Selectivity in the electron transfer catalyzed Diels-Alder reaction of (R)-α-phellandrene and 4-methoxystyrene

Article abstract of DOI:10.1021/jo8002562

(Chemical Equation Presented) Electron transfer catalysis is an effective method for the acceleration of Diels-Alder reactions between two substrates of similar electron density. The dependence of the selectivity of the Diels-Alder reaction between (R)-α-phellandrene and 4-methoxystyrene catalyzed by photoinduced electron transfer with tris(4-methoxyphenyl) pyrylium tetrafluoroborate is studied. Despite the fact that the radical ions involved are highly reactive species, complete regioselectivity favoring attack on the more highly substituted double bond is observed. The endo/exo selectivity and the periselectivity between [4 + 2] and [2 + 2] cycloaddition is found to be solvent-dependent. Stereochemical analysis showed that the periselectivity is correlated with the facial selectivity, with attack trans to the isopropyl group leading to the [4 + 2] product and cis attack leading to the formation of the [2 + 2] product. A good correlation between the dielectric constant of the solvent and the endolexo ratio is found, but more polar solvents lead to lower periselectivity. The effect of reactant and catalyst concentrations is found to be smaller. These results are rationalized in the context of the relative stability of the ion-molecule complexes and the singly linked intermediate of the reaction.

Full text of DOI:10.1021/jo8002562

Selectivity in the Electron Transfer Catalyzed Diels-Alder Reaction
of (R)-r-Phellandrene and 4-Methoxystyrene
Christo S. Sevov and Olaf Wiest*
Department of Chemistry and Biochemistry, UniVersity of Notre Dame, Notre Dame, Indiana 46556
owiest@nd.edu
ReceiVed January 31, 2008
Electron transfer catalysis is an effective method for the acceleration of Diels-Alder reactions between
two substrates of similar electron density. The dependence of the selectivity of the Diels-Alder
reaction between (R)-R-phellandrene and 4-methoxystyrene catalyzed by photoinduced electron
transfer with tris(4-methoxyphenyl) pyrylium tetrafluoroborate is studied. Despite the fact that the
radical ions involved are highly reactive species, complete regioselectivity favoring attack on the
more highly substituted double bond is observed. The endo/exo selectivity and the periselectivity
between [4 + 2] and [2 + 2] cycloaddition is found to be solvent-dependent. Stereochemical analysis
showed that the periselectivity is correlated with the facial selectivity, with attack trans to the isopropyl
group leading to the [4 + 2] product and cis attack leading to the formation of the [2 + 2] product.
A good correlation between the dielectric constant of the solvent and the endo/exo ratio is found,
but more polar solvents lead to lower periselectivity. The effect of reactant and catalyst concentrations
is found to be smaller. These results are rationalized in the context of the relative stability of the
ion-molecule complexes and the singly linked intermediate of the reaction.
Introduction
substrates of similar electron density, and the fact that, unlike
most other pericyclic reactions of radical cations, it could be
extended to a variety of heterosubstituted compounds.6
The acceleration of pericyclic reactions through electron
transfer catalysis (ETC) has been one of the most fruitful areas
of radical ion chemistry.1 Using a sequence of one-electron
oxidation, pericyclic reaction, and back electron transfer, a wide
range of electrocyclic reactions,2 cycloadditions,3 cyclorever-
sions,4 and sigmatropic shifts5 can be accelerated by up to 13
orders of magnitude. As is the case in the thermal pericyclic
reactions, the radical cation Diels-Alder reaction has been
particularly useful because of the rapid construction of an
important structural element, the possibility to react two
Starting from the seminal work of Bauld and co-workers,7
the mechanism of the radical cation Diels-Alder reaction has
been studied extensively using a variety of experimental8 and
computational methods.9 Stereochemical studies showed that
for different substituted systems, the reaction proceeds through
a stepwise process that allows the scrambling of stereochem-
istry.10 This is confirmed by the calculation of the reaction
pathways at different levels of theory by our group11 and others12
10.1021/jo8002562 CCC: $40.75
Published on Web 09/12/2008
2008 American Chemical Society
J. Org. Chem. 2008, 73, 7909–7915 7909

Sevov and Wiest
that indicate the presence of acyclic intermediates in different
direct and indirect radical cation Diels-Alder pathways. The
relative stabilities of these intermediates can be used to
rationalize the experimentally observed regioselectivity of the
reaction.13 The structures of stationary points and the overall
shape of the potential energy surface strongly resembles the
ones calculated for the stepwise, biradical pathways of the
thermal Diels-Alder reaction. In contrast, the concerted,
symmetric transition structures analogous to the one calculated
for the thermal reaction are distorted by vibronic coupling to
low-lying excited states in a pseudo-second-order Jahn-Teller
distortions and are therefore second-order saddle points much
higher in energy.1d,e,11
stereoselective. Even for combinations of nonpolar diene-
dienophile combinations, good to excellent endo/exo ratios were
obtained.8 It is this dichotomy between the widely agreed upon
stepwise pathway of electron transfer reactions, which can
provide an assortment of products, versus the empirically
observed selectivity that makes a variation of reaction conditions
a useful study. When a chiral diene was used, complete facial
diastereoselectivity of the reaction was observed.14 In a rare
application of a radical cation Diels-Alder reaction to a natural
product synthesis, the ETC reaction of phenyl vinyl sulfide with
a chiral diene only gave a 1.9:1.3:1 diastereoselectivity.
Nevertheless, this reaction is still much more selective than the
corresponding thermal reaction, which gives 11 isomers in
approximately equal amounts.15 These findings are surprising
considering that the radical cations involved are highly reactive
species and the calculated energy differences between the
different stationary points on the potential energy surfaces of
these reactions are typically very small.11-13 It is therefore not
obvious what the origin of the observed selectivity is and how
it depends on the reaction conditions.
So far, there have been very few systematic investigations
of the selectivity of radical cation Diels-Alder reactions.
