Formation of transient and long-lived cyclopropenyl anions [6]

Full text of DOI:10.1021/ja991224o

7272
J. Am. Chem. Soc. 1999, 121, 7272-7273
Formation of Transient and Long-Lived
Cyclopropenyl Anions
Tina L. Arrowood and Steven R. Kass*
Department of Chemistry
UniVersity of Minnesota
a cyclopropenyl anion intermediate, such a species is not required
mechanistically. A hypervalent penta- or hexacoordinate silicon
anion cannot be ruled out.
Minneapolis, Minnesota 55455
ReceiVed April 19, 1999
To distinguish between these possibilities a chiral cyclopropene
(4) was synthesized by photochemical addition of L-menthyl
(trimethylsilyl)diazoacetate to tert-butyl-phenylacetylene (eq 3).
Cyclobutadiene has been viewed as the Mona Lisa of organic
chemistry by Cram et al.1 This description also applies to
cyclopropenyl anion since it too has captured the imagination of
chemists, inspires wonder, and has been the subject of consider-
able effort.2-14 However, in contrast to cyclobutadiene, cyclo-
propenyl anion remains largely unknown. A more apt analogy,
therefore, might be to compare it to the pyramids since it is unclear
how they were constructed or how a long-lived cyclopropenyl
anion can be prepared in condensed media.
In 1994 we reported the first preparation of a stable cyclopro-
penyl anion by reacting fluoride ion with 3-carbomethoxy-3-
trimethylsilylcyclopropene in the gas phase (eq 1).15 In this work
we provide condensed-phase mechanistic evidence that the
fluoride-induced desilylation of a 3-trimethylsilylcyclopropene
derivative leads to a transient cyclopropenyl anion intermediate.
The first preparation of a long-lived cyclopropenyl anion in
solution and its corresponding UV-visible spectrum also are
presented.
The 2 diastereomers (4a and 4b) were separated by MPLC to
give the optically pure (g98%) esters. These substrates were
reacted with TBAF and Schwesinger’s P2-F phosphazenium
fluoride source; the latter reagent is a “naked” and extremely
reactive source of fluoride.19,20 If a cyclopropenyl anion inter-
mediate is formed, racemization at C-3 is expected although it is
not required because of the chiral menthyl group (pathway 1,
Scheme 1). However, if a penta- or hexacoordinate silicon
intermediate reacts with the electrophile then retention of con-
figuration at C-3 should result (pathway 2, Scheme 1). In both
cases when the desilylation reaction was carried out in the
presence of benzaldehyde four benzylic alcohols were formed in
1
a 1:1: 0.7:0.7 ratio as indicated by H NMR (eq 4).
Valuable information regarding the structure and reactivity of
cyclopropenyl anions can, in principle, be obtained by studying
the fluorodesilylation of 3-trimethylsilylcyclopropenes. In an
important experiment, Borden et al. looked at the desilylation of
13C-labeled 1,2,3-triphenyl-3-trimethylsilylcyclopropene with tetra-
n-butylammonium fluoride (TBAF) in tetrahydrofuran.16 They
found that the 13C label scrambled around the ring in the proton-
trapped product. We have explored the regioselectivity in unsym-
metrical substrates and have found that carbon electrophiles can
be used to give good yields of the trapped cyclopropene (eq 2).17,18
While these results are consistent with the putative formation of
Separation of two of the benzylic alcohols and a mixture of
the other two was achieved via HPLC, but subsequent chemical
transformations were carried out on equimolar 5a/5b and 5c/5d
mixtures. Oxidation of the trapped alcohols with MnO2 affords
the corresponding phenyl ketones in which a chiral center has
been eliminated. Both sets of compounds afforded diastereomeric
ketones, which indicates that 5a and 5b, as well as 5c and 5d,
have different configurations at C-3. In a similar manner, reduction
of the menthyl esters with DIBAL at -78 °C afforded enanti-
omers. Again, this shows that racemization occurred at C-3. The
protodesilylation products (6a and 6b), which are formed with
either fluoride source due to the presence of protic impurities,21
also were produced in a 1:1 ratio.22 This indicates that complete
racemization at C-3 takes place. Pathway 1 in Scheme 1,
consequently, appears to be operating, and a cyclopropenyl anion
intermediate is formed in these desilylation reactions. This
(1) Cram, D. J.; Tanner, M. E.; Thomas, R. Angew. Chem., Int. Ed. Engl.
1991, 30, 1024-1027.
(2) Breslow, R. Angew. Chem., Int. Ed. Engl. 1968, 7, 565-570.
(3) Breslow, R. Chem. Br. 1968, 4, 100-101.
(4) Breslow, R. Pure Appl. Chem. 1971, 28, 111-130.
(5) Breslow, R. Acc. Chem. Res. 1973, 6, 393-398.
(6) Breslow, R. Pure Appl. Chem. 1982, 54, 927-938.
(7) Winkelhofer, G.; Janoschek, R.; Fratev, F.; Spitznagel, G. W.; Chan-
drasekhar, J.; Schleyer, P. v. R. J. Am. Chem. Soc. 1985, 107, 332-337.
(8) Garratt, P. J. Aromaticity; John Wiley and Sons: New York, 1986.
(9) Schleyer, P. v. R.; Kaufmann, E.; Spitznagel, G. W. Organometallics
1986, 5, 79-85.
(10) Li, W.-K. J. Chem. Research 1988, 220-221.
(11) Minkin, V. I.; Glukhovtsev, M. N.; Simkin, B. Y. Aromaticity and
Antiaromaticity: Electronic and Structural Aspects; Wiley-Interscience: New
York, 1994.
(12) Glukhovtsev, M. N.; Laiter, S.; Pross, A. J. Phys. Chem. 1996, 100,
17801-17806.
(13) Klicic, J.; Rubin, Y.; Breslow, R. Tetrahedron 1997, 53, 4129-4136.
(14) Merrill, G. N.; Kass, S. R. J. Am. Chem. Soc. 1997, 119, 12322-
12337.
(19) The structures of P2-F and P5-5 are as follows: (Me2N)3PdN+d
P(NMe2)3 F- and ((Me2N)3PdN)4P+ F-, respectively. Schwesinger, R.; Link,
R.; Thiele, G.; Rotter, H.; Honert, D.; Limbach, H.; Ma¨nnle, F. Angew. Chem.,
Int. Ed. Engl. 1991, 30, 1372-1375.
(20) Link, R.; Schwesinger, R. Angew. Chem., Int. Ed. Engl. 1992, 31,
850.
(21) When THF-d8 was used as the solvent, no deuterium was incorporated
into 6. This strongly suggests that this product does not arise via a radical
pathway.
(22) Trapping in DMSO also leads to a 1:1 mixture of proton-trapped
products.
(15) Kass, S. R.; Sachs, R. K. J. Am. Chem. Soc. 1994, 116, 783-784.
(16) Koser, H. G.; Renzoni, G. E.; Borden, W. T. J. Am. Chem. Soc. 1983,
105, 6359-6360.
(17) Han, S.; Kass, S. R. J. Chem. Soc., Perkin Trans 1 1999, 1553-
1558.
(18) Han, S. Ph.D. Thesis; University of Minnesota, 1997.
10.1021/ja991224o CCC: $18.00 © 1999 American Chemical Society
Published on Web 07/27/1999

