Anodic coupling reactions: Exploring the generality of Curtin-Hammett controlled reactions

Article abstract of DOI:10.1021/ol200182f

Intramolecular anodic olefin coupling reactions can be compatible with the presence of dithioketal protecting groups even though the dithioketal oxidizes at a lower potential than either of the groups participating in the cyclization. In such cases, product formation is controlled by the Curtin-Hammett Principle. In this study, the generality of such reactions is examined along with the use of alternative reaction conditions to suppress unwanted side reactions.

Full text of DOI:10.1021/ol200182f

ORGANIC
LETTERS
2011
Vol. 13, No. 7
1678–1681
Anodic Coupling Reactions: Exploring the
Generality of Curtin-Hammett Controlled
Reactions
Alison Redden and Kevin D. Moeller*
Department of Chemistry, Washington University, St. Louis, Missouri 63130,
United States
moeller@wustl.edu
Received January 20, 2011
ABSTRACT
Intramolecular anodic olefin coupling reactions can be compatible with the presence of dithioketal protecting groups even though the dithioketal
oxidizes at a lower potential than either of the groups participating in the cyclization. In such cases, product formation is controlled by the
Curtin-Hammett Principle. In this study, the generality of such reactions is examined along with the use of alternative reaction conditions to
suppress unwanted side reactions.
Anodic olefin coupling reactions provide a unique op-
portunity for both synthesizing new ring skeletons and
probing the chemistry of reactive radical cation
intermediates.1,2 In these reactions, the radical cation
formed at the anode can undergo a reversible, intramole-
cular electron-transfer reaction.3,4 In such cases, the pro-
duct formed from the oxidation is governed by the
Curtin-Hammett Principle. An example of this behavior
isillustratedinScheme1.3 Thedithioketalin substrate1 has
an oxidation potential of Ep/2 = þ1.1 V vs Ag/AgCl while
the enol ether has an oxidation potential of Ep/2 = þ1.4 V
vs Ag/AgCl. Hence, oxidation of 1 generates radical cation
2. This radical cation undergoes an electron-transfer reac-
tion to form an enol ether radical cation (3) that in turn
triggers a cyclization and the formation of 4 in a 70%
isolated yield. If the alcohol in the substrate is not present,
then the only product observed is derived from radical
cation 2.
Curtin-Hammett controlled reactions of this type are
potentially very useful. Consider the retrosynthesis of
artemisolide outlined in Scheme 2.5,6 A Michael reac-
tion-anodic oxidation sequence analogous to the one
developed by the Wright and Trauner groups is proposed.7
In this case, the presence of a dithioketal in 5 would both
aid the Michael reaction by stabilizing the anion and serve
as a carbonyl equivalent for setting up C7 in the final
product. The cyclization would take advantage of an enol
ether-furan coupling reaction.8 The success of this
(5) (a) For the isolation of Artemisolide, see: Dim, J. H.; Kim, H.-K.;
Jeon, S. B.;Son, K.-H.;Kim, E. H.;Kang, S. K.;Sung, N.-D.;Kwon, B.-M.
Tetrahedron Lett. 2002, 43, 6203. (b) For the isolation of the related
Arteminolides, see: Lee, S.-H.; Kim, H.-K.; Seo, J.-M.; Kang, H.-M.;
Kim, J. H.; Son, K.-H.; Lee, H.; Kwon, B.-M. J. Org. Chem. 2002, 67,
7670.
(6) For synthetic studies targeting the related Arteminolides, see: (a)
Sohn, J.-H. Bull. Korean Chem. Soc. 2010, 31, 1841. (b) Sohn, J.-H. Bull.
Korean Chem. Soc. 2009, 30, 2517. (c) Lee, H.-Y.; Sohn, J.-H.; Kim,
H. Y. Tetrahedron Lett. 2001, 42, 1695.
(7) For lead references, see: (a) Wright, D. L.; Whitehead, C. R.;
Sessions, E. H.; Ghiviriga, I.; Frey, D. A. Org. Lett. 1999, 1, 1535. (b)
Miller, A. K.; Hughes, C. C.; Kennedy-Smith, J. J.; Gradl, S. N.;
Trauner, D. J. Am. Chem. Soc. 2006, 128, 17057.
(8) Please see: Wu, Honghui; Moeller, Kevin D. Org. Lett. 2007, 9,
4599 and references therein.
(1) For a recent review, see: Moeller, K. D. Synlett. 2009, 8, 1208.
(2) For reviews covering the use of electrochemical methods in
synthesis, see: (a) Sperry, J. B.; Wright, D. L. Chem. Soc. Rev. 2006,
35, 605. (b) Yoshida, J.; Kataoka, K.; Horcajada, R.; Nagaki, A. Chem.
Rev. 2009, 108, 2265.
(3) Duan, S.; Moeller, K. D. J. Am. Chem. Soc. 2002, 124, 9368.
(4) (a) For an intramolecular electron-transfer reaction involving a
radical cation and a 4-methoxyphenyl ring as an electron donor, see:
Chiba, K.; Miura, T.; Kim, S.; Kitano, Y.; Tada, M. J. Am. Chem. Soc.
2001, 123, 11314. (b) For an earlier example involving the anodic
oxidation of an aromatic ring and an amide as an electron donor, see:
Moeller, K. D.; Wang, P. W.; Tarazi, S.; Marzabadi, M. R.; Wong, P. L.
J. Org. Chem. 1991, 56, 1058.
r
10.1021/ol200182f
Published on Web 03/04/2011
2011 American Chemical Society