Martiny et al. studied the endo/exo selectivity in the radical
cation Diels-Alder reaction of simple model systems as a
function of the reaction conditions using the sensitizer 2,4,6-
tris(4-methoxy phenyl) pyrylium tetrafluoroborate, which has
been shown to successfully catalyze [4 + 2] and [2 + 2]
cycloaddition reactions in the absence of oxygen.16 This electron
transfer sensitizer was deemed particularly useful in the study
of the reaction conditions because the neutral pyrylium radical
formed by the electron transfer does not strongly complex to
the substrate radical cation, thus yielding free radical cations
and less complex reaction mechanisms. Variable concentrations,
stoichiometric ratios, solvents, and sensitizers were used in
reactions to expose possible trends and mechanistic pathways.
Depending on these variables, endo/exo ratios of 2.4:1 up to
24.3:1 were found for the electron transfer Diels-Alder reaction
of 1,3-cyclohexadiene and styrenes.16a
Another useful feature of the radical cation Diels-Alder
reaction is the empirical finding that the reaction is highly
(1) For overviews, compare: (a) Bauld, N. L. Tetrahedron 1989, 45, 5307–
5363. (b) Schmittel, M.; Burghart, A. Angew. Chem., Int. Ed. Engl. 1997, 36,
2550–2589. (c) Fagnoni, M. Heterocycles 2003, 60, 1921–1958. (d) Saettel, N. J.;
Oxgaard, J.; Wiest, O. Eur. J. Org. Chem. 2003, 1429–1439. (e) Donoghue,
P. J.; Wiest, O. Chem.-Eur. J. 2006, 12, 7018–7026. (f) Fagnoni, M.; Dondi,
D.; Ravelli, D.; Albini, A. Chem. ReV. 2007, 107, 2725–2756. (g) Mu¨ller, F.;
Mattay, J. Chem. ReV. 1993, 93, 99–117. (h) Fukuzumi, S. Org. Biomol. Chem
2003, 1, 609–620. (i) Linker, T.; Schmittel, M. Radikale und Radikalionen in
der Organischen Synthese; Wiley-VCH: Weinheim, 1998. (j) Bauld, N. L.
Radicals, Ion Radicals, and Triplets; Wiley & Sons: New York, 1997.
(2) For example: (a) Wiest, O. J. Am. Chem. Soc. 1997, 119, 7513–7519.
(b) Sastry, G. N.; Bally, T.; Hrouda, V.; Carsky, P. J. Am. Chem. Soc. 1998,
120, 9323–9334. (c) Barone, V.; Rega, N.; Bally, T.; Sastry, G. N. J. Phys.
Chem. A 1999, 103, 217–219. (d) Kawamura, Y.; Thurnauer, M.; Schuster, G. B.
Tetrahedron 1986, 42, 6195. (e) Takahashi, Y.; Miyamoto, K.; Sakai, K.; Ikeda,
H.; Miyashi, T.; Ito, Y.; Tabohashi, K. Tetrahedron Lett. 1996, 37, 5547–5550.
(f) Pasto, D. J.; Yang, S.-H. J. Org. Chem. 1989, 54, 3544–3549. (g) Aebischer,
J. N.; Bally, T.; Roth, K.; Haselbach, E.; Gerson, F.; Qin, X.-Z. J. Am. Chem.
Soc. 1989, 111, 7909–7914.
(3) For example: (a) Mlcoch, J.; Steckhan, E. Angew. Chem., Int. Ed. Engl.
1985, 5, 412–414. (b) Jia, X. D.; Han, B.; Zhang, W.; Jin, X.; Yang, L.; Liu,
Z.-L. Synthesis 2006, 2831–2836. (c) Gonzalez-Bejar, M.; Stiriba, S.-E.;
Domingo, L. R.; Perez-Prieto, J.; Miranda, M. A. J. Org. Chem. 2006, 71, 6932–
6941.
(4) (a) Wiest, O. J. Phys. Chem. A 1999, 103, 7907–7911. (b) Schroeter, K.;
Schro¨der, D.; Schwarz, H.; Reddy, G. D.; Wiest, O.; Carra, C.; Bally, T. Chem.-
Eur. J. 2000, 6, 4422–4430.
(5) (a) Oxgaard, J.; Wiest, O. J. Am. Chem. Soc. 1999, 121, 11531–11537.
(b) Dinnocenzo, J. P.; Conlon, D. A. J. Am. Chem. Soc. 1988, 110, 2324–2325.
(c) Dinnocenzo, J. P.; Conlon, D. A. Tetrahedron Lett. 1995, 36, 7415–7418.
(6) (a) Zhou, L.; Jia, X. D.; Li, R.; Liu, Z. G.; Liu, Z. L.; Wu, L. M.
Tetrahedron Lett. 2005, 46, 8937–8939. (b) Haberl, U.; Steckhan, E.; Blechert,
S.; Wiest, O. Chem.-Eur. J. 1999, 5, 2859–2865. (c) Wiest, O.; Steckhan, E.
Angew. Chem., Int. Ed. Engl. 1993, 32, 901–903.
(7) (a) Bellville, D. J.; Bauld, N. L. J. Am. Chem. Soc. 1982, 104, 2665–
2667. (b) Pabon, R.; Bauld, N. L. J. Am. Chem. Soc. 1983, 105, 633–634. (c)
Bauld, N. L.; Bellville, D. J.; Harirchian, B.; Lorenz, K. T.; Pabon, R. A.;
Reynolds, W.; Wirth, D. D.; Chiou, H. S.; Marsh, B. K. Acc. Chem. Res. 1987,
20, 371–378. (d) Bauld, N. L.; Mirafzal, G. A. J. Am. Chem. Soc. 1991, 113,
3613–3614. (e) Mirafzal, G. A.; Liu, J. P.; Bauld, N. L. J. Am. Chem. Soc. 1993,
115, 6072–6077. (h) Yueh, W.; Bauld, N. L. J. Am. Chem. Soc. 1995, 117, 5671–
5676.
