An experimental and computational study of 1,2-hydrogen migrations in 2-hydroxycyclopentylidene and its conjugate base

Article abstract of DOI:10.1021/jo001038x

Thermal decomposition of α-hydroxydiazirine 2 gives primarily cyclopentanone and some allylic alcohol, in similar amounts as the known cyclohexyl analogue 1. Calculations (B3LYP/6-31+G*) also show cyclopentanone to be the major product of this carbene rearrangement. Diazirine 2 and the lithium salt of the corresponding conjugate base 3 were decomposed by photolysis. The proportion of ketone formed increases with deprotonation, a trend also found computationally. In comparison, the base-induced isomerization of cyclopentene oxide, which proceeds via α-elimination to a carbenoid intermediate similar to that obtained from 3, yields primarily allylic alcohol rather than ketone; neither ring size nor charge thus accounts for the unusual product distribution observed. Interestingly, the calculations reveal that in the gas phase with no counterion, the singlet, oxyanionic carbene, and the α-deprotonated epoxide are the same, rather than discrete structures. This intramolecular complexation stablilizes the oxyanionic carbene by 20-25 kcal/mol.

Full text of DOI:10.1021/jo001038x

1
600
J . Org. Chem. 2001, 66, 1600-1606
An Exp er im en ta l a n d Com p u ta tion a l Stu d y of 1,2-Hyd r ogen
Migr a tion s in 2-Hyd r oxycyclop en tylid en e a n d Its Con ju ga te Ba se
Kathleen M. Morgan,* Matthew J . O’Connor, J onathan L. Humphrey, and
Kerry E. Buschman
Department of Chemistry, The College of William & Mary, P.O. Box 8795,
Williamsburg, Virginia 23187-8795
kmmorg@wm.edu
Received J uly 10, 2000 (Revised Manuscript Received December 27, 2000)
Thermal decomposition of R-hydroxydiazirine 2 gives primarily cyclopentanone and some allylic
alcohol, in similar amounts as the known cyclohexyl analogue 1. Calculations (B3LYP/6-31+G*)
also show cyclopentanone to be the major product of this carbene rearrangement. Diazirine 2 and
the lithium salt of the corresponding conjugate base 3 were decomposed by photolysis. The proportion
of ketone formed increases with deprotonation, a trend also found computationally. In comparison,
the base-induced isomerization of cyclopentene oxide, which proceeds via R-elimination to a
carbenoid intermediate similar to that obtained from 3, yields primarily allylic alcohol rather than
ketone; neither ring size nor charge thus accounts for the unusual product distribution observed.
Interestingly, the calculations reveal that in the gas phase with no counterion, the singlet, oxyanionic
carbene, and the R-deprotonated epoxide are the same, rather than discrete structures. This
intramolecular complexation stablilizes the oxyanionic carbene by 20-25 kcal/mol.
In tr od u ction
The most common reaction of singlet alkyl and dialkyl
Sch em e 1
carbenes is 1,2-hydrogen migration to form an alkene.1-3
Understanding the fundamental features that influence
the selectivity of this reaction has been an ongoing goal
of carbene chemists. Electron-donating bystander sub-
stituents B have been shown to increase the migratory
aptitude of adjacent groups (eq 1).1,4,5 These substituents
are thought to act by affecting the bond strength to the
migrating hydrogen through stabilization of the partial
positive charge which forms on carbon as the hydrogen
migrates. Other factors, such as stereoelectronic effects
and mode of carbene generation, are also known to
influence carbene reactivity.
Sch em e 2
The goal of this investigation is to better understand
the unexpected regioselectivity of the carbene/oid inter-
mediate implicated in the base-induced rearrangement
of cyclopentene oxide to 2-cyclopentenol and cyclopen-
tanone (Scheme 1). Deuterium-labeling studies show that
the allylic alcohol is formed via R-elimination in nonpolar
solvents, rather than â-elimination (Scheme 2).6 The
preference for R-elimination over â-elimination in cyclo-
pentene oxide is believed to arise from the unusual
acidity of the epoxide R-proton due to increased angle
strain, not because of poor orbital overlap for the compet-
ing syn-E2 reaction; cyclopentene oxide reacts faster via
the carbene/oid mechanism than does cyclohexene oxide
via syn-E2. Note that the addition of diisopropylamine
to the epoxide or the carbene is not observed under these
reaction conditions.
7
*
To whom correspondence should be addressed. Fax: (757) 221-
715.
1) Moss, R. A. In Advances in Carbene Chemistry; Brinker, U. H.,
Ed.; J AI Press Inc.: Greenwich, CT, 1994; Vol. 1, pp 59-88.
2) Platz, M. S. In Advances in Carbene Chemistry; Brinker, U. H.,
Ed.; J AI Press Inc.: Stamford, CT, 1998; Vol. 2, pp 133-174.
