% of the products result from reaction of the initially
cis methyl groups. (The isotope effect here is nearly negligible.47)
1
understatement. In this way, the O2 ene mechanism is related
to a number of thermal hydrocarbon rearrangement reactions,
such as 1,3-sigmatropic shifts in vinylcyclopropanes and
vinylcyclobutanes.51-53 In these rearrangement reactions, there
is a clear separation between an initial bond-breaking stage and
a final bond-making stage. The predicted potential energy
surfaces typically exhibit only a single saddle point, but between
the stages, there is a nearly flat region corresponding to a
diradical “intermediate.” In the 1O2 ene mechanism, the separate
“stages” are not the result of a flat potential energy surface, but
rather, the stages are intrinsically related to the topology of a
surface having two adjacent saddle points. For both types of
reactions, product distributions will be decided by dynamics
effects54-56 rather than separate barriers for alternative reactions.
We would argue that a distinction should be made in the
description of reaction mechanisms when the bonding changes
become so disconnected that experimental probes may affect
(45) Singleton, D. A.; Hang, C.; Szymanski, M. J.; Greenwald, E. E. J. Am.
Chem. Soc. 2003, 125, 1176.
(46) Rousseau, G.; LePerche, G.; Conia, P. M. Tetrahedron Lett. 1977, 2517.
(47) Orfanopoulos, M.; Grdina, M. J.; Stephenson, L. M. J. Am. Chem. Soc.
1979, 101, 275.
(48) Schulte-Elte, K. H.; Rautenstrauch, V. J. Am. Chem. Soc. 1980, 102, 1738.
(49) Prein, M.; Adam, W. Angew. Chem., Int. Ed. Engl. 1996, 35, 477-94.
(50) Dewar, M. J. S. J. Am. Chem. Soc. 1984, 106, 209.
(51) Nendel, M.; Sperling, D.; Wiest, O.; Houk, K. N. J. Org. Chem. 2000, 65,
3259.
(52) Beno, B. R.; Wilsey, S.; Houk, K. N. J. Am. Chem. Soc. 1999, 121, 4816.
Wilsey, S.; Houk, K. N.; Zewail, A. H. J. Am. Chem. Soc. 1999, 121, 5772.
(53) Leber, P.; Baldwin, J. E. Acc. Chem. Res. 2002, 35, 279.
(54) Carpenter, B. K. Angew. Chem., Int. Ed. 1998, 37, 3340.
To investigate the cis effect, a grid search was conducted for
the two possible rate-limiting transition structures for attack of
1O2 on 2-methyl-2-butene. This search employed, as earlier,
CCSD(T)/6-31G* single-point energies on B3LYP geometries.
(44) For a similar example, see: Bach, R. D.; Glukhovtsev, M. N.; Gonzalez,
C. J. Am. Chem. Soc. 1998, 120, 9902.
9
1326 J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003

Mechanism of Ene Reactions of Singlet Oxygen
A R T I C L E S
one of the bonding changes without the other. Despite the lack
of an intermediate in the 1O2 ene reaction, the mechanism
involves two kinetically distinguishable steps. The steps are
distinguishable in that the overall rate of the reaction and the
product distribution may be influenced separately, that is,
through isotopic substitution. This is very unlike most “two-
stage” mechanisms, such as an asynchronous Diels-Alder
reaction. Kinetically, the ene reaction of 1O2 with simple alkenes
behaves in every way like a regular two-step mechanism. We
would therefore describe the reaction as involving a “two-step
no-intermediate mechanism”.
The only directly observable difference between a two-step
no-intermediate mechanism and a regular two-step mechanism
involving a discrete intermediate occurs if the intermediate is
trappable. Early reports of trapping an intermediate in the ene
reaction of 1O2 with tetramethylethylene were later discounted.57
Extensive efforts by one of our groups to trap an intermediate
perepoxide have failed with any simple alkene capable of
undergoing the ene reaction. Of course, these observations only
show that an intermediate, if present, has a very short lifetime.
In two-step mechanisms, there is a continuum of possible
intermediate lifetimes, down to a few femtoseconds. From this
viewpoint, a two-step no-intermediate mechanism is just the
extreme of the continuum.