(8) (a) Schmittel, M.; Wohrle, C.; Bohn, I. Chem.-Eur. J. 1996, 2, 1031–
1040. (b) Schmittel, M.; Wohrle, C. J. Org. Chem. 1995, 60, 8223–8230. (c)
Mattay, J.; Gersdorf, J.; Mertes, J. Chem. Commun. 1985, 1088–1090. (d) Hu,
H.; Wenthold, P. G. J. Phys. Chem. A 2002, 106, 10550–10553. (e) Goebbert,
D. J.; Liu, X.; Wenthold, P. G. J. Am. Soc. Mass. Spectrom. 2004, 15, 114–120.
(f) van der Hart, W. J. Int. J. Mass Spectrom. 2001, 208, 119–125. (g) Gassmann,
P. G.; Singleton, D. A. J. Am. Chem. Soc. 1984, 106, 7993. (h) Saettel, N. J.;
Wiest, O.; Singleton, D. A.; Meyer, M. P. J. Am. Chem. Soc. 2002, 124, 11552.
(9) For example: (a) Lorentz, K. T.; Bauld, N. L. J. Am. Chem. Soc. 1987,
109, 1157–1160. (b) Bauld, N. L.; Bellville, D. J.; Pabon, R. A.; Chelsky, R.;
Green, G. J. J. Am. Chem. Soc. 1983, 105, 2378–2382.
Here, we present the first systematic study of chemo-, peri-,
and stereoselectivity of electron transfer catalyzed Diels-Alder
reactions of a chiral diene as a function of the reaction
conditions. After identifying the main products, the regio- and
diastereoselectivity of reacting (R)-R-phellandrene (1) and
electron-rich 4-methoxystyrene (2) while varying solvent polar-
ity, concentration, electron transfer catalysts, and catalyst
concentration will be discussed. Although 1 has been used
numerous times as a diene in radical cation Diels-Alder
reactions,6c,10b,14,17 the ETC reaction with the prototypical 2 as
well as the factors controlling the chemo-, regio-, and diaste-
reoselectivity of the reaction have to the best of our knowledge
not been studied.
(14) (a) Gieseler, A.; Steckhan, E.; Wiest, O.; Knoch, F. J. Org. Chem. 1991,
56, 1405–1411. (b) Gieseler, A.; Steckhan, E.; Wiest, O. Synlett 1990, 275–
277.
(10) (a) Bauld, N. L.; Yang, J. K. Tetrahedron Lett. 1999, 40, 8519–8522.
(b) Wiest, O.; Steckhan, E. Tetrahedron Lett. 1993, 34, 6391–6394. (c) Schmittel,
M.; Wohrle, C.; Bohn, I. Acta Chim. Scand. 1997, 51, 151–157.
(11) Haberl, U.; Steckhan, E.; Wiest, O. J. Am. Chem. Soc. 1999, 121, 6730–
6736.
(12) (a) Hofmann, M.; Schaefer, H. F. J. Am. Chem. Soc. 1999, 121, 6719–
6729. (b) Hofmann, M.; Schaefer, H. F. J. Phys. Chem. A 1999, 103, 8895–
8905. (c) Bouchoux, G.; Nguyen, M. T.; Salpin, J. Y. J. Phys. Chem. A 2000,
104, 5778–5786.
(15) Harirchian, B.; Bauld, N. L. J. Am. Chem. Soc. 1989, 111, 1826–1828.
(16) (a) Martiny, M.; Steckhan, E.; Esch, T. Chem. Ber. Receuil. 1993, 126,
1671–1682. Compare also: (b) Mlcoch, J.; Steckhan, E. Tetrahedron Lett. 1987,
28, 1081–1084.
(17) (a) Gonza´lez-Be´jar, M.; Stiriba, S.-E.; Miranda, M. A.; Perez-Prieto, J.
ARKIVOC 2007, 4, 344–355. (b) Gonza´lez-Be´jar, M.; Stiriba, S.-E.; Miranda,
M. A.; Perez-Prieto, J. J. Org. Chem. 2006, 71, 6932–6941. (c) Davies, A. G.;
Hay-Motherwell, R. J. Chem. Soc. 1988, 2099–2013. For an example of the
thermal Diels-Alder reaction of 1, see for example: (d) Pickering, M. J. Chem.
Educ. 1990, 67, 524–525. (e) Littmann, E. R. J. Am. Chem. Soc. 1936, 58, 1316–
1317.
(13) (a) Valley, N. A.; Wiest, O. J. Org. Chem. 2007, 72, 559–566. (b) Wiest,
O.; Steckhan, E.; Grein, F. J. Org. Chem. 1992, 57, 4034–4037.
7910 J. Org. Chem. Vol. 73, No. 20, 2008

Electron Transfer Catalyzed Diels-Alder Reaction
FIGURE 1. Main products of the ETC reaction of 1 and 2 and key NOE contacts.
TABLE 1. Effect of Absolute Concentration in CH2Cl2
and 5, respectively, in Supporting Information), and the relative
configuration was assigned using the ROESY cross peaks of
the benzylic proton with olefinic and exo protons in 4 and 5,
respectively, as indicated in Figure 1. Compounds 4 and 5 are
therefore assigned to be the exo and endo products of the
cycloadditions, respectively. It is important to note that no
significant amounts of the product resulting from the attack on
the other side of the diene are observed under these reaction
conditions. This very high facial selectivity is not only syntheti-
cally useful but also mechanistically interesting and can be
understood on the basis of the calculations of simple model
systems. The reaction is generally accepted to be stepwise, and
the relative configuration is set in the formation of the singly
linked intermediate. Earlier theoretical work13 has shown that
the reaction of the methyl-substituted 1,3-diene radical cation
is ∼7 kcal/mol exothermic. Consequently, an early transition
state would be expected, but the bond lengths for the formation
of the first bond in the transition structure is with ∼2.0-2.1 Å
shorter than expected. This is due to the initial formation of an
ion-molecule complex that is approximately isoenergetic with
the singly linked intermediate. These calculations also predicted
very similar energies for the endo and exo pathways, in
agreement with experiment.13
After elucidating the structure of the two major isomers, we
then turned our attention to the effect of the reaction conditions
on the diastereoselectivity of the reactions. First, we investigated
the effect of the absolute concentration of diene and dienophile
on the reaction. For this purpose, GC was used to quantify the
relative amounts of 4 and 5, as well as the minor isomers 6-10.
