Isotope effects and the mechanism of deoxygenation of epoxides with dichlorocarbene

Article abstract of DOI:10.1016/j.tetlet.2005.01.153

The deoxygenation of styrene oxide with dichlorocarbene is studied by a combination of isotope effects and theoretical calculations. A normal ¹³C isotope effect of 1.016 is observed for the α-carbon of the styrene oxide but a surprising inverse isotope effect of 0.995 is observed at the β-carbon. This is indicative of a highly asynchronous process in which the C_α-O bond is broken without any progress in the breakage of C_β-O bond. Theoretical calculations support this interpretation. This coarctate reaction is formally concerted as it avoids involving a high-energy intermediate, but it appears uninfluenced by transition state aromaticity.

Full text of DOI:10.1016/j.tetlet.2005.01.153

Tetrahedron
Letters
Tetrahedron Letters 46 (2005) 2033–2036
Isotope effects and the mechanism of deoxygenation of
epoxides with dichlorocarbene
Daniel A. Singleton* and Zhihong Wang
Department of Chemistry, Texas A&M University, PO Box 30012, College Station, TX 77842, USA
Received 19 September 2004; accepted 27 January 2005
Abstract—The deoxygenation of styrene oxide with dichlorocarbene is studied by a combination of isotope effects and theoretical
calculations. A normal 13C isotope effect of 1.016 is observed for the a-carbon of the styrene oxide but a surprising inverse isotope
effect of 0.995 is observed at the b-carbon. This is indicative of a highly asynchronous process in which the Ca–O bond is broken
without any progress in the breakage of Cb–O bond. Theoretical calculations support this interpretation. This coarctate reaction is
formally concerted as it avoids involving a high-energy intermediate, but it appears uninfluenced by transition state aromaticity.
Ó 2005 Published by Elsevier Ltd.
Coarctate reactions are those in which two bonds are
simultaneously formed and broken at an atom.1 A sur-
prisingly large number of reactions involve transforma-
tions that are formally coarctate processes—in fact
about 2% of organic reactions based on a computerized
search.2 A theory of coarctate transition states has been
described, based on a topological transformation of the
Dewar/Zimmerman concept of transition state
aromaticity.1
tivity of the product alkene.3–6 As a result, the yield of
alkene from these reactions is typically low. However,
the deoxygenation itself appears to be both facile and
clean. The reaction is stereospecific, retaining the cis or
trans stereochemistry of the original epoxide.5,6 Ylide
intermediates (i.e., 1) have often been proposed in these
reactions,3,5,6 but an attempt to observe such an ylide
was not successful.5
In our view the most chemically interesting question for
coarctate reactions is the degree to which concert is fa-
vored by transition state aromaticity. What portion of
the large number of formally coarctate reactions are in
fact concerted processes? If they are concerted, is there
a stabilization of synchronous bonding transformations,
or do the reactions involve highly asynchronous bond-
ing changes, with concertness only being a way of avoid-
ing high-energy intermediates? We have initiated an
experimental and computational examination of coarc-
tate reactions with the goal of addressing these ques-
tions. We describe here our results with one very
simple coarctate reaction—the deoxygenation of epox-
ides with carbenes.
Our goals in applying kinetic isotope effects (KIEs) to
study these reactions were to distinguish rate-limiting
formation of an ylide versus a rate-limiting deoxygen-
ation process, and to use the KIEs as a gauge of the
asynchronicity of the breaking of the two C–O bonds.
The latter goal requires the use of an unsymmetrically
substituted epoxide, as a symmetrical epoxide will lead
to symmetrical isotope effects due to averaging. The
deoxygenation of styrene oxide (2) with dichlorocarbene
was chosen for study.
The deoxygenation of epoxides with carbenes has often
been observed but it is complicated by the further reac-
Keywords: Isotope effects; Mechanism; Deoxygenation; Epoxide;
Coarctate.
*
Corresponding author. Tel.: +1 979 845 9166; fax: +1 979 845
4719; e-mail: singleton@mail.chem.tamu.edu
0040-4039/$ - see front matter Ó 2005 Published by Elsevier Ltd.
doi:10.1016/j.tetlet.2005.01.153