The composite use of CCSD(T) single-point energies on a
grid of B3LYP geometries lead to an excellent prediction of
the experimental intermolecular KIEs and intramolecular KIE
with 3. This strongly supports the accuracy of the composite
surface in the area around the rate-limiting transition state and,
by extension, supports the accuracy of the rest of the predicted
reaction surface. The unique feature of this surface is that it
involves two adjacent saddle points with no intervening
intermediate. Accompanying dynamics calculations support the
consistency of the intramolecular KIEs in 1 and 2 with
selectivity around the valley-ridge inflection of this surface.45
The paradox where experiment supports an intermediate but
theory does not is resolved; there is no intermediate, but a two-
step no-intermediate mechanism can exhibit product selectivity
not related to the rate-limiting transition state.
The results here provide the first experimental support for a
reaction surface involving two adjacent transition states. How-
ever, from the growing number of theoretically predicted cases,
there is no reason to think that surfaces involving adjacent saddle
points should be rare. When experimental observations are
interpreted, the inadequacy of the two-dimensional picture of
Figure 1a should be remembered.
Experimental Section
This continuum is illustrated further by the consideration of
unsymmetrical alkenes. The shape of the potential surfaces
shown in Figures 1 and 5 can occur for the singlet oxygen
reaction only when the transition state leads to a valley-ridge
inflection that can form two products, such as with tetrameth-
ylethylene or cis-2-butene. With trans-2-butene, isobutylene,
propene, or related unsymmetrical alkenes, the valley on one
side of the ridge necessarily disappears, and the potential surface
becomes closely related to other two-stage mechanisms involv-
ing one transition state and a plateau that leads to product
without a second barrier. The reaction coordinates for the sym-
metrical and unsymmetrical alkenes may be very similar, but
for the unsymmetrical alkenes the second stage would not have
experimental consequences.
Ene Reaction of 6. Example Procedure. A stream of O2 was slowly
bubbled through a mixture of 19.71 g (141 mmol) of 2,4-dimethyl-3-
isopropyl-2-pentene (6),59 14.66 g of 1,2-dichloroethane (as GC
standard), 2.0 g (2.0 mmol) of Rose Bengal, 210 mL of methanol, and
210 mL of 2-propanol at ∼10 °C while the mixture was being irradiated
with a 300-W sunlamp. After 26 h, the reaction was found to be 76 (
2% complete by GC analysis. The irradiation was stopped, and 23 mL
(24.2 g, 195 mmol) of trimethyl phosphite was added dropwise at 5-10
°C. The reaction mixture was stirred overnight at 25 °C, at which time
no residual hydroperoxide could be detected by GC. The volatiles were
then collected by a crude distillation (6 appears to form an azeotrope
with the alcohols) and partitioned between 200 mL of pentane and 400
mL of water. The aqueous layer was extracted with a second 200-mL
portion of pentane. The combined pentane layers were rinsed with four
250-mL portions of water, dried (MgSO4), and passed through a 6-in.
pad of silica, rinsing with pentane. The resulting solution was subjected
to two successive fractional distillations using 30-cm and 10-cm
Vigreaux columns to afford 2.35 g of the unreacted 6 (12%, 50%
recovery, bp 148-150 °C).
Conclusion
Qualitatively, the intermolecular and intramolecular 13C
isotope effects for reactions of 6 and tetramethylethylene,
respectively, support the general picture of the mechanism that
An analogous reaction using 19.0 g of 2,4-dimethyl-3-isopropyl-2-
pentene was taken to 77 ( 2% conversion, and 3.4 g of unreacted
alkene was recovered.
2
had been previously concluded from intramolecular H KIEs,
1
and they provide additional detail. The initial attack of O2 on
Ene Reaction of Tetramethylethylene. Example Procedure. A
stream of O2 was slowly bubbled through a mixture of 20.64 mL (14.61
g, 173.6 mmol) of tetramethylethylene and 112 mg (0.11 mmol) of
Rose Bengal in 100 mL of dry methanol at 0 °C while being irradiated
with a 300-W sunlamp. After 4 d, the irradiation was discontinued,
7.18 g (190 mmol) of NaBH4 was added, and the mixture was stirred
at 25 °C for 2 h. After the addition of 100 mL of water, the mix-
ture was extracted with three 50-mL portions of pentane. The com-
6 occurs at the center of the π-bond with an early transition
state. After the rate-limiting transition state, the mechanism
involves a species with the effective symmetry of a perepoxide.