Coinjection of isolated and structurally characterized samples
was used to distinguish the respective retention times of the
hetero- and homodimers and assign the isomers. Although the
minor isomers were not formed in sufficient amounts to allow
for their detailed structural identification, analysis of the GC/
MS data (see Supporting Information) showed that 6-10 are
isomers of 4 and 5.
concentration [M]
1.51
0.50
0.15
0.05
dimers of 1
mixed adducts
dimers of 2
1.00
0.33
∼0
1.00
0.43
0.04
1.00
2.67
0.38
1.00
8.12
1.32
4
5
6
29%
55%
2%
28%
53%
2%
37%
60%
2%
35%
63%
2%
0%
0%
7
8
9
10
8%
2%
2%
3%
9%
3%
2%
2%
1%
trace
trace
trace
0%
trace
Results and Discussion
As a model diene for the study of the diastereoselectivity of
the radical cation Diels-Alder reaction, the commercially
available (R)-R-phellandrene (1) was chosen because it can
reasonably be expected that only steric interactions will be
relevant for the reaction outcome, simplifying the analysis of
the results. GC/MS analysis of the reaction of 1 with 4-meth-
oxystyrene (2) in dichloromethane under electron transfer
catalysis induced by irradiation at 420 nm in the presence of
2% 2,4,6-tris(4-methoxyphenyl) pyrylium tetrafluorborate (3)
leads to the formation of seven different isomers of the mixed
adduct, 4-10, in a combined isolated yield of 29% of which
only two are present in sufficient amounts to be isolated and
structurally characterized. Control experiments show that, in
agreement with previous studies on related systems,16 the
presence of both 3 and irradiation is necessary for product
formation, thus confirming the electron transfer catalysis of the
reaction. Several dimers of 1 were formed in appreciable
amounts as well (for a typical GC trace, see Supporting
Information). Besides 4 and 5, the other isomers were formed
in small amounts. This and the very similar polarity of the
isomers of essentially pure hydrocarbons made the isolation
challenging. HPLC and normal phase silica column chroma-
tography showed no useful separation of the mixture because
of their similar polarity. However, effective separation was
achieved using argentated silica. This technique, where silica
gel is pretreated with silver by forming strong Si-O-Ag+
interactions, has been used most commonly in separations of
molecules with steroidal backbones.18 The well-known ability
of silver to complex with a double bond leads to a resolution
of the diastereomers based on their steric accessibility. A
combination of standard chromatography and this method can
be used on a preparative scale and yielded 96% and 97% pure
(by NMR) samples of 4 and 5 in 2% and 4% yield, respectively.
The structures of 4 and 5 were determined on the basis of
the assignments of the protons by 2D NMR (see (5.91 ppm,
2.75 ppm) and (1.39 ppm, 2.83 ppm) ROESY crosspeaks for 4
The results for the variation of the absolute concentration of
a 1:1 mixture of diene and dienophile in the reaction in
dichloromethane are summarized in Table 1. The reaction is
less selective with respect to heterodimer versus homodimer
formation at high substrate concentrations, with the selectivity
reversing over the concentration range studied here. The fact
that both homodimers are found is consistent with the fact that
although the exact oxidation potentials of 1 and 2 are not known
due to the irreversibility of the redox reaction under typical
cyclic voltametry conditions, the peak potentials are sufficiently
close to indicate that both radical cations 1•+ and 2•+ are formed.
In contrast, the concentration dependence of the relative yields
for the homodimers as well as the mixed adducts is more
difficult to understand. The finding that the relative amount of
(19) (a) Johnston, L. J.; Schepp, N. P. In AdVances in Electron Transfer
Chemistry; Mariano, P. S., Ed.; JAI Press: New York, 1996; Vol. 6, pp 41-102.
(b) Mattes, S. L.; Farid, S. Org. Photochem. 1983, 6, 233–326. (c) Lewis, F. D.;
Kojima, M. J. Am. Chem. Soc. 1988, 110, 8664–8670. (d) O’Neil, L. L.; Wiest,
O. J. Org. Chem. 2006, 71, 8926–8933.
(18) (a) Li, T. S.; Li, J. T.; Li, H. Z. J. Chromatogr. A 1993, 715, 372–375.
(b) Cert, A.; Moreda, W. J. Chromatogr. A 1998, 823, 291–297.
J. Org. Chem. Vol. 73, No. 20, 2008 7911

Sevov and Wiest
TABLE 2. Solvent Effect on ETC Reaction of 1 and 2
conditions
CH2Cl2
acetone
acetonitrile
DMSO
dimers of 1
mixed adducts
dimers of 2
1.00
2.67
0.38
1.00
2.19
0.21
1.00
0.99
0.04
1.00
0.67
0.02
4
5
6
7
8
9
10
37.4%
59.5%
1.8%
0.7%
0.1%
0.2%
0.3%
34.9%
57.9%
1.4%
3.3%
1.2%
1.0%
0.3%
21.5%
45.3%
0.7%
12.1%
4.8%
9.5%
29.8%
5.4%
26.4%
9.9%
3.4%
12.3%
7.2%
11.7%
FIGURE 2. Endo/exo ratio of 4 to 5 as a function of dielectric constant.
homodimers of 1 is lowest at low concentrations could be
rationalized by the hypothesis that formation of these most
sterically hindered homodimers relies on reduction of the singly
linked intermediate to the biradical, which would then rapidly
close. This hypothesis would be in analogy to the concentration
dependence of the formation of the strained cyclobutane through
ETC dimerization of 2.19 It is, however, noteworthy that from
the practical point of view, the reaction should be run at lower
concentrations of diene and dienophile to ensure the predominant
formation of 4 and 5 and that this stepwise reaction of a highly
reactive intermediate, which could be expected to have poor
selectivity, shows in fact a ∼98% facial selectivity in dilute
solutions to give 4 and 5.