2034
D. A. Singleton, Z. Wang / Tetrahedron Letters 46 (2005) 2033–2036
Previous deoxygenation of 2 using phase-transfer/emul-
sifying conditions had been plagued by the side reaction
of basic hydrolysis of the epoxide.6 The use of solid
KOH, with no discrete aqueous phase, and 18-crown-6
as the phase transfer catalyst appears to alleviate this
problem. Under these conditions, no phenylethylene
glycol could be observed, and the formation of 3 was
nearly quantitative based on NMR analysis versus an
internal standard.
Theoretical calculations were used to interpret these
results in greater detail. The deoxygenation of 2 with
dichlorocarbene was studied in B3LYP calculations
employing a 6-311++G(2d,p) basis set. Three transition
structures were located, two with the carbene attacking
the epoxide oxygen anti to the phenyl group and one
syn to the phenyl group. The best of these structures
(4) is 5.5 kcal/mol (including zpe) above separate start-
ing materials. Structure 4 notably places the lone pair
of the carbene moiety anti to the breaking Ca–O bond.
This is favored by 4.3 kcal/mol over the second best
structure (given in Supplementary data) in which the
carbene lone pair is syn to the breaking Ca–O bond.
Similar results were obtained in B3LYP, mPW1PW91
and MP2 calculations employing smaller basis sets, as
described in Supplementary data.
The 13C KIEs for deoxygenation of 2 were studied by
NMR methodology at natural abundance.7 A total of
four reactions of 2 at 40 °C were taken to 72–82% con-
version, and the starting 2 was recovered by an aqueous
workup followed by flash chromatography and microdi-
stillation. The 13C NMR of the samples of recovered 2
was analyzed along with standard samples that had
not been subjected to the reaction conditions. The
change in isotopic composition in each position was
determined relative to the para aromatic carbon, with
the assumption that isotopic fractionation of this carbon
was negligible. From the percentage conversions and the
changes in isotopic composition, the KIEs were calcu-
lated as previously described.7
Figure 1a shows the average KIEs from the four inde-
pendent determinations. A substantial normal (>1) 13C
KIE was observed at the epoxide carbon adjacent to
the aromatic ring (Ca). In contrast, the KIE at the distal
carbon (Cb) is surprisingly inverse (<1). The KIEs for
the aromatic ring carbons are within experimental error
of unity.
If the deoxygenation involved rate-limiting formation
of an epoxide–carbene ylide complex, the Ca and
Cb C KIEs would be near unity. The observation of
As expected from the isotope effects, transition structure
4 is undergoing cleavage of the Ca–O bond but not the
Cb–O bond. To test the consistency of this transition
structure with the experimental isotope effects, the KIEs
for 4 were predicted from the scaled theoretical vibra-
tional frequencies8 using conventional transition state
theory by the method of Bigeleisen and Mayer.9 Tunnel-
ing corrections were applied using the one-dimensional
infinite parabolic barrier model.10 Such KIE predictions
including a one-dimensional tunneling correction have
proven highly accurate in reactions not involving hydro-
gen transfer, so long as the calculation accurately de-
picts the mechanism and transition state geometry.11
13
a substantial Ca KIE rules this out—an ylide may still
be involved but it is not kinetically important. The
standard qualitative interpretation of the Ca KIE is
that the Ca–O bond is being broken in the rate-limiting
step. However, the inverse Cb-KIE suggests that the
Cb–O bond has gotten stronger at the transition state
than it was in the starting material. Together, these re-
sults are indicative of a transition state that would be
best described as a ring-opening process rather than a
synchronous deoxygenation. This does not rule out a
formally concerted deoxygenation, but the two C–O
bonds are clearly breaking in different stages of the
process.
The results are shown in Figure 1b. It is apparent that
the predicted KIEs are quite similar in pattern to those
observed experimentally. This supports the qualitative
interpretation of the isotope effects and the approximate
accuracy of 4. Since some degree of charge buildup at
the transition state would be stabilized in solution, the
gas phase structure 4 cannot be expected to be perfectly
accurate. Considering this limitation of the calculations,
the agreement between predicted and experimental KIEs
is in fact very good.
Figure 1. (a) Experimental 13C KIEs (k12C/k13C) for the deoxygenation
of styrene oxide at 40 °C. Standard deviations in the last digit from
four determinations are shown in parentheses. (b) Predicted 13C KIEs
based on transition structure 4.
No stationary point corresponding to an ylide geometry
could be located in these calculations. A very loose
˚
epoxide–CCl2 complex with an O–C distance of 2.60 A
was found on the potential energy surface, but the en-