3) Kirmse, W. Carbene Chemistry, 2nd ed.; Academic Press: New
York, 1971.
2
(
(
The intermediacy of carbenoids in base-promoted
isomerizations of some cyclic and bicyclic epoxides in
(
(
4) Nickon, A. Acc. Chem. Res. 1993, 26, 84-89.
(6) Morgan, K. M.; Gajewski, J . J . J . Org. Chem. 1996, 61, 820-
821.
(5) Keating, A. E.; Garcia-Garibay, M. A.; Houk, K. N. J . Phys.
Chem. A 1998, 102, 8467-8476.
(7) Morgan, K. M.; Gronert, S. J . Org. Chem. 2000, 65, 1461-1466.
1
0.1021/jo001038x CCC: $20.00 © 2001 American Chemical Society
Published on Web 01/31/2001

1
,2-Hydrogen Migrations in 2-Hydroxycyclopentylidene
J . Org. Chem., Vol. 66, No. 5, 2001 1601
Sch em e 3
Sch em e 6
Sch em e 7
Sch em e 4
Sch em e 5
lize an adjacent carbocation. The results presented show
that neither ring size nor charge favor the formation of
allylic alcohol and that the adjacent anionic oxygen
stabilizes the carbene by 20-25 kcal/mol in the gas
phase.
Resu lts a n d Discu ssion
Syn th esis of Dia zir in es. As shown in Scheme 7,
diazirine 2 was synthesized from 2-hydroxycyclopen-
tanone, which was produced via the acyloin condensation
of diethylglutarate.13 Following Schmitz’s procedures,
the ketone was treated with hydroxylamine-O-sulfonic
acid (HOSA) in aqueous ammonia to give the diaziridine,
which was oxidized with silver oxide. Diazirine 3 was
produced by treating diazirine 2 with excess lithium
hydride in THF.
12
nonpolar solvents is well established, as seen through
deuterium-labeling studies and the formation of cross-
ring insertion products (Scheme 3).8,9 Carbene-like reac-
tivity is also observed in an R-lithiated cyclopentene oxide
having a tethered alkene, in which a cyclopropane ring
is formed (Scheme 4).10 In several cases, the initial
R-deprotonation was found to be reversible.6
Im p or ta n ce of Rin g Size: Decom p osition of Di-
a zir in e 2. The decomposition of 2 was first performed
under the same conditions used by Schmitz for 1.
,8
The R-elimination of cyclopentene oxide with LDA in
nonpolar solvents gives mostly the allylic alcohol and 15-
-
3
Thermal decomposition of a 1 × 10 M solution of 2 in
nitrobenzene at 200 °C, followed by GC analysis, gave
only cyclopentanone and 2-cyclopentenol, in a 2:1 ratio
(Table 1). These two products and cyclopentene oxide
were found to be stable to the thermal reaction condi-
tions. Schmitz observed a slightly greater amount of
2
5% of cyclopentanone (Scheme 1). It is not unusual for
carbenes to rearrange with low selectivity. However, it
is counterintuitive that the ketone is the minor product
because the oxygen bystander group should strongly
stabilize the adjacent positive charge during 1,2-hydrogen
migration to enolate.11 A methoxy bystander substituent
is known to promote 1,2-hydrogen migration by a factor
of 73.5 as compared to H.4 In addition, the product
distribution observed in this rearrangement of cyclopen-
tene oxide differs significantly from a similar carbene 1
previously studied by Schmitz, generated thermally from
a diazirine precursor (Scheme 5).12
12
ketone from 1 (Scheme 5). The greatest difference
between the cyclopentyl and cyclohexyl systems is that
no epoxide is produced from the cyclopentyl carbene. This
may be explained by the greater strain in cyclopentene
oxide compared to cyclohexene oxide.1
4,15
The preference
for 1,2-hydrogen migration rather than insertion into a
16,17
tethered O-H bond has been observed previously.
The three main differences between the Schmitz car-
bene and that generated from cyclopentene oxide are ring
size, charge, and carbene/oid precursor. To explore the
first two variables, the reactions of diazirines 2 and 3
The product distribution is not highly dependent on
temperature; almost identical results are obtained upon
heating 2 in nitrobenzene at 200 °C for 15 min and at
150 °C for 1 h. The thermal decomposition of 2 was also
performed in triglyme, with similar results. The com-
plexation of the ether oxygen with the carbene to form
an ylide is either slower than 1,2-hydrogen migration or
does not greatly influence the products observed.
(Scheme 6) are described. Product distributions are
compared to 1 and to the base-promoted rearrangement
of cyclopentene oxide. Density functional theory calcula-
tions were performed on the carbenes, their rearrange-
ment products, and the transition states to rearrange-
ment. In addition, model compounds were calculated to
assess the extent to which various oxygen groups stabi-
(13) Bloomfield, J . J .; Nelke, J . M. Organic Syntheses; Wiley: New
York, 1988; Collect. Vol. VI.