Following the study of the concentration effects, various
solvents were tested at a concentration of 0.15 M to observe
the effect of solvent polarity on the reaction. It can be
hypothesized that the solvent could have an effect at two points
of the reaction pathway: first, a more polar solvent would solvate
the substrate radical cation better and weaken the interactions
in the initial ion-molecule complex, which are thought to con-
tribute to the stereoselection; second, the solvent could stabilize
the localized charges in the acyclic intermediate, which could
lead to stereochemical scrambling. Due to the cyclic nature of
the diene, no trans addition would be possible in the reaction
studied here, but a rotation around the C1-C2 bond could
interconvert the endo and exo products independently of the
orientation of the initial attack. The results of these studies are
summarized in Table 2.
Increasing the polarity of the solvent favors the dimerization
of the diene 1 over the formation of mixed adducts. The free
radical cation formed in a more polar solvent is more likely to
react to form either a mixed or a homoadduct, giving the poorer
selectivity that is observed. This conversely suggests that the
ion-molecule complexes present in less polar solvents have a
higher selectivity. The other chemoselectivity trend observed
is that the amount of the homodimer of the dienophile 2
increases relative to the dimerization of the diene 1 in less polar
solvents, even though the homodimer of 1 is still favored. This
selectivity trend indicates that both substrates can be oxidized
by the excited state of 3 in a diffusion controlled fashion but
that in polar solvents 2•+ has a sufficiently long lifetime to
oxidize 1. The alternative explanation that the relative redox
potentials of 1 and 2 changes as a function of the solvent appears
to be less likely but cannot be excluded.
correspond to [2 + 2] cycloadditions of 1 and 2 and are most
likely formed through a back electron transfer of the singly
linked intermediate to form the biradical, which then closes the
cyclobutane ring.19 Second, the solvent polarity has a substantial
effect on the endo/exo ratio as shown in Figure 2. In contrast
to the decreasing selectivity for the formation of 4 and 5, an
increase in stereoselectivity is observed with increasing polarity
of the solvent. This and the data in Table 2 indicate that the
effect of the solvent polarity is different for the pathway leading
the exo isomer 4, where the yield decreases 4-fold, than for the
pathway leading the endo isomer 5, where the yield decreases
only 2-fold. This finding could be rationalized by the hypothesis
that the larger solvent stabilization of the exo form of the
ion-molecule complex or singly linked intermediate relative
to the corresponding endo diastereomer can lead to larger
amount of other isomers than the corresponding endo form. In
polar solvents, the exo isomers are therefore more likely to react
through other pathways than the endo isomers, which are more
likely to collapse to 5.
Finally, we investigated the effect of the sensitizer. Varying
the reaction time and catalyst concentration in solution was
tested and had no impact on mixed adduct selectivity (see
Supporting Information). It is generally accepted that the use
of a cationic sensitizer or cosensitizer in reactions catalyzed by
oxidative electron transfer is beneficial because it does not lead
to ion pairs of the substrate radical cation with the radical anion
of the sensitizer after ET.16 Thus, diffusion is not hindered by
Coulomb attraction and will rapidly lead to the formation of
free radical ions, which can then react with the respective neutral
molecules. We tested this hypothesis by comparing the chemo-
and stereoselectivity for the standard reaction conditions in
dichloromethane and acetonitrile catalyzed by 3+ and 1,10-
dicyanoanthracene 11. The effects of the solvent on ET reactions
catalyzed by 3+ have already been discussed earlier. In the
reactions catalyzed by 11, the selectivity for the formation of
mixed adducts is higher even though the total amount of all
products after the same reaction time is much smaller than in
the case of the reactions catalyzed by 3+ as shown by
comparison of the GC peak areas to the ones of unreacted
substrate. The selectivity is even higher in acetonitrile, but
significant amounts of dimers of 2 are also formed in this case.
Interestingly, the endo/exo ratio is at ∼2:1 nearly constant
in all cases. The main differences in stereoselectivity between
the two sensitizer systems is observed in the larger amounts of
the minor products 6-10 that are formed. The reaction in
dichloromethane sensitized by 3+ is found to be the most
selective, forming essentially only the two isomers of the mixed
The effect of solvent polarity on the product distribution of
the mixed adducts is two-fold. First, significantly larger amounts
other isomers are formed in more polar solvent and 7 and 10
become the second and third most abundant product in DMSO.
As will be discussed in more detail below, these two isomers
7912 J. Org. Chem. Vol. 73, No. 20, 2008

Electron Transfer Catalyzed Diels-Alder Reaction
TABLE 3. Variation of Sensitizer and Solvent
conditions
3+/CH2Cl2 3+/acetonitrile 11/CH2Cl2 11/acetonitrile
dimers of 1
mixed adducts
dimers of 2
1.00
2.67
0.38
1.00
0.99
0.04
1.00
4.27
0.85
1.00
8.44
3.18
FIGURE 3. Structure of isomer 7.
4
5
6
7
8
9
10
37.4%
59.5%
1.8%
0.7%
0.1%
0.2%
0.3%
21.5%
45.3%
0.7%
12.1%
4.8%
22.3%
50.8%
4.3%
5.8%
1.1%
24.6%
53.9%
0.8%
1.0%
0.4%
because they would force the formation of localized spin and/
or charge. Previous computational studies on 2-methyl-
substituted 1,3-butadiene radical cations13a indicate that attack
at C1 is favored. More importantly, the experimentally observed
regiochemistry of 4, 5, and 7 indicates that the major pathways
attack at C1 because the formation of the experimentally
observed major products via attack at C4 would require the
formation of a primary radical or cation.