Communications to the Editor
J. Am. Chem. Soc., Vol. 121, No. 31, 1999 7273
Scheme 1
methodology thus may be useful for measuring the inversion
barrier of a pyramidal cyclopropenyl anion.
Substitution of a phenyl group for the tert-butyl substituent in
4 will lead to a more stable cyclopropenyl anion. We therefore
decided to examine the fluorodesilylation of 3-carbomethoxy-
1,2-diphenyl-3-trimethylsilylcyclopropene (1) in the absence of
an added electrophile. TBAF is unsuitable for this purpose because
of the residual water in the reagent; protodesilylation takes place
in this case.17 Schwesinger’s P2-F and P5-F fluoride sources,
however, can be made anhydrous and relatively free of protic
(alcohol) contamination.19,20 Consequently, P2-F or P5-F was
reacted with 1 at -78 °C in the absence of an added electrophile.
Upon mixing the reagents the solution turns dark blue/green and
the color persists for extended periods of time before turning
yellow.23 A water quench of the reaction mixture causes the
yellow color to disappear and leads to the recovery of proton-
trapped material along with o- and m-tetraphenylterephthalate (eq
5); the meta isomer predominates at low temperature whereas
Figure 1. UV/visible spectrum of 7 at -72 °C (solid line) and the
spectrum after quenching with benzaldehyde (dashed line).
solution, the color rapidly disappears, and a small amount of
trapped alcohol (2) is obtained along with 3, 8, and 9.
A low temperature (-72 °C) UV-visible spectrum of the blue/
green solution was obtained (Figure 1). Upon addition of the
fluoride source to the cyclopropene new bands at 380, 480, and
600 nm were simultaneously observed. Subsequent addition of
benzaldehyde led to the rapid disappearance of the 480 and 600
nm absorptions but not the feature at 380 nm. Independent
preparation of (TMSF2)- P5+ via the reaction of TMSF with P5-F
revealed that this compound is responsible for the high energy
band at 380 nm. As none of the starting materials or products
absorb above 350 nm, we attribute the absorbances at 480 and
600 nm to the cyclopropenyl anion 7. CIS calculations on
3-carbomethoxycyclopropenyl anion (7 without the two phenyl
substituents) are in good accord with this assignment in that
absorbances at 426 and 545 nm are predicted and the absence of
the phenyl groups should lead to a blue shift.28
the ortho product and the terephthalates are favored at higher
temperature.24-27 If benzaldehyde is added to the blue/green
(23) The lifetime of the blue/green color depends on the amount of methanol
or 2-propanol present in the fluoride source. We have prepared solutions which
persist for up to 1 h, but longer times are possible since all of our fluoride
reagents have had some protic impurities.
(24) The meta product can be accounted for by a pathway previously
suggested by Breslow (refs 25-27) involving the addition of 7 to either 1 or
3. The ortho product requires a different pathway, and we suggest that at
elevated temperatures 7 decomposes to its corresponding radical which rapidly
dimerizes. The dimer subsequently undergoes a fluoride-induced rearrangement
to 8.
Additional experiments and computations on transient and long-
lived cyclopropenyl anions will be reported in due course.
Acknowledgment. Helpful conversations with Prof. T. Hoye and Dr.
C. Toia along with the use of Professor W. Tolman’s low temperature
UV-vis apparatus are gratefully acknowledged. Support from the National
Science Foundation, the donors of the Petroleum Research Foundation,
as administered by the American Chemical Society, and the Minnesota
Supercomputer Institute are thankfully acknowledged.
(25) Breslow, R.; Dowd, P. J. Am. Chem. Soc. 1963, 85, 2729-2735.
(26) Breslow, R.; Gal, P.; Chang, H. W.; Altman, L. J. J. Am. Chem. Soc.
1965, 87, 5139-5144.
(27) Breslow, R.; Cortes, D. A.; Jaun, B.; Mitchell, R. D. Tetrahedron Lett.
1982, 23, 795-798.
(28) Additional bands at 336, 353, 387, and 478 nm also are predicted,
but the intensities are relatively small, or the absorptions fall in a region where
they would be obscured.
JA991224O