(14) Slayden, S. W.; Liebman, J . F. In Supplement E2: The
Chemistry of Hydroxyl, Ether and Peroxide Groups; Patai, S., Ed.; J ohn
Wiley & Sons: New York, 1993.
(
(
(
8) Crandall, J . K.; Apparu, M. Org. React. 1983, 29, 345-443.
9) Satoh, T. Chem. Rev. 1996, 96, 3303-3325.
(15) Kozina, M. P.; Timofeeva, L. P.; Luk’yanova, V. A.; Pimenova,
S. M.; Kas'yan, L. I. Russ. J . Phys. Chem. 1988, 62, 609-612.
(16) Stevens, I. D. R.; Liu, M. T. H.; Soundararajan, N.; Paike, N.
Tetrahedron Lett. 1989, 30, 481-484.
(17) Stevens, I. D. R.; Liu, M. T. H.; Soundararajan, N.; Paike, N.
J . Chem. Soc., Perkin Trans. 2 1990, 661-667.
10) Doris, E.; Dechoux, L.; Mioskowski, C. J . Am. Chem. Soc. 1995,
1
17, 12700-12704.
(
(
11) Taylor, K. G. Tetrahedron 1982, 38, 2751-2772.
12) Schmitz, E.; Stark, A.; H o¨ rig, C. Chem. Ber. 1965, 98, 2509-
2
515.

1
602 J . Org. Chem., Vol. 66, No. 5, 2001
Ta ble 1. P r od u cts Obta in ed fr om Decom p osition of Dia zir in es 2 a n d 3
Morgan et al.
compd
conditions
cyclopentanone (%)
2-cyclopentenol (%)
cyclopentene oxide (%)
2
2
2
2
2
3
NO2Ph, 200 °C
NO2Ph, 150 °C
triglyme, 200 °C
THF, hν
66.4 ( 4.6
67.8 ( 1.0
64.2 ( 0.8
53.4 ( 1.2
56.7
33.6 ( 4.6
32.2 ( 1.4
35.8 ( 0.8
28.1 ( 1.5
26.1
0
0
0
18.5 ( 1.1
17.2
7.9 ( 0.3
, correcteda
THF, hν
THF, hν
69.2 ( 3.3
22.9 ( 3.6
a
Data are corrected to reflect the 14% degradation of cyclopentanone in THF upon photolysis.
Sch em e 8
Diazirine 2 in THF was also decomposed photochemi-
cally. The ratio of ketone to allylic alcohol remains 2:1.
The most significant difference is that nearly 20% cyclo-
pentene oxide is formed by photolysis. One possible
mechanism for the formation of epoxide is shown in
Scheme 8. Upon photolysis, the diazirine isomerizes to
the diazo compound. A rapid intramolecular proton
transfer generates a diazonium ion which could easily
close to the epoxide.18 Intermolecular proton transfer
could also give rise to a diazonium ion, but due to the
low concentration of diazirine, this option is less likely.
Recent studies suggest that some rearrangements thought
to occur from a carbene might instead be occurring from
an excited state of the diazirine or from the isomeric diazo
compound.2 This might also explain the formation of
epoxide on photolysis but not thermolysis.
,19
The photostability of the products was tested on a
sample with known composition. After 1 h of photolysis,
which is the time it took to decompose the diazirine, the
amounts of epoxide and allylic alcohol did not change,
F igu r e 1. Optimized geometries for singlet and triplet 4,
TS1-TS3, and 4-ylide, B3LYP/6-31+G*. Selected bond lengths
are given in Å.
but the ketone decreased by 14%, possibly by forming a
THF adduct.20 A correction for this decomposition of
cyclopentanone was made by increasing the relative
amount of ketone by 14%. As seen in Table 1, these
corrected values are not substantially different from the
uncorrected data.
A computational study (B3LYP/6-31+G*//B3LYP/6-
1+G*) of 2-hydroxycyclopentylidene 4 reveals that there
3
are two conformers of the singlet carbene, one having
H-O-C-Ccarbene syn, the other anti. The anti conformer
is less stable by 4 kcal/mol, possibly due to greater steric
interference between the hydroxyl group and the meth-
ylene hydrogens. The syn conformer of triplet-4 is again
more stable than the anti conformer, but only by 1 kcal/
mol (UB3LYP; S**2 is 2.0058). The carbene has a singlet
ground state, with a singlet/triplet energy gap of 10.4
kcal/mol. This is similar to cyclopentylidene itself, which
is also calculated to have a singlet ground state and a
singlet/triplet gap of 8.3 kcal/mol.