This attack can proceed trans or cis to the isopropyl group,
initially leading to the two possible ion-molecule complexes
12•+ and 13•+, which are essentially equal in energy within the
accuracy that can be reasonably expected for the method used
here. On the basis of previous computational studies,11,12 the
potential energy surface around 12•+ and 13•+ is expected to
be quite flat, resulting in several, presumably rapidly intercon-
verting structures of the ion-molecule complexes. One such
structure is shown for illustrative purposes in Figure 5 on the
left. The first bond to yield the trans and cis singly linked
intermediates 14•+ and 15•+ is formed through the transition
structures TS12,14 and TS13,15 with activation energies of 2.4
and 3.8 kcal/mol, respectively. The lower activation energy for
TS12,14 as well as the increased energy difference between the
two singly linked intermediate is consistent with a stronger steric
repulsion between the styryl and the isopropyl group. Consistent
with earlier studies,13a the activation energy for the closure to
the [4 + 2] product is at 0.6 kcal/mol very low.
Interestingly, we were unable to locate a transition structure
for the ring closure to the [2 + 2] products 7 or a hypothetical
trans analog 7′. Attempts to calculate the structures of either
7•+ or 7′•+ led to a spontaneous opening of the cyclobutane
ring, forming a “long bond intermediate”19 that is stabilized by
both benzylic and allylic resonance. It is therefore clear that
the formation of the [2 + 2] products proceeds through the
previously proposed back electron transfer (BET)-ring closure
sequence.19 This is further supported by the fact that the
calculated activation energy for the closure of 15•+ to the [4 +
2] product, which is experimentally not observed, is at 6.5 kcal/
mol substantially higher than in the case of 14•+. This is due to
the larger steric repulsion in the case of 15•+, as can be seen in
Figure 5. The increased lifetime of 15•+ therefore allows BET,
which leads to larger amounts of 7. If back electron transfer is
not feasible or too slow, 14•+ and 15•+ could possibly
equilibrate, giving larger amounts of the thermodynamically
more stable 4 and 5. This could also explain the larger amounts
of 7 formed in polar solvents and, qualitatively, in the reactions
using 11 as a sensitizer, even though it is not clear if the radical
ions will occur in the form of free ion pairs, as discussed earlier.
An alternative and parallel explanation for the increased
formation of 7 as a function of solvent polarity is the hypothesis
that in polar solvents 1•+ is present as a free radical cation that
has a lower facial selectivity, whereas in less polar solvents, it
reversibly forms an ion-molecule complex that has a higher
facial selectivity for attack. The high selectivity observed in
dichloromethane would thus be a result of a double stereose-
lection in the two independent steps of the formation of the
ion-molecule complex and the formation of the singly linked
3.4%
12.3%
0.4%
15.2%
0.2%
19.2%
DielssAlder product 4 and 5. For the reactions sensitized by
11, substantial amounts of 10 are formed even in dichlo-
romethane. These conditions, in which reaction is most likely
to proceed via an ion pair 1•+/11•-, also give a relatively large
amount of 6. Interestingly, the only other set of reaction
conditions that lead to the formation of substantial amounts of
6 is the reaction in the relatively viscous DMSO.
Three observations emerge from these experiments that
provide some insights into the origin of the selectivity. First is
the chemoselectivity of the formation of mixed cycloaddition
products versus the homodimers of 1 and 2. As can be seen
from the results in Tables 2 and 3, polar solvents or sensitizers
that are less likely to complex with an oxidized diene like 3
following electron transfer increase the amount of dimers of 1
relative to the mixed adducts. This data from catalyst variation
supports the idea presented earlier that the formation of free
ion pairs lead to an essentially statistical mixture of products
that reflects the probability of collision of 1•+ with either 1 or
2. The behavior of other forms of 1•+, e.g., as an ion pair with
11•- or as a solvent-separated ion pair, is more complex but
apprarently leads to the preferential formation of the mixed
cycloadditions products.
The second question is the periselectivity of the reaction,
leading to virtually exclusive formation of 4 and 5 in dichlo-
romethane but yielding a substantial amount of other isomers
in more polar solvents, as shown in Tables 2 and 3. Although
none of the other isomers could initially be isolated under the
standard reaction conditions, repeating the reaction in a 50:50
mixture of dichloromethane and acetone followed by rapid
workup gave sufficient material for the spectroscopic charac-
terization of 7. The structure of 7 was then elucidated by 2D
NMR and decoupling experiments that allowed the assignment
of the cis (6.8 Hz) and trans (11.4 Hz) coupling for the hydrogen
at C4 (see Supporting Information). As shown in Figure 3, 7 is
the result of a dienophile attack on the same side as the isopropyl
group forming a cis-diastereomer. Based on the similar MS (see
Suppporting Information) and the fact that formation of 7 and
10 for the most part correlate as a function of solvent polarity
as shown in Table 2, it can be hypothesized that 10 is an isomer
of 7 with the methyl and anisyl groups in the cis position.
However, multiple attempts to isolate 10 in an amount sufficient
for structural characterization were not successful.
The formation of 7 and its dependence on the solvent polarity
can be understood by considering the reactive intermediates
involved. The electronic structure of 1•+ may be represented as
shown in Figure 4 on the left, which is consistent with a B3LYP/
6-31G* computed ChelpG charge of +0.18 at C4 and a spin
density of 0.46 at C1. Attack of 2 could occur in principle at
either position, whereas attacks at C2 and C3 can be excluded
J. Org. Chem. Vol. 73, No. 20, 2008 7913

Sevov and Wiest
FIGURE 4. Possible reaction pathways with energies relative to 12•+ (parenthesis), ChelpG charges (plain text), and spin densities (italics).