Scheme 1
Scheme 3
Scheme 4
Scheme 2
CH2Cl2 solvent, 2,6-lutidine as a proton scavenger, and a
constant current of 8 mA), a 60% isolatedyield of the cyclic
product 11a was obtained. A 1H NMR of the crude reac-
tion material showed only the desired product suggesting that
the lower isolated yield for the reaction resulted from instability
of the product to chromatography. Clearly, formation of the
seven-membered ring did not interfere with Curtin-Hammett
control of the oxidation. An oxidation of the eight-membered
ring substrate 10b did not lead to the same conclusion. In this
case, the reaction led to a mixture of products, none of which
were consistent with cyclized material.
Attention was then turned toward probing the compat-
ibility of the cyclizations with less reactive carbon-based
nucleophiles. This work began by examining bis-enol ether
substrate 12.10 When 12 was oxidized using the same
conditions described above (Scheme 4), the reaction led
to cyclic product 13 in a 21% yield along with 16% of the
acyclic dimethoxyketalproduct14(derived fromoxidation
of the dithioketal) and 19% of the overoxidized cyclic
product 15. The formation of the acyclic product 14 high-
lighted how much slower the C-C bond forming reaction
is relative to alcohol trapping of the radical cation. For the
carbon-carbon bond forming reaction, trapping of the
dithioketal radical cation by methanol competed with the
cyclization even when the cyclization afforded a five-
membered ring instead of the seven-membered ring gener-
ated from the oxidation of 10a.
cyclization would require a Curtin-Hammett type scenario
since the oxidation potential of the dithioketal is significantly
lower than that of either the enol ether or the furan ring.
But will the oxidation of 7 really give rise to a reaction
consistent with the Curtin-Hammett Principle? Anodic
coupling reactions forming carbon-carbon bonds are
typically slower than reactions leading to carbon-oxygen
bonds,9 and the formation of the seven-membered ring
proposed in Scheme 2 should be significantly slower than
formation of the six-membered ring generated from the
oxidation of 1. Is the proposed cyclization still fast enough
to drain the reaction toward the formation of 8? How
general is the conclusion reached for the oxidation of 1?
Initially, this question was addressed for the formation of
medium-size rings. To this end, both seven- and eight-
membered ring substrates (Scheme 3) were synthesized in a
fashion directly analogous to 1.10 When the seven-mem-
bered ring substrate 10a was oxidized using conditions
identical to those employed for the oxidation of 1
(a reticulated vitreous carbon (RVC) anode, a carbon
cathode, lithium perchlorate electrolyte, 20% MeOH/
Cyclic voltammetry data suggested a method for aiding
-
the cyclization originating from 12 (Figure 1).11 The Ep/2
value measured for an isolated methoxy enol ether
(compound 16) was þ1.40 V vs a Ag/AgCl reference
(9) For an example, see: Frey, D. a.; Reddy, S. H. K.; Wu, N.;
Moeller, K. D. J. Org. Chem. 1999, 64, 2805.
(10) Synthetic details are contained in the Supporting Information.
(11) Ep/2 values were measured using a Ag/AgCl reference electrode,
a Pt anode, a 0.1 N LiClO4 in CH3CN electrolyte solution, and a sweep
rate of 25 mV/s.
Org. Lett., Vol. 13, No. 7, 2011
1679