D. A. Singleton, Z. Wang / Tetrahedron Letters 46 (2005) 2033–2036
2035
ergy of this complex is only 1.6 kcal/mol below separate
reactants and its free energy is 7.2 kcal/mol above reac-
tants at standard state. Such a complex is unlikely to be
a minimum on the free-energy surface and is mechanis-
tically irrelevant.
breaking C–O bond. This arrangement is reminiscent of
the anomeric effect and may be viewed as maximizing
overlap between the lone pair and the r* orbital of
the breaking C–O bond, stabilizing the transition state.
An alternative transition structure in which the lone pair
is syn to the breaking bond (Supplementary data) was
9.5 kcal/mol higher in energy. The stabilizing effect of
the anti lone pair appears to enforce asynchronicity in
the reaction, as the lone pair can only aid in breaking
one of the two C–O bonds.
When the steepest–descent reaction path in mass-weight
coordinates is followed onward from 4, the Ca–O bond
˚
distance increases rapidly to >2 A while the Cb–O dis-
˚
tance at first stays below 1.5 A. However, there is ulti-
mately no barrier to dissociation of the products,
phosgene and styrene, which are downhill from 4 by
86.8 kcal/mol, so the Cb–O bond breaks without there
being an intermediate. The deoxygenation is thus overall
a formally concerted process.
The picture that emerges from these results may be
understood with reference to a More OÕFerrall–Jencks
diagram12 for the deoxygenation (Fig. 2). The transition
state is early and breakage of the two C–O bonds is
highly asynchronous. The reaction path starts out in a
direction that would lead to the intermediate at the adja-
cent corner of the diagram. However, some time after
the transition state, the reaction path avoids the high-en-
ergy intermediate (either a diradical or a zwitterion) by
curving toward the extremely stable products. We note
that if a synchronous coarctate transition state were sub-
stantially stabilized by transition state aromaticity, the
reaction path would be expected to pass in some degree
toward the middle of the More OÕFerrall–Jencks dia-
gram. Based on the isotope effects and the calculational
predictions, this is not the case.
Dewar has pointed out that there is an intrinsic prefer-
ence for reactions to avoid synchronous bonding
changes.13 This preference may be overcome by special
factors in transition states, most particularly transition
state aromaticity for pericyclic reactions. The degree to
which transition state aromaticity aids synchronicity in
coarctate reactions remains to be seen. For the deoxy-
genation of epoxides with carbenes, it appears that the
preference for asynchronous bonding changes wins
out. Among coarctate reactions, deoxygenation may
not be typical—its high exothermicity and resulting
early transition state make it arguably non-ideal for
observing any special stabilization of a synchronous
transition state—but for this coarctate reaction at least,
there is no apparent influence of transition state
aromaticity.
Would this highly asynchronous pathway apply to a
symmetrical epoxide? As described above, this question
cannot be addressed well experimentally with isotope ef-
fects. However, calculations suggest that the preference
for an asynchronous pathway is general. Transition
structure 5 was located for the deoxygenation of cis-
2,3-dimethyloxirane in B3LYP/6-311++G(2d,p) calcula-
tions. The barrier in this reaction is predicted to be 6.2,
0.7 kcal/mol higher than with 2. The transition structure
is somewhat later, but it is still very early. As was true in
4, 5 places the lone pair of the carbene moiety anti to the
Acknowledgements
We thank NIH grant # GM-45617 and The Robert A.
Welch Foundation for support of this research.
Supplementary data
Experimental procedures, NMR integration results, cal-
culational procedures, and energies and geometries of all
calculated structures. Supplementary data associated
with this article can be found, in the online version, at
doi:10.1016/j.tetlet.2005.01.153.
References and notes
Figure 2. More OÕFerrall–Jencks diagram for the deoxygenation of
styrene oxide with dichlorocarbene.
1. Herges, R. Angew. Chem., Int. Ed. Engl. 1994, 33, 255–
276.

2036
D. A. Singleton, Z. Wang / Tetrahedron Letters 46 (2005) 2033–2036
2. Herges, R. J. Chem. Comput. Sci. 1994, 34, 91–102.
9. (a) Bigeleisen, J.; Mayer, M. G. J. Chem. Phys. 1947, 15,
261–267; (b) Wolfsberg, M. Acc. Chem. Res. 1972, 5, 225–
233; (c) Bigeleisen, J. J. Chem. Phys. 1949, 17, 675–
678.
3. Nozaki, H.; Takaya, H.; Noyori, R. Tetrahedron 1966, 22,
3393–3401; Halton, B.; Russell, S. G. G. J. Org. Chem.
1991, 56, 5553–5556.
4. Wittig, G.; Schlosser, M. Tetrahedron 1962, 18, 1023–
1028; Mannfov, G. G.; Berdnikov, E. A.; Samuilov, Y. D.
Russ. J. Org. Chem. 2001, 37, 339–344.
5. Shields, C. J.; Schuster, G. B. Tetrahedron Lett. 1987, 28,
853–856.
6. Tabushi, I.; Kuroda, Y.; Yoshida, Z. Tetrahedron 1976,
32, 997–1000.
7. Singleton, D. A.; Thomas, A. A. J. Am. Chem. Soc. 1995,
117, 9357–9358.
8. The calculations used the program QUIVER Saunders,
M.; Laidig, K. E.; Wolfsberg, M. J. Am. Chem. Soc. 1989,
111, 8989–8994, Becke3LYP frequencies were scaled by
0.9614.
10. Bell, R. P. The Tunnel Effect in Chemistry; Chapman &
Hall: London, 1980; pp 60–63.
11. (a) Beno, B. R.; Houk, K. N.; Singleton, D. A. J. Am.
Chem. Soc. 1996, 118, 9984–9985; (b) Meyer, M. P.;
DelMonte, A. J.; Singleton, D. A. J. Am. Chem. Soc. 1999,
121, 10865–10874; (c) DelMonte, A. J.; Haller, J.; Houk,
K. N.; Sharpless, K. B.; Singleton, D. A.; Straßner, T.;
Thomas, A. A. J. Am. Chem. Soc. 1997, 119, 9907–9908;
(d) Singleton, D. A.; Merrigan, S. R.; Liu, J.; Houk, K. N.
J. Am. Chem. Soc. 1997, 119, 3385–3386.
12. More OÕFerrall, R. A. J. Chem. Soc. B 1970, 274; Jencks,
W. P. Chem. Rev. 1972, 72, 705.
13. Dewar, M. J. S. J. Am. Chem. Soc. 1984, 106, 209–219.