FIGURE 5. B3LYP/6-31G* calculated structures of ion-molecule complex 12•+ (left), 15•+ (middle), and reactive conformation of 14•+ (right).
intermediate, which should be faster from the preferentially
formed ion-molecule complex with a trans stereochemistry of
the isopropyl group and 2. However, these effects are likely to
be too small to be studied by implicit solvent models.19d
Considering the structure of 14•+ also helps to rationalize the
third observation, the solvent dependence of the endo/exo ratio
of 5 to 4 shown in Figure 2. Shown in Figure 5 on the right is
the B3LYP/6-31G* calculated structure of the conformation of
14•+ leading to the formation of the endo product 5. In order
for the [4 + 2] ring closure to occur, the vinyl anisole group
has to be in the axial position. The preferred pseudoequatorial
position of the isopropyl group then leads to a steric repulsion
of the anisyl group with the axial hydrogen as shown in Figure
5. This repulsive interaction will be even stronger for the
conformation leading to the exo isomer 4. The solvent depen-
dence of the endo/exo ratio can thus be rationalized as a
competition between the ring closure, which will be faster in
less polar solvents, and single bond rotation in 14•+, which is
stabilized by more polar solvents and thus gives a higher amount
of the more stable endo isomer 5. Thus, the endo/exo ratio is
not governed by secondary orbital overlap as in the case of
neutral Diels-Alder reactions but rather by the steric interactions
in the singly linked intermediate 14•+.
chemoselectivity favoring the formation of mixed Diels-Alder
products relative to homodimers of 1 as well as other mixed
cycloadducts is obtained in dilute (<0.15 M] dichloromethane
solutions using 3+ as a sensitizer. Other reaction conditions lead
to the formation of substantial amounts of other isomers of the
mixed cycloadducts. Surprisingly, the endo/exo ratio is found
to be highly solvent-dependent, with the highest diastereose-
lectivity obtained for DMSO, the most polar solvent studied
here.
From the mechanistic viewpoint, the results of the study can
be rationalized by considering a double stereoselection at the
stage of an ion-molecule complex as well as at the stage of
the singly linked intermediate. The fact that the highest
selectivities are observed for low-polarity solvents is consistent
with the idea that the ion-molecule complexes would be more
stable and that the singly linked intermediate 14•+ would rapidly
collapse to give the product radical cations such as 4•+ and 5•+.
The involvement of the ion-molecule complex could rationalize
the almost complete facial selectivity of the dimerization. More
polar solvents would stabilize free radical cations as well as
the singly linked intermediate long enough to allow other
reaction channels, including back electron transfer for form the
biradical, which could then close to form the [2 + 2] cycload-
ducts such as 7. Such a partitioning of the reaction pathway is
similar to the better-studied case of the dimerization of 2.19
Conclusions
Electron transfer catalysis of the reaction of (R)-R-phellan-
drene with 4-methoxystyrene leads to the rapid formation of
the mixed cycloaddition products in a surprisingly regio-, peri-,
and stereoselective fashion. The systematic variation of the
reaction conditions makes it possible to draw two sets of
conclusions. From the synthetic point of view, the highest
Experimental Section
General Procedure. Commercially available, enantiomerically
pure (95% GC) (R)-R-phellandrene 1 (24 µL) and 4-methoxystyrene
2 (20 µL) were dissolved in 1 mL of dichloromethane. After
addition of 1.6 mg (2 mol %) of 2,4,6-tris(4-methoxyphenyl)
7914 J. Org. Chem. Vol. 73, No. 20, 2008

Electron Transfer Catalyzed Diels-Alder Reaction
pyrylium tetrafluoroborate 316a as electron transfer sensitizer, the
mixture was irradiated for 3 h under nitrogen with light of a
wavelength of λ ) 422 ( 20 nm in a photoreactor equipped with
eight UV lamps. The sensitizer was removed by passing the reaction
through a short silica column with a mobile phase of methylene
chloride and hexane in a 1:1 ratio. The reaction mixtures were then
analyzed in a gas chromatograph using a standard 25 m column.
These general reaction conditions were modified as specified in
the text.
This procedure was scaled up to the synthesis of isomers 4, 5,
and 7 used in the structure elucidation as follows. Using a microliter
syringe 200 µL of 4-methoxystyrene (2) was added to 200 µL of
(R)-(-)-R-phellandrene (1) in 15 mL of CH2Cl2 with 15 mg of 3
(1.9 mol %). Prior to individual isomeric separation, 97 mg of mixed
adducts was collected for a 29% yield of cross addition. For the
synthesis of 7, a 50:50 mixture of dichloromethane and acetone
was used as the solvent. The reaction mixture was degassed and
irradiated at 420 nm for 3 h, and 57 mg of cross adducts (17%
cross addition yield) was isolated using flash chromatography with
5% ethyl acetate in hexanes to separate the mixed adducts from
other reaction products. Isomer separation was accomplished with
argentated silica. AgNO3 (2.75 g) was dissolved in 40 mL of water
and added to 30 g of 60 Å silica. The mixture was stirred until
homogeneity was achieved and left to oven dry overnight while
covered with perforated aluminum foil. When dry, the silver-
impregnated silica was slurry packed as one would with normal
silica, and 2 mL fractions were collected.
Hz, 10.0 Hz, 1H), 1.89 (d J ) 1.4 Hz, 3H), 1.78 (ddd, J ) 12.0
4
2
3
Hz, J ) 2.4 Hz, 9.2 Hz, 1H), 1.44 (m, range ) 18.6 Hz, 1H),
1.40 (m, range ) 29.4 Hz, 1H), 1.11 (m, 3J ) 6.6 Hz, 2.9 Hz, 1H),
1.01 (m, range ) 19.9 Hz, 1H) 0.88 (d, 3J ) 6.6 Hz, 3H), 0.84 (d,
3J ) 6.6 Hz, 3H). 13C NMR (500 MHz, CDCl3) δ 20.3, 20.8, 21.4,
31.7, 33.5, 35.3, 37.0, 41.0, 44.9, 48.5, 55.4, 113.5, 122.1, 128.9,
141.0, 143.4 157.7 ppm. IR (CCl4) 1041.3, 1178.9, 1248.6, 1461.4,
1511.2, 1610.9, 2927.3, 3029.7 cm-1. FABMS [M + H]+ (relative
intensity) 271 (24), 199 (34), 163 (23), 134 (83) 121 (100), 105
(30). Exact mass calculated for C19H26O 270.1984, found 270.1979.
