The fragmentation of polyfluorinated benzylic alcohols: the first observation of pentafluorophenyl anion as a good leaving group

Article abstract of DOI:10.1016/j.tetlet.2006.08.069

Treatment of a series of pentafluorobenzylic alcohols with sodium methoxide in DMSO results in a rapid carbon-carbon cleavage reaction, yielding a ketone or aldehyde as well as pentafluorobenzene, which undergoes subsequent nucleophilic aromatic substitution (NAS). In methanol solvent, the fragmentation is very slow, and NAS without fragmentation occurs almost exclusively. In methanol/DMSO mixtures, both processes are observed simultaneously. This appears to be the first report of the fragmentation of pentafluorobenzylic alcohols.

Full text of DOI:10.1016/j.tetlet.2006.08.069

Tetrahedron Letters 47 (2006) 7405–7407
The fragmentation of polyfluorinated benzylic alcohols: the
first observation of pentafluorophenyl anion as a good leaving group
Charles M. Garner* and Henry C. Fisher
Department of Chemistry and Biochemistry, Baylor University, Waco, TX 76798, United States
Received 11 July 2006; accepted 18 August 2006
Abstract—Treatment of a series of pentafluorobenzylic alcohols with sodium methoxide in DMSO results in a rapid carbon–carbon
cleavage reaction, yielding a ketone or aldehyde as well as pentafluorobenzene, which undergoes subsequent nucleophilic aromatic
substitution (NAS). In methanol solvent, the fragmentation is very slow, and NAS without fragmentation occurs almost exclusively.
In methanol/DMSO mixtures, both processes are observed simultaneously. This appears to be the first report of the fragmentation
of pentafluorobenzylic alcohols.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Reactions that result in the cleavage of carbon–carbon
bonds are relatively rare. Most of these reactions involve
alkoxides fragmenting to give ketones and generally
require stabilized carbanion leaving groups.1 Some
common examples include the haloform reaction and
the retro-aldol reaction. However, examples involving
less-stabilized2 or even non-stabilized3 carbanions are
known, though extreme steric crowding3b or high tem-
peratures3c are often required.
and GC–MS, generally with comparison to authentic
materials.
The reaction of 2 (ꢀ0.1 M) with 2 equiv of sodium
methoxide (25 wt% solution in methanol) in polar apro-
tic solvents was very fast (Scheme 2). In DMSO and
DMF, the reaction was complete in <4 min. Acetonitrile
gave a somewhat slower reaction (complete in 5–
10 min), THF slower yet (ꢀ65% complete in 60 min),
and tert-butanol, dichloromethane and benzene were
all much less reactive (ꢀ10% complete at 60 min),
though some insolubility of the methoxide was evident
in the case of benzene.
Several trans-1-aryl-4-tert-butylcyclohexanols are effec-
tive gelling agents for nonpolar organic solvents.4 In
an attempt to prepare derivatives of trans-1-pentafluoro-
phenyl-4-tert-butylcyclohexanol (1) by nucleophilic
aromatic substitution (NAS), we found that the reaction
with sodium methoxide in DMSO resulted in the forma-
tion of 4-tert-butylcyclohexanone and 2,3,5,6-tetraflu-
oroanisole (Scheme 1). Neither the fragmentation nor
the substitution of pentafluorobenzylic alcohols appears
to have been previously reported,5 with the only refer-
ence to a substitution reaction having first required pro-
tection of the hydroxy group.6 We initiated a study of
this reaction using the more-available (and non-gelling)
cis isomer 2. Although GC–MS was initially used (and
verified the formation of ketone), the best technique
for studying these reactions proved to be 19F NMR.
The identities of all species were established by NMR
We found that the very rapid cleavage reaction observed
in DMSO slows considerably with the addition of
increasing amounts of methanol, and NAS without
cleavage becomes the dominant reaction. The reaction
of trans isomer 1 was slow enough in 35:65 (v/v)
DMSO-d6:CH3OH at room temperature (ꢀ50% com-
plete in 1.5 h) that it could be conveniently monitored
by 19F NMR. Under these conditions (ꢀ0.1 M in 1,
2 equiv NaOMe), we could observe both the NAS prod-
uct before cleavage (3) and pentafluorobenzene, indicat-
ing that the reaction was clearly proceeding by two
pathways. The 19F spectra revealed that the initial step
of path A (Scheme 3) was 5–10 times faster than the
initial step of path B. Pentafluorobenzene serves as a
reactive intermediate, present only at low levels (620%
of all fluorinated species) with its initial rate of forma-
tion being only somewhat faster than its conversion to
*
Corresponding author. Tel.: +1 254 710 6862; fax: +1 254 710
2403; e-mail: charles_garner@baylor.edu
0040-4039/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2006.08.069