F igu r e 2. Relative energies (∆H298) of 4, TS1-TS3, and
products, including zero-point vibrational energy and thermal
corrections, B3LYP/6-31+G*//B3LYP/6-31+G*.
The C-Ccarbene-C bond angles in the syn conformers
of singlet-4 and triplet-4 are 102.8° and 116.5°, respec-
tively. Selected bond lengths for the syn conformers are
given in Figure 1. Intramolecular stabilization of singlet-
empty p-orbital to form 4-ylid e, was found to cost 4.4
kcal/mol. Structural parameters for 4-ylid e, an energy
minimum, are also given in Figure 1.
4
, via donation of an oxygen lone pair into the carbene
Calculations on the 1,2-hydrogen migration reactions
for the lower energy syn conformer of singlet-4 show enol,
hence ketone, to be the primary product. Figure 2
presents relative enthalpies at 298 K. The transition
state to generate the enol (TS1) was calculated to be of
similar energy as singlet-4. The difference in energy, ∆E,
(
18) Kirmse, W. In Advances in Carbene Chemistry; Brinker, U. H.,
Ed.; J AI Press: Greenwich, CT, 1994; Vol. 1, pp 1-57.
19) Bonneau, R.; Liu, M. T. H. In Advances in Carbene Chemsitry;
Brinker, U. H., Ed.; J AI Press Inc.: Stamford, CT, 1998; Vol. 2, pp
-28.
20) Elad, D. In The Chemistry of the Ether Linkage; Patai, S., Ed.;
Interscience Publishers: New York, 1967; pp 353-372.
(
1
(

1
,2-Hydrogen Migrations in 2-Hydroxycyclopentylidene
J . Org. Chem., Vol. 66, No. 5, 2001 1603
Sch em e 9
Sch em e 10
Ta ble 2. Ca lcu la ted ∆H298 for Isod esm ic Rea ction
shows the carbene to be more stable by 0.4 kcal/mol. With
the inclusion of zero-point vibrational energies, TS1
drops to 0.1 kcal/mol below the carbene, and with the
thermal energy correction, TS1 is 0.4 kcal/mol lower in
energy than the carbene.
(k ca l/m ol)
M
∆H298
Li
Na
K
-51.8
-68.0
-77.8
Two pathways to allylic alcohol, one via migration of
the H syn to the alcohol (TS2), and one anti (TS3), are
both higher in energy relative to singlet-4 by 6.0 and 8.8
kcal/mol, respectively. The optimized structures of TS1-
TS3 are shown in Figure 1. No transition state was found
for the concerted formation of cyclopentene oxide at the
B3LYP/6-31+G* level of theory, but AM1 calculations
found this transition state to be 42.5 kcal/mol above
singlet-4. For comparison, B3LYP/6-31+G* calculations
find that the transition state for 1,2-hydrogen migration
from cyclopentylidene to cyclopentene is 5.9 kcal/mol
higher in energy than the carbene, suggesting that in 4
the oxygen substituent does facilitate 1,2-hydrogen mi-
gration substantially.
In TS1, the C-H bond length of the migrating hydro-
gen has elongated to 1.19 Å from 1.14 Å, and the distance
from migrating hydrogen to the carbene carbon shrinks
to 1.59 Å from 1.84 Å. The Henol-C-Ccarbene bond angle
has compressed to 73.9° from 88.9° in 4. Note also that
the C-O bond length in TS1 has shortened to 1.38 Å,
suggesting that the oxygen is stabilizing the adjacent
carbon via electron donation. TS2 and TS3 occur later
on the reaction coordinate. In TS2, the migrating syn
hydrogen is nearly equidistant from the migration origin,
none
-178.9
ate, rather it closed directly to cyclopentene oxide. Since
formation of the epoxide is not observed in the thermal
decomposition of 2, it is unlikely that this proton-transfer
mechanism occurs in preference to carbene chemistry.
Im p or ta n ce of Ch a r ge: Decom p osition of Dia zi-
r in e 3. Given that nitrobenzene reacts explosively when
heated with alkali,21 the conditions used by Schmitz were
not considered for the thermolysis of 3. Thermal decom-
position of 3 in triglyme was attempted, but the experi-
ment failed due to the rapid reaction of the base with
solvent at high temperatures. A solution of diazirine 3
in THF was successfully photolyzed. As seen in Table 1,
a greater percentage of ketone is formed, which is
consistent with the oxyanion being a stronger electron-
donating bystander group compared to the hydroxyl
group. However, the ratio of ketone to allylic alcohol
formed from 3 is 3:1, not dramatically different than the
2
:1 ratio observed from 2. Interestingly, a small amount
of epoxide is produced.