Scheme 5
Figure 1. Potentials measured by CV relative to a Ag/AgCl
reference electrode.
thereby controlling the nature of the solution surrounding the
electrode.14 Double layers formed by hydrophobic electro-
lytes exclude methanol from the surface of the anode. The
more hydrophobic the electrolyte used, the more methanol
excluded. The oxidation of 12 provides a nice example of this
situation. When 12 was oxidized using a 0.1 M tetraethyl-
ammonium tosylate solution in place of the lithium perchlo-
rate solution used in the intitial experiment, a 51% isolated
yield of the cyclic dithioketal product 13 was obtained along
with only 6% of the acyclic dithioketal product 14 and 10%
of the cyclic overoxidized material 15 (Scheme4).Theuseofa
0.5 M Et4NOTs electrolyte solution improved the reaction
even further producing a 60% isolated yield of 13. Only 2%
of 14 was formed along with 7% of 15. While the oxidation
potentials given in Figure 1 suggest that the overoxidized
product 15 can be produced from either the oxidation of the
dithioketal in 13 or an oxidative cyclization of 14, the
observation that the yield of 14 drops off faster than the yield
of 15 is consistent with 15 arising primarily from overoxida-
tion of 13. In either case, the oxidation leads to a 67% isolated
yield of the cyclic product.
The influence of the electrolyte on the reaction was also
observed using a less reactive allylsilane trapping group for
the enol ether radical cation (Scheme 5).15 Once again, the
yield of cyclized product dramatically improved with the
use of tetraethylammonium tosylate as the electrolyte.
Using the quaternary ammonium salt, a 61% combined
yield of cyclic product was isolated.
Because furan rings are among the most reactive anodic
olefin coupling partners studied to date,8 the success of the
cyclizations arising from 12 and 18 bode well for the
proposed enol ether-furan coupling reaction highlighted
in Scheme 2. With this in mind, six-membered ring sub-
strate 22 was synthesized10 and oxidized (Scheme 6).
The reactivity of the furan ring was immediately evident
when the reaction led to a good yield of cyclic product even
with lithium perchlorate as the electrolyte. No evidence
was observed for methanol trapping of the dithioketal
radical cation 25. The initial cyclization product 23 was
not isolated but instead treated with toluenesulfonic acid
to directly obtain the tricyclic furan.7,10 Like previous
Curtin-Hammett controlled cyclizations, the reaction
electrode. For a substrate leading to a fast cyclization, a
lower oxidation potential would be expected.12 This occurs
because fast radical cation initiated cyclizations happen at
or near the electrode surface. Hence, the cyclization removes
the radical cation intermediate from the surface of the anode
resulting in a shift in the equilibrium at the electrode surface
toward the product. The shift in equilibrium lowers the
potential measured for the substrate. The faster the cycliza-
tion reaction, the lower the potential measured. For exam-
ple, consider substrate 17 in Figure 1. In this case, the
formation of a five-membered ring and a gem-dialkyl effect
afford a fast cyclization and a 300 mV drop in the oxidation
potential measured for the substrate. For comparison, the
potential measured for a similar substrate without the gem
methyls is þ1.2 V vs Ag/AgCl (a drop of only 200 mV
relative to the enol ether).13 The faster cyclization (gem
methyls present) has a lower observed potential. As ex-
pected, the oxidation potential measure for 14 shows a drop
similar to that measured for 17. In the case of 12, a lower
potential is observed, but this potential reflects oxidation of
the dithioketal and not simply the cyclization. Nevertheless,
the oxidative cyclization arising from 12 should proceed in a
fashion very similar to that of both 17 and 14. In other
words, it should also occur at or near the electrode surface.
In such reactions, decreasing the amount of methanol
close to the anode should favor the cyclization. A lower
concentration of methanol by the anode means slower
competitive trapping of the dithioketal radical cation.
However, decreasing the concentration of methanol in
the electrolysis is not necessarily an option. Methanol is
needed in sufficient concentration to serve as both the
substrate reduced at the cathode and as the nucleophile to
trap the oxonium ions resulting from the cyclization.
Fortunately, methanol can serve both of these purposes
while having its effective concentration selectively lowered
near the surface of the anode by employing a less-polar
electrolyte. The electrolyte forms a double layer at the anode
(12) For examples, please see: Xu, H.; Moeller, K. D. Angew. Chem.,
Int. Ed. 2010, 49, 8004 and references therein.
(13) Moeller, K. D.; Tinao, L. V. J. Am. Chem. Soc. 1992, 114, 1033.
(14) (a) For general discussions of how the adsorption of a substrate
and an electrolyte onto an electrode surface can exclude solvent and
protect a reactive intermediate, see: Organic Electrochemistry: 4th ed.,
Revised and Expanded; Lund, H., Hemmerich, O., Eds.; Marcel Dekker:
New York-Basel, 2001; p 802. Synthetic Organic Electrochemistry, 2nd
ed.; Fry, A. J., Ed.; John Wiley and Sons: New York, 1989; p 126. (b) For a
detailed example, see: Fry, A. J.; Reed, R. G. J. Am. Chem. Soc. 1972, 94,
8475.
(15) For evidence showing the lower reactivity of allylsilane trapping
groups, see: (a) Huang, Y.; Moeller, K. D. Tetrahedron 2006, 62, 6536.
(b) Hudson, C. M.; Marzabadi, M. R.; Moeller, K. D.; New, D. G.
J. Am. Chem. Soc. 1991, 113, 7372. (c) Tinao-Wooldridge, L. V.;
Moeller, K. D.; Hudson, C. M. J. Org. Chem. 1994, 59, 2381.
1680
Org. Lett., Vol. 13, No. 7, 2011