Quantitatively, the isotope effects have allowed us to triage
the differing mechanisms predicted by various theoretical
methods. All of the geometry-optimizable methods employed
lead to poor predictions of the experimental KIEs. It is
particularly noteworthy that our best efforts at CASSCF
calculations did not afford accurate results. The limitations of
these calculations have been previously documented, but the
results here suggest caution in the interpretation of recent
(56) Doubleday, C., Jr.; Bolton, K.; Hase, W. L. J. Am. Chem. Soc. 1997, 119,
5251. Doubleday, C.; Nendel, M.; Houk, K. N.; Thweatt, D.; Page, M. J.
Am. Chem. Soc. 1999, 121, 4720. Doubleday, C. J. Phys. Chem. 2001,
105, 6333. Doubleday, C., Jr.; Bolton, K.; Hase, W. L. J. Phys. Chem. A
1998, 102, 3648. Doubleday, C.; Li, G.; Hase, W. L. Phys. Chem. Chem.
Phys. 2002, 4, 304.
(57) Gollnick, K.; Haisch, D.; Schade, G. J. Am. Chem. Soc. 1972, 94, 1747.
(58) (a) Bobrowski, M.; Liwo, A.; Oldziej, S.; Jeziorek, D.; Ossowski, T. J.
Am. Chem. Soc. 2000, 122, 8112. (b) Maranzana, A.; Ghigo, G.; Tonachini,
G. J. Am. Chem. Soc. 2000, 122, 1414.
(59) Lion, C.; Dubois, J.-E. Bull. Soc. Chim. Fr. 1976, 1875. Newman, M. S.;
Arkell, A.; Fukunaga, T. J. Am. Chem. Soc. 1960, 82, 2498.
1
CASSCF calculations looking at other O2 reactions.58
(55) Carpenter, B. K. J. Am. Chem. Soc. 1995, 117, 6336. Reyes, M. B.;
Carpenter, B. K. J. Am. Chem. Soc. 2000, 122, 10163. Reyes, M. B.;
Lobkovsky, E. B.; Carpenter, B. K. J. Am. Chem. Soc. 2002, 124, 641.
Carpenter, B. K. J. Am. Chem. Soc. 1996, 118, 10329.
9
J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003 1327

A R T I C L E S
Singleton et al.
bined pentane layers were dried (MgSO4) and fractionally distilled
through a 10-cm Vigreaux column to afford 3.49 g (20%) of 9, bp
112-115 °C, contaminated with ∼2% of 2,3-dimethyl-2-butanol and
0.5% of pentane.
150-s delays between pulses, collecting 480 000 points and zero-filling
to 512 K. The longest T1 value in 9 was 18.8 s. For comparisons of
each relevant pair of peaks, a set of spectra was taken with the
transmitter centered between the two peaks. The average results from
six measurements on each of two independent samples are shown in
Figure 3, with standard deviations shown in parentheses.
NMR Measurements. All samples were prepared using a constant
amount of analyte (1.70 g for samples of 6, 1.5 g for samples of 9) in
10-mm NMR tubes filled with CDCl3 to a constant height of 5.0 cm.
The 13C spectra were recorded at 100.58 MHz with inverse gated
decoupling. A T1 measurement was performed on each sample to ensure
that the relaxation rates did not change from sample to sample.
Integrations were determined numerically using a constant integration
region for each peak of 10 times the width at half-height. A zeroth
order baseline correction was generally applied, but in no case was a
first order (tilt) correction applied.
The 13C spectra of 6 were taken at a controlled temperature of 50
°C, using 244-s delays between calibrated 45° pulses, a 11.899-s
acquisition time, and collecting 486 656 points. The results from six
measurements taken on each of the two samples of recovered 6 and
samples of 6 that were not subjected to the reaction conditions are
summarized in the Supporting Information.
Acknowledgment. D.A.S., C.H., M.J.S., and M.P.M. thank
NIH Grant No. GM-45617 and The Robert A. Welch Founda-
tion for support and NSF Grant No. CHE-9528196 and the
Texas A&M University Supercomputing Facility for computa-
tional resources. The NSF (CHE-0077917) provided funds for
the purchase of NMR instrumentation. A.L., K.T.K., J.S.C.,
A.G., and K.N.H. thank NSF Grant CHE-9986344. K.T.K.
acknowledges computational support from the NCSA facility
at the University of Kentucky.
Supporting Information Available: Theoretical structures and
energies. This material is available free of charge via the Internet
at http://pubs.acs.org.
For the intramolecular isotope effects with 9, some more special
acquisition parameters were used. This included calibrated 90° pulses,
a sweep width of 40 000 Hz, a filter bandwidth of 28 000 Hz, and
JA027225P
9
1328 J. AM. CHEM. SOC. VOL. 125, NO. 5, 2003