We had hoped to study the importance of counterion
on the decomposition of 3. Unfortunately, the corre-
sponding sodium and potassium salts did not give useful
data, rather they produced a dark solution within min-
utes at room temperature. Instead, one effect of the
counterion was calculated using the isodesmic reaction
shown in Scheme 10. This equation is designed to show
how the C-H bond migration may be affected by the
environment at oxygen. A C-H bond next to a stronger
electron donor would be expected to migrate more readily,
giving rise to a greater amount of enol(ate). An exother-
mic isodesmic reaction indicates that the oxyanion sta-
bilizes the adjacent positive charge to a greater extent
than the alcohol would stabilize the charge. The calcu-
lated reaction energies are all exothermic, with ∆H298
1
C
9
.29 Å, and the carbene carbon, 1.32 Å. The Hsyn-C-
carbene bond angle is 58.6° in TS2, quite different from
8.0° in 4. The hydrogen migration in TS3 is not quite
as advanced as in TS2.
The importance of including polarization functions on
hydrogen was also investigated for TS1-TS3. Activation
enthalpies calculated at the B3LYP/6-31+G** level of
theory differed by less than 1 kcal/mol compared to
energies obtained at the lower level of theory (Supporting
Information, Table S-2). Note that diffuse functions were
included in all calculations on the neutral carbene
surface, for later comparison with the anionic carbene.
Transition states for 1,2-hydrogen migrations from
anti-4 were also calculated. These alternative transition
state conformers TS1a n ti, TS2a n ti and TS3a n ti were
calculated to be 0.3, 5.5, and 8.7 kcal/mol higher in energy
than anti-4. The structures of the anti carbenes, anti
transition states and anti alcohol products are included
in the Supporting Information.
(
B3LYP/6-31+G*) ranging from -51.8 kcal/mol for M )
Li to -178.9 kcal/mol for the bare anion (Table 2).
These calculations parallel the trend observed experi-
mentally, that the lithium salt forms enolate more than
alcohol 4 forms enol. However, the magnitudes are quite
different. TS1 and TS4, the transition state to enolate
formation, are both early; thus, the large substituent
effects found in the isodesmic reaction should be expected
to be smaller in the carbene rearrangements. In addition,
the isodesmic calculations were conducted in the gas
phase, and in solution the differences will be greatly
attenuated.
It is possible that rearrangement of singlet-4 might
occur instead through a carbocation intermediate, formed
via proton transfer from the alcohol to the carbene
(
Scheme 9).18 Deuterium labeling experiments would not
distinguish these pathways. Thus, the cationic mecha-
nism was investigated computationally. No stable struc-
ture was found for the oxyanion/carbocation intermedi-
(
21) Lewis, R. J ., Sr. Hazardous Chemicals Desk Reference, 4th ed.;
J ohn Wiley & Sons: New York, 1997.

1
604 J . Org. Chem., Vol. 66, No. 5, 2001
Morgan et al.
singlet carbenes and are due to intramolecular stabiliza-
tion. The geometry of singlet-5 is a hybrid of the oxiranyl
anion and the oxyanionic carbene (eq 2). Especially in
the gas phase, the singlet carbene will obtain significant
stabilization through the interaction of the vacant p-
orbital and the oxyanion. The most relevant structural
features are shown in Figure 3. The C-O single bond in
singlet-5 is 1.42 Å and the C-O partial bond is 1.58 Å,
compared to C-O bonds of 1.44 Å in cyclopentene oxide.
The C-O partial bond in singlet-5 must break in order
for the carbene p-orbital to be available for 1,2-hydrogen
migrations; this cost is reflected in the greater than
normal activation energies. Also, the O-C-Ccarbene bond
angle in singlet-5 is 65.9°, greater than the 59.4° O-C-C
bond angle in cyclopentene oxide. Note that the triplet
oxyanionic carbene 5, which cannot benefit from such
intramolecular stabilization, is very similar in structure
to the triplet hydroxycarbene 4. Triplet-5 lies 32.2 kcal/
mol higher in energy than singlet-5. Also, as noted above,
all the transition states to 1,2-hydrogen migration are
at least 23 kcal/mol above singlet-5, which suggests that
the gas-phase oxyanionic carbene is stabilized by 20-25
kcal/mol from this intramolecular complexation. Another
measure of the unusual stability of singlet-5 was obtained
by calculating the energy of the oxyanionic carbene at
the geometry of alcohol 4; this hypothetical structurally
frozen 4-a n ion was calculated to be 25.0 kcal/mol greater
than that of singlet-5, consistent with the stabilization
energy discussed above.
F igu r e 3. Optimized geometries for singlet and triplet 5 and
TS4-TS6, B3LYP/6-31+G*. Selected bond lengths are given
in Å.
Structurally, TS4-TS6 are fairly similar to TS1-TS3.
In all cases, the oxygen has moved away from the carbene
carbon, and these nonbonded distances are 2.38 Å in TS4,
2
.51 Å in TS5, and 2.28 Å in TS6. The O-C-Ccarbene bond
angles have also opened up, the smallest being 103.2° in
TS6. TS4 has the earliest transition state, with the C-H
bond elongating only 0.05 Å.