Scheme 6
Scheme 7
Scheme 8
dichloromethane to expand the central ring in the desired
direction.16 The reaction led to a 60% isolated yield of two
seven-membered ring products in a 1.6:1 ratio. A HMQC-
TOCSY experiment was used to assign the connectivity of
the two products showing that the major isomer was the
desired 29 and the minor isomer the product of ring expan-
sion in the opposite direction. Clearly, the Curtin-Hammet
controlled anodic cyclization can provide access to the ring
skeleton needed for the synthesis of artemisolide, but work
is still needed to improve the yield and regioselectivity of the
ring-expansion reaction.
In conclusion, we have extended the scope of Curtin-
Hammett controlled anodic cyclizations by examining the
compatibility of coupling reactions conducted in the pre-
sence of a dithioketal protecting group with both the
formation of medium-size rings and the generation of
carbon-carbon bonds. Cyclizations utilizing an alcohol
trapping group are fast and show no signs of competitive
trapping of the dithioketal radical cation even when forming
a seven-membered ring. Carbon-carbon bond forming
reactions utilizing reactive furan trapping groups are also
fast, although not fast enough to allow for seven-membered
ring formation. Slower carbon-carbon bond forming reac-
tions benefit strongly from the use of less polar electrolytes
that selectively lower the concentration of methanol in the
region of the reaction surrounding the anode. Applications
of these reactions to total syntheses efforts are underway.
benefited from elevated temperature, presumably by fa-
voring the intramolecular cyclization over bimolecular
methanol trapping.3 Surprisingly, the reaction did not pro-
ceed as well when tetraethylammonium tosylate was used as
the electrolyte. It appears that the use of the less polar
electrolyte did too good of a job excluding methanol from
the reaction medium surrounding the anode. Reactions
involving furan rings require sufficient methanol to be present
in order to trap the initial cyclic cation leading to 23. If this
trapping reaction does not occur, then the elimination of a
proton regenerates an electron-rich furan ring with a lower
oxidation potential than the substrate. Eventually, it was
found that the use of a 0.5 M LiClO4 electrolyte solution and
elevated temperature afforded a 70% isolated yield of the
tricyclic furan after treatment of the initial product with acid.
The reaction could not be extended to the formation of a
seven-membered ring (Scheme 7).10 The reaction again
benefited from the use of LiClO4 as the electrolyte, but under
the optimal conditions (in this case 0.1 M electrolyte) only a
9% isolated yield of the tricyclic product 28 was obtained.
None of the product derived from simple oxidation of the
dithioketal was observed. Instead, the reaction led to a com-
plex mixture of products. Related experiments suggest that an
elimination of Ha from the enol ether radical cation can
compete with the cyclization. However, no evidence for any
one alternative product was obtained from the oxidation of 27.
The failure of the seven-membered ring cyclization
suggested that the synthesis of artemisolide would best
take advantage of the six-membered ring cyclization fol-
lowed by a ring expansion (Scheme 8). To this end, the
tricyclic electrolysis product was reduced, the resulting
alcohol protected, and the dithioketal cleaved to deprotect
the six-membered ring ketone. The ketone was then
Acknowledgment. We thank the National Science Founda-
tion (CHE-0809142) for their generous support of this work.
Supporting Information Available. A sample experi-
mental procedure is included for the oxidation reaction
along with characterization data for all new compounds.
This material is available free of charge via the Internet at
http://pubs.acs.org.
treated with trimethylsilyldiazomethane and BF3 Et2O in
3
Org. Lett., Vol. 13, No. 7, 2011
(16) Saito, T.; Nakata, T. Org. Lett. 2009, 11 (1), 113.
1681