7406
C. M. Garner, H. C. Fisher / Tetrahedron Letters 47 (2006) 7405–7407
F
F
F
F
OCH
3
F
F
F
F
NaOCH3
DMSO
O
+
OH
F
H
H
H
1
Scheme 1.
OCH
H
OH
F
3
F
F
F
F
O
F
NaOMe
+
various
solvents
H
F
H
F
F
2
Scheme 2.
F
OCH
3
F
F
F
F
F
F
OCH
H
3
F
NaOCH3
F
NaOCH3
F
F
F
F
O
path A
+
OH
OH
F
H
H
H
3
1
NaOCH3
F
F
F
O
path B
+
F
NaOCH3
H
H
Scheme 3.
tetrafluoroanisole. Under these conditions, synthetically
useful amounts (ꢀ60–70%) of substitution product 3
accumulated; methoxy substitution clearly inhibits sub-
sequent cleavage. However, slow cleavage of compound
3 was also evident, confirmed by submitting isolated
samples to the same reaction conditions.
However, in DMSO-d6/methanol mixtures, differences in
the reactivity of the various alcohols became apparent.
Conveniently slow reaction rates were obtained
in 45:55 (v/v)% DMSO-d6:CH3OH using 0.1 M substrate
and 5 equiv of sodium methoxide at 25 ꢁC. The reactions
were monitored by 19F NMR to >90% disappearance of
starting alcohol in all cases. The pseudo first-order plots
exhibited good linearity (r2 P 0.98) and an almost 10-
fold difference in reactivities (Table 1). In particular,
the axial aryl group in 1 makes that substrate easily the
most reactive due to relief of strain resulting from the
fragmentation reaction.
A variety of pentafluorobenzylic alcohols (4–7, Scheme
4) were prepared to evaluate the generality of the
reaction, that is, to determine whether ring size, acyclic
nature, and aldehyde versus ketone products had a
significant effect. These were made by the addition of
pentafluorophenylmagnesium bromide7 to the corre-
sponding ketone or aldehyde, and all have been reported
previously8 except 5.9 Upon exposure to sodium meth-
oxide in DMSO-d6, all of these underwent rapid cleav-
age (100% complete in <4 min) to tetrafluoroanisole.
Table 1. Pseudo first-order kinetics for the disappearance of pentaflu-
orobenzylic alcohols after treatment with 5 equiv of NaOMe in 45:55
(v/v)% DMSO-d6:CH3OH
Compound
Rate constant
(minÀ1
Linearity (r2)
Half-life
(min)
C F
C F
)
HO
6
5
C F
6
5
HO
6
5
C F
HO
6
5
1
2
4
5
6
7
4.7 · 10À2
5.1 · 10À3
9.4 · 10À3
7.4 · 10À3
1.4 · 10À2
9.4 · 10À3
0.978
0.996
0.997
0.998
0.989
0.991
15
132
74
94
51
OH
CH
H C
3
3
4
5
6
7
74
Scheme 4.