One should be concerned that the oxyanionic carbene
might behave like an anion, thus 1,2-hydrogen migra-
tions would be forbidden. The transition states to the 1,2-
hydrogen migrations of 5 were obtained carefully, start-
ing from deprotonated versions of TS1-3. Each transition
state was connected to reactant and product by displacing
the transition state geometry by 10% along the imaginary
vibrational frequency, both toward the reactant and
F igu r e 4. Relative energies (∆H298) of 5, TS4-TS6, and
products, including zero-point vibrational energy and thermal
corrections, B3LYP/6-31+G*//B3LYP/6-31+G*.
The rearrangement of the oxyanionic carbene singlet-5
was also calculated. The bare oxyanion was studied to
maximize the effects of deprotonation on the reaction
profile. The optimized structures are shown in Figure 3
and relative enthalpies at 298 K are shown in Figure 4.
The formation of enolate (TS4) is again most favorable,
22
toward the product. The resulting structure was then
carefully reoptimized, and in all cases the carbene/
oxiranyl anion hybrid structure 5 was obtained as the
reactant.
2
3.2 kcal/mol above singlet-5. Two pathways to the allylic
alcoholate were also found, with the migration of the
hydrogen syn to the oxygen (TS5) and anti (TS6) nearly
isoenergetic, +31.1 and 30.3 kcal/mol, respectively. These
results also predict that enolate formation from singlet-5
will be more favorable than enol formation from singlet-
To further address this concern, atomic charges were
calculated using Reed and Weinhold’s natural population
23,24
analysis.
Charges for five structures are shown in
Figure 5: singlet 2-hydroxycyclopentylidene 4, 4-ylid e,
cyclopentene oxide, singlet oxyanionic carbene 5, and
4
. TS4 is 7.1 kcal/mol lower in energy than the best
transition state to form the allylic alcoholate, while TS1
is 6.4 kcal/mol more favorable than the lower-energy
transition state to allylic alcohol.
The activation energies calculated for singlet-5 are
extraordinarily large for 1,2-hydrogen migrations of
(
22) We thank Professor C. M. Hadad, The Ohio State University,
for this program.
23) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J . Chem. Phys. 1985,
3, 735-746.
24) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88,
899-926.
(
8
(

1
,2-Hydrogen Migrations in 2-Hydroxycyclopentylidene
J . Org. Chem., Vol. 66, No. 5, 2001 1605
of allylic alcohol in 2-hydroxycyclopentylidene 4. Density
functional theory calculations support these results. The
calculations reveal that the gas-phase singlet oxyanionic
carbene 5 is the same as the oxiranyl anion. The
oxyanionic carbene is calculated to be stabilized by at
least 20 kcal/mol due to the nearby oxygen donor, and
undergoes 1,2-hydrogen migration with an activation
energy of greater than 23 kcal/mol. It appears that the
way in which the oxyanionic carbene/oid reactive inter-
mediate is generated is important in determining the
products obtained.
Exp er im en ta l Section
Gen er a l Meth od s. Unless otherwise specified, reagents
were purchased from Aldrich. Dichloromethane (Fisher) was
shaken with aqueous NaHCO
distilled from sodium benzophenone ketyl. Chloroform-d was
distilled from K CO . NMR spectra were obtained on a Varian
3
before use. THF (Fisher) was
F igu r e 5. NPA atomic charges, B3LYP/6-31+G*.
-a n ion . The charge on the carbene carbon is nearly zero
2
3
4
Mercury 400. TLC was performed on silica HLF plates
purchased from Analtech, and visualization was accomplished
using vanillin stain. UV/VIS spectra were obtained on a
Beckman DE-90 spectrometer using airtight quartz cuvettes.
GC was performed on a Hewlett-Packard 5890 with FID,
interfaced to a Hewlett-Packard 3396A integrator, with an
HP-1 cross-linked methyl siloxane column, 50 m × 0.2 mm ×
for all structures except 5, which has a charge of -0.22.
These charges suggest that the electron density in 5 is
unusual for a carbene, but certainly not that of a true
carbanion. The oxygen charges are also interesting. In
neutral molecules 4 and 4-ylid e, the oxygen charges are
nearly the same, -0.75 and -0.72. The oxygen charges
in 4 (-0.75) and 4-a n ion (-0.93) are not all that different
despite the formal charge on oxygen in the latter. In 5,
the oxygen charge is -0.68. Thus, in the hybrid structure
of 5, about 0.25 electrons have been donated from the
oxygen to the carbene carbon; electronically, this com-
pound is more like a carbene than a carbanion.
0
.33 µm film thickness.