(1S,4R,6R,8S)-4-Isopropyl-8-(4-methoxyphenyl)-1-methylbi-
cyclo[4.2.0]oct-2-ene 7. Compound 7 was formed following the
general procedure but with a different solvent system. A 50:50
mixture of CH2Cl2/acetone was dried over MgSO4. A yield of 3
mg (1% isolated yield; 62% pure by NMR) was collected as a clear
1
3
liquid. H NMR (500 MHz, CDCl3) δ 7.07 (d, J ) 8.7 Hz, 2H),
6.85 (d, 3J ) 8.7 Hz, 2H), 5.64 (d, 3J ) 10.4 Hz, 1H) 5.04 (dd, 3J
3
) 10.4 Hz, 2.7 Hz, 1H) 3.81 (s, 3H), 3.29 (t, J ) 9.4 Hz, 1H),
2.34 (m, range ) 23.8 Hz, 1H), 2.11 (m, range ) 22.6 Hz, 1H),
3
3
2.01 (dd, J ) 9.9 Hz, 8.5 Hz, 2H), 1.67 (m, J ) 6.8 Hz, 1H),
1.60 (m, range ) 22.5 Hz, 1H), 1.22 (m, range ) 29.0 Hz, 1H),
1.21 (s, 3H), 0.94 (d, 3J ) 6.8 Hz, 3H), 0.92 (d, 3J ) 6.8 Hz, 3H).
13C NMR (500 MHz, CDCl3) δ 158.0, 143.5, 133.4, 128.8, 113.4,
55.5, 51.4, 47.0, 43.7, 40.0, 37.9, 37.1, 32.0, 23.8, 19.8, 19.5 ppm.
IR (CCl4) 855.3 1052.3, 1175.9, 1233.4, 1455.4, 1507.2, 1623.1,
2934.2, 3033.7 cm-1. FABMS [M + H]+ (relative intensity) 271
(19), 269 (22), 255 (7), 199 (8), 149 (22), 137 (100), 121 (67), 105
(23). Exact mass calculated for C19H26O 270.1984, found 270.1986.
Computational Studies. All computational studies were per-
formed using the G03 series of programs20 at the B3LYP/6-31G*
level of theory. All geometries were fully optimized without
constraints and the nature and identity of the stationary points was
verified using harmonic frequency calculations and animation of
the negative frequency. Energies are zero-point corrected and given
in kcal/mol relative to the ion-molecule complex 12•+. Partial
charges were calculated using the CHelpG method.
(5S)-5-Isopropyl-8-((4R)-4-methoxyphenyl)-2-methylbicyclo-
[2.2.2]oct-2-ene 4. A yield of 8 mg (2% isolated yield; 96% pure
1
by NMR) was collected as a clear liquid. H NMR (600 MHz,
3
3
CDCl3) δ 7.20 (d, J ) 8.8 Hz, 2H), 6.86 (d, J ) 8.8 Hz, 2H),
5.91 (dd, 3J ) 6.5 Hz, 4J ) 1.5 Hz, 1H), 3.81 (s, 3H), 2.75 (m, 3J
3
) 2.6 Hz, 6.5 Hz, 11.3 Hz, 1H), 2.53 (ddd, J ) 2.4 Hz, 2.2 Hz,
6.6 Hz, 1H), 2.38 (m, range ) 13.2 Hz, 1H), 1.81 (d, 4J ) 1.6 Hz,
2
3
3H), 1.75 (m, range ) 19.5 Hz, 1H), 1.70 (ddd, J ) 12.4 Hz, J
2
3
) 2.9 Hz, 9.2 Hz, 1H) 1.59 (ddd, J ) 12.8 Hz, J ) 2.1 Hz, 6.4
Hz, 1H) 1.35 (m, range ) 14.0 Hz, 1H) 1.07 (m, 3J ) 2.5 Hz, 6.6
Hz, 1H) 0.98 (m, range ) 22.8 Hz, 1H) 0.71 (d, 3J ) 6.6 Hz, 3H)
Acknowledgment. We thank the National Science Founda-
tion (Grant CHE-0415344) for financial support of this work
and Pfizer for a Summer Undergraduate Research Fellowship
for C.S.S.
0.69 (d, J ) 6.6 Hz, 3H). 13C NMR (500 MHz, CDCl3) δ 20.1,
3
20.3, 20.9, 29.9, 33.3, 33.4, 36.7, 38.5, 39.4, 43.0, 55.2, 113.4,
125.8, 129.0, 136.7, 142.4, 157.4 ppm. IR (CCl4) 1041.5, 1178.1,
1247.4, 1511.2, 1610.4, 2868.9, 3030.7 cm-1. FABMS [M + H]+
(relative intensity) 271 (17), 270 (27), 199 (25), 163 (26), 154 (30),
134 (100), 121 (58), 105 (17). Exact mass calculated for C19H26O
270.1984, found 270.1978.
Supporting Information Available: Full citation for ref 20,
Cartesian coordinates and energies for all structures discussed,
GC/MS spectrum of the crude reaction, and 1D and 2D spectra
of 4, 5, and 7. This material is available free of charge via the
Internet at http://pubs.acs.org.
(5S)-5-Isopropyl-8-((4S)-4-methoxyphenyl)-2-methylbicyclo-
[2.2.2]oct-2-ene 5. A yield of 15 mg (4% isolated yield; 97% pure
1
by NMR) was collected as a clear liquid. H NMR (600 MHz,
3
3
CDCl3) δ 7.08 (d, J ) 8.6 Hz, 2H), 6.81 (d, J ) 8.6 Hz, 2H),
JO8002562
3
3
5.64 (d, J ) 6.2 Hz, 1H), 3.80 (s, 3H), 2.83 (ddd, J ) 1.8 Hz,
5.6 Hz, 9.7 Hz, 1H), 2.59 (ddd, 3J ) 1.8 Hz, 1.8 Hz, 6.2 Hz, 1H),
2.42 (m, range ) 11.7 Hz, 1H), 2.02 (ddd, 2J ) 12.8 Hz, 3J ) 2.8
(20) Frisch, M. J. et al. Gaussian 03, ReVision B.05; Gaussian, Inc.: Pittsburgh,
PA, 2003.
J. Org. Chem. Vol. 73, No. 20, 2008 7915