C. M. Garner, H. C. Fisher / Tetrahedron Letters 47 (2006) 7405–7407
7407
The mechanism of the first step in pathway B is clearly
Acknowledgments
formation of the alkoxide followed by cleavage to the
ketone and the pentafluorophenyl anion, the latter pro-
tonating to form the observed C6F5H. In theory, the
pentafluorophenyl anion could re-add to the carbonyl
group, and an experiment designed to test this was done.
The trans alcohol 1 in THF was treated with butyl-
lithium, and besides cleavage products, small amounts
of the cis isomer 2 were observable by GC–MS. Thus,
the pentafluorophenyl anion appears to be an intermedi-
ate in the cleavage.
We thank the Robert A. Welch Foundation (Grant
#AA-1395) for partial support of this work, and the
National Science Foundation (Award #CHE-0420802)
for funding the purchase of our 500 MHz NMR.
References and notes
1. Wietzerbin, K.; Bernadou, J.; Meunier, B. Eur. J. Inorg.
Chem. 2000, 1391–1406.
2. See, for example, allylic anion leaving groups in: Benkeser,
R. A.; Siklosi, M. P.; Mozdzen, E. C. J. Am. Chem. Soc.
1978, 100, 2134–2139.
3. (a) Lomas, J. S.; Dubois, J.-E. J. Org. Chem. 1984, 49,
2067–2069; (b) Arnett, E. M.; Small, L. E.; McIver, R. T.;
Miller, J. S. J. Org. Chem. 1978, 43, 815–817; (c) Zook, H.
D.; March, J.; Smith, D. F. J. Am. Chem. Soc. 1959, 81,
1617–1620.
4. Garner, C. M.; Terech, P.; Allegraud, J.-J.; Mistrot, B.;
Nguyen, P.; de Geyer, A.; Rivera, D. J. Chem. Soc.,
Faraday Trans. 1998, 94, 2173–2179.
5. As of this writing, Scifinder appears to have no references
to the decomposition of pentafluorobenzylic alcohols under
any conditions to form either pentafluorobenzene or the
pentafluorophenyl anion.
Given the propensity for pentafluorobenzylic alkoxides
to undergo cleavage, it is interesting to consider how
Grignard reactions, which initially produce alkoxides,
are able to produce the corresponding alcohols with
any efficiency. The answer appears to involve the
degree of negative charge on the alkoxide oxygen,
and depends on the nature of the solvent and of the
metal. As observed in our initial solvent studies, alkox-
ides are much more prone to cleavage in the more
highly ionizing dipolar aprotic solvents rather than in
the ether solvents used for Grignard reactions. Previ-
ous studies3a on the cleavage of alkoxides by loss of
unstabilized carbanions found that DMSO promoted
the reactions, and HMPT greatly so. In addition, the
more covalent magnesium alkoxides are less prone to
cleavage than are the more ionic sodium salts. The
latter point was established by treating alcohol 2 with
magnesium methoxide in DMSO; only partial cleavage
(ꢀ50%) occurred in 15 min, as opposed to complete
cleavage occurring almost immediately using sodium
methoxide.
6. Crowley, P. J.; Eur. Pat. Appl. 1982, EP 82-300641 (CAN
98:125610).
7. Preparation of C6F5MgBr, see: Respess, W. L.; Tamborski,
C. J. Organomet. Chem. 1968, 11, 619–622. This reagent is
also available commercially from Aldrich or Fluka.
8. Preparation of 4, see: Ramachandran, P. V.; Jennings, M. P.
Org. Lett. 2001, 3, 3789–3790; Preparation of 6, see: Allen, A.
D.; Kwong-Chip, J. M.; Mistry, J.; Sawyer, J. F.; Tidwell, T.
T. J. Org. Chem. 1987, 52, 4164–4171; Preparation of 7, see:
Filler, R.; Kang, H. H. J. Org. Chem. 1975, 40, 1173–1175.
9. NMR (CDCl3) data for compound 5: 1H: 1.80 (m, 2H), 1.93
(m, 2H), 2.09 (m, 3H, incl. OH), 2.36 (m, 2H). 13C: 22.4, 40.4
(t, J = 4.5), 81.2, 119.2 (tm, J = 63), 137.7 (dm, J = 238),
138.9 (dm, J = 240), 145.3 (dm, J = 242). 19F (relative to
C6H5F at À113.26): À138.1 (approx. dd, J = 23, 6, 2F),
À157.8 (tt, J = 23, 6, 1F), À163.7 (approx. td, J = 23, 6, 2F).
We have established the ease and generality of cleavage
of polyfluorobenzylic alcohols, as well as conditions
under which substitution without cleavage can be
accomplished. Those wishing to accomplish synthetic
manipulations of these types of compounds should be
aware of the pertinent considerations.