3
,3-[r-Hyd r oxytetr a m eth ylen e]d ia zir in e 2.12 2-Hydroxy-
cyclopentanone (4 g, 0.04 mol), prepared via the acyloin
condensation of diethylglutarate,13 was dissolved in 65 mL of
4
methanol and 25 mL of 15 N NH OH. Hydroxylamine-O-
sulfonic acid (90% pure, 8 g, 0.06 mol) was added in portions
over 30 min at -20 to -15 °C. The reaction mixture was
warmed to room temperature and was stirred for 3 h. The
white solid was removed by filtration and the solids were
rinsed with methanol. The combined methanol solutions were
carefully concentrated by rotary evaporation.
The idea that the carbene is the same as the oxiranyl
anion, rather than the product of ring opening, appears
to be new. In fact, the nucleophilic behavior of oxiranyl
anions at -70 °C or below9 appears to contradict this
hypothesis. The explanation in the literature is that ring
opening of the unstable oxiranyl anion to the carbene
does not occur at low temperature. However, the tem-
perature dependence on reactivity could also be explained
by a smaller free energy of activation for the bimolecular,
nucleophilic reaction at low temperature, while the
unimolecular carbene reactions, found computationally
to have unusually large barriers in the gas phase, become
more favorable at higher temperatures. This type of
behavior is analogous to the ambiphilic carbenes which
react faster with alkenes having either electron with-
drawing or electron donating substituents, rather than
simple alkenes.2 The substituents on ambiphilic car-
benes such as chloromethoxycarbene stabilize the car-
bene through resonance, in the same way that 5 is
stabilized by the oxyanion. Note also that in solution the
intramolecular stabilization of the oxyanionic carbene
will likely be diminished; thus, it is possible that both
the calculations and previous experimental conclusions
are correct.
,25
The crude diaziridine (0.03 mol assumed) was dissolved in
50 mL of deionized water. A solution of AgNO (10.2 g, 0.06
3
mol) in 20 mL water was added, then 30 mL of 2 N NaOH
was added dropwise. A heavy black precipitate formed im-
mediately. After 30 min, the reaction mixture was analyzed
by TLC, and no diaziridine was observed. The reaction mixture
was filtered (note: Celite was found to decompose the diazirine),
and the solid residue was rinsed with water and then ether.
The aqueous solution was extracted with 3 × 50 mL ether,
and the combined organic extracts were dried over Na SO .
2
4
Diazirine 2 was purified by column chromatography (1:1
dichloromethane/pentane on silica) and was stored as a dilute
solution. Decomposition of 2 to give cyclopentanone occurs on
exposure to acid, and the absence of cyclopentanone in the
sample was verified by UV, NMR, and/or TLC (CH
eluent) prior to decomposition studies. To perform a solvent
exchange, an aliquot of the diazirine in CH Cl /pentane was
2 2
Cl as
6
2
2
combined with the solvent of interest, and the original solvents
were removed preferentially by rotary evaporation. The diazi-
rine was not evaporated to dryness due to the potential for
explosion.
1H NMR (NO
Ph-d
.17 (m, 1H), 1.95 (m, 2H), 1.74 (m, 1H), 1.34 (m, 2H). C NMR
CDCl ): δ ppm 73.6, 37.3, 35.5, 29.4, 21.8. UV (λmax) 337 nm
THF).
): δ ppm 3.72 (m, 1H), 2.46 (br, 1H),
2
5
13
2
(
(
3
Con clu sion s
3,3-[r-Hyd r oxytetr a m eth ylen e]d ia zir in e, Lith iu m Sa lt
. Excess LiH power was added to a solution of diazirine 2 in
3
The goal of this investigation was to discover why the
base-induced rearrangement of cyclopentene oxide, via
a carbene/oid intermediate, gave primarily allylic alcohol
rather than ketone. The experimental studies show that
neither ring size nor deprotonation favor the formation
dry THF under nitrogen, and the suspension was stirred at
room temperature for 30 min. After the solids settled, the THF
solution was removed by syringe and transferred to an airtight
quartz cuvette. Although a small amount of LiH was trans-
ferred to the cuvette with the THF solution, a clean UV
spectrum was still obtained.
Th er m a l Decom p osition of Dia zir in e 2. An NMR tube
(
25) Satoh, T.; Kobayashi, S.; Nakanishi, S.; Horiguchi, K.; Irisa, S.
Tetrahedron 1999, 55, 2515-2528.
26) Moss, R. A. Acc. Chem. Res. 1980, 13, 58-64.
containing a solution of diazirine 2 in nitrobenzene-d
5
was
(
placed into a preheated, stirred oil bath. Samples were

1
606 J . Org. Chem., Vol. 66, No. 5, 2001
Morgan et al.
analyzed by NMR before GC analysis to ensure that no
diazirine remained. At 200 °C, decomposition was complete
in 15 min, and at 150 °C, the reaction took 60 min. The bath
temperature was not rigorously controlled, ( 3 °C, due to the
observed temperature independence on the product ratios. The
stability of the primary reaction products was ascertained by
heating a mixture containing a known amount of cyclopen-
tanone and cyclopentene oxide, or cyclopentanone and the
allylic alcohol, for 15 min at 200 °C.
to the transition state; if the minimization took a step uphill
in energy, the calculation was stopped, then restarted using
the last structure in the continuous descent. Triplet carbenes
were calculated at the unrestricted level of theory. Atomic
charges were obtained by the natural population analysis
method,2
3,24
using the wave function obtained at the B3LYP/
6-31+G* level of theory. Some preliminary calculations includ-
ing conformational analyses were performed at the RHF/3-
21G level of theory using PC Spartan Plus.32
P h otolysis of Dia zir in es. A THF solution of diazirine,
-
3
approximately 1 × 10 M, was irradiated in a quartz cuvette
at room temperature using a 450 W Hanovia lamp, and
reaction progress was monitored by UV. The reaction was
complete in an hour. The photostability of the products was
also studied by photolyzing a sample of known proportions for
Ack n ow led gm en t. We thank the J effress Memorial
Trust and The College of William & Mary for support
of this work. Gaussian 94 was purchased with an award
from Research Corp. K.M.M. thanks Professor Chris-
topher Hadad, Ohio State, for helpful discussions.
1
h. The amounts of epoxide and allylic alcohol did not change,
but 14% of the ketone was destroyed.
P r od u ct An a lysis. Each reaction was repeated at least
three times, except the thermal decomposition of 2 in ni-
trobenzene at 150 °C, which was repeated twice. The uncer-
tainty intervals reported in Table 1 are twice the standard
deviation from the mean. The product ratios were determined
by gas chromatography. The identity of each product peak was
ascertained by spiking with an authentic sample. Using
solutions containing known amounts of ketone, epoxide and
allylic alcohol, GC response factors were found to be 1:1:1.
Solutions containing the lithium salts were quenched with
Su p p or tin g In for m a tion Ava ila ble: Cartesian coordi-
nates and energies of all computed structures; selected bond
angles (B3LYP/6-31+G*) (Table S-1); comparison of B3LYP/
1
13
6
-31+G* and B3LYP/6-31+G** results (Table S-2); H and
C
NMR spectra of diazirine 2. This information is available free
of charge via the Internet at http://pubs.acs.org.
J O001038X
(27) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
water and dried over Na
2 4
SO prior to GC analysis.
J ohnson, B. G.; Robb, M. A.; Cheeseman, J . R.; Keith, T.; Petersson,
G. A.; Montgomery, J . A.; Raghavachari, K.; Al-Laham, M. A.;
Zakrzewski, V. G.; Ortiz, J . V.; Foresman, J . B.; Cioslowski, J .;
Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala,
P. Y.; Chen, W.; Wong, M. W.; Andres, J . L.; Replogle, E. S.; Gomperts,
R.; Martin, R. L.; Fox, D. J .; Binkley, J . S.; Defrees, D. J .; Baker, J .;
Stewart, J . P.; Head-Gordon, M.; Gonzalez, C.; Pople, J . A. Revision
D.4 ed., Gaussian, Inc., Pittsburgh, PA, 1995.
Com p u ta tion a l Meth od s. The calculations were per-
2
7
formed using Gaussian 94. The geometry of each structure
2
8,29
was optimized using the B3LYP functional
at the 6-31+G*
3
0
level of theory. Diffuse functions were included to more
3
0
accurately model the anions. Frequency calculations were
also performed at this level to verify that each structure was
a minimum or transition state, as well as to obtain zero-point
vibrational and thermal energy corrections to 298K. Zero-point
vibrational energies were scaled by 0.9806. The geometry of
each transition state was displaced (10% along the normal
coordinate for the imaginary vibrational frequency using the
(28) Becke, A. D. J . Chem. Phys. 1993, 98, 5648-5652.
(29) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789.
(30) Hehre, W. J .; Radom, L.; Schleyer, P. v. R.; Pople, J . A. Ab Initio
Molecular Orbital Theory; J ohn Wiley & Sons: New York, 1986.
3
1
(
(
31) Scott, A. P.; Radom, L. J . Phys. Chem. 1996, 100, 16502-16513.
32) Deppmeier, B. J .; Driessen, A. J .; Hehre, W. J .; Hume, T. T.;
2
2
program geomfreq. These structures were then carefully
J ohnson, J . A.; Klunzinger, P. E.; Lou, L.; Yu, J . PC Spartan Plus,
reoptimized (opt ) calcfc) to obtain the minima which connect
Version 1.5.3; Wavefunction, Inc.: Irvine, CA, 1996-1998.