A Theoretical and Experimental Study on Acid-Catalyzed Isomerization of 1-Acylaziridines to the Oxazolines. Reexamination of a Possible SNi Mechanism by Using ab Initio Molecular Orbital Calculations

Article abstract of DOI:10.1021/jo961089n

The S_Ni mechanism, which was previously proposed for the isomerization of 1-acylaziridines to the oxazolines, was reexamined theoretically by performing molecular orbital (MO) calculations of 1-formylaziridine and its derivatives as model compounds and experimentally by using 1 (R)-[α-methoxy-α-(trifluoromethyl)phenylacetyl]-2 (S)-methylaziridine (5). At the MP2/6-31G**//RHF/6-31G* level, the activation energy was estimated to be 38.9 kcal mol^-1 for the S_Ni mechanism in which N-protonated 1-formylaziridine 8a(NH⁺) isomerizes to the N-protonated oxazoline 9a(NH⁺). Intrinsic reaction coordinate calculations showed that this reaction proceeds with retention of the ring carbon configuration. Methyl substitution in the aziridine ring reduces the activation energy by 10 kcal mol^-1. The ring closure of N-(2-chloroethyl)formamide (10a) to the oxazoline, which is a model reaction of the rate-determining step for the addition-elimination mechanism, was estimated to have an activation energy of 45.4 kcal mol^-1. The results of these MO calculations are consistent with the observation that the isomerization of the acylaziridine 5 to the oxazoline 6 is facilitated in the presence of weak nucleophiles such as with BF₃·OEt₂ while the formation of 6 is very slow in the presence of stronger nucleophiles such as p-toluenesulfonate. Both theoretical and experimental results suggest that the S_Ni mechanism explains well the isomerization of (R,S)-5 to the oxazoline with BF₂·OEt₂ in refluxing benzene.

Full text of DOI:10.1021/jo961089n

J . Org. Chem. 1997, 62, 3081-3088
3081
A Th eor etica l a n d Exp er im en ta l Stu d y on Acid -Ca ta lyzed
Isom er iza tion of 1-Acyla zir id in es to th e Oxa zolin es.
Reexa m in a tion of a P ossible SNi Mech a n ism by Usin g a b In itio
Molecu la r Or bita l Ca lcu la tion s
Kenzi Hori,*,1a Takeshi Nishiguchi,1b and Aiko Nabeya1c
Institute for Fundamental Research of Organic Chemistry, Kyushu University, Fukuoka 812-81, J apan,
Faculty of Education, Yamaguchi University, Yamaguchi 753, J apan, and Department of Chemistry,
Turumi University, School of Dental Medicine, Yokohama 230, J apan
Received J une 10, 1996 (Revised Manuscript Received February 5, 1997X)
The SNi mechanism, which was previously proposed for the isomerization of 1-acylaziridines to the
oxazolines, was reexamined theoretically by performing molecular orbital (MO) calculations of
1-formylaziridine and its derivatives as model compounds and experimentally by using 1(R)-[R-
methoxy-R-(trifluoromethyl)phenylacetyl]-2(S)-methylaziridine (5). At the MP2/6-31G**//RHF/6-
31G* level, the activation energy was estimated to be 38.9 kcal mol-1 for the SNi mechanism in
which N-protonated 1-formylaziridine 8a (NH+) isomerizes to the N-protonated oxazoline 9a (NH+).
Intrinsic reaction coordinate calculations showed that this reaction proceeds with retention of the
ring carbon configuration. Methyl substitution in the aziridine ring reduces the activation energy
by 10 kcal mol-1. The ring closure of N-(2-chloroethyl)formamide (10a ) to the oxazoline, which is
a model reaction of the rate-determining step for the addition-elimination mechanism, was
estimated to have an activation energy of 45.4 kcal mol-1. The results of these MO calculations
are consistent with the observation that the isomerization of the acylaziridine 5 to the oxazoline 6
is facilitated in the presence of weak nucleophiles such as with BF3‚OEt2 while the formation of 6
is very slow in the presence of stronger nucleophiles such as p-toluenesulfonate. Both theoretical
and experimental results suggest that the SNi mechanism explains well the isomerization of (R,S)-5
to the oxazoline with BF3‚OEt2 in refluxing benzene.
In tr od u ction
decomposition of ROSOCl using both theoretical and
experimental methods.6 They termed the decomposition
with R ) tert-butyl and 1-adamantyl the SN1i mechanism
(i.e., the SN1-like SNi) mechanism.
Convincing experimental evidence for the SNi mecha-
nism only exists for a relatively few reactions.2,3 Two of
the most important examples are the decomposition of
alkyl chlorosulfites (ROSOCl) and chloroformates (RO-
COCl) to the alkyl chlorides. For these reactions, four-
centered cyclic transition states, leading to front-side
substitutions and resulting in the retention of the con-
figuration, have been proposed.4 However, the existence
of such transition states has been challenged. An “ion
pair, multi-stage substitution” has been advanced to
explain the acceleration of the reactions in polar solvents
and the change of products depending on the reaction
conditions.5 Recently, Schreiner et al. elucidated the
We earlier reported another reaction which possibly
proceeds by the SNi mechanism (eq 1).7a More specifi-
cally, (S)-1-(N-phenylcarbamoyl)-2-methylaziridine (1)
isomerizes into (S)-2-anilino-5-methyl-2-oxazoline (2)
with complete retention of configuration at the asym-
metric carbon when it was heated in refluxing benzene
with BF3‚OEt2. In this reaction, 4-methyl isomer was
also observed and the ratio of 5-methyl/4-methyl products
was 88/12. Similarly a benzene solution of boron trifluo-
X Abstract published in Advance ACS Abstracts, April 1, 1997.
(1) (a) Kyushu University. (b) Yamaguchi University. (c) Tsurumi
University.
(2) Gould, E. S. Mechanism and Structure in Organic Chemistry;
Henry Holt and Company Inc.: New York, 1960; p 294.
(3) March, J . Advanced Organic Chemistry: Reactions, Mechanisms,
and Structure; 4th ed.; Wiley-Interscience: New York, 1992; pp 308,
326.
ride free of diethyl ether was found to smoothly isomerize
the aziridine derivatives to the oxazolines. This reaction
also preserved the carbon configuration in the five-
membered ring of the product completely (eq 2): cis-1-
(4) (a) Cowdrey, W. A.; Hughes, E. D.; Ingold, C. K.; Masterman,
S.; Scott, A. D. J . Chem. Soc. 1937, 1252. (b) Hughes, E. D.; Ingold,
C. K.; Whitfield, I. C. Nature 1941, 147, 206. (b) White, E. H. J . Am.
Chem. Soc. 1954, 76, 4497. (b) White, E. H.; Aufdermarsh, J r. J . Am.
Chem. Soc. 1961, 83, 1179. (c) Lewis, E. S.; Herndon, W. C.; Duffy, D.
C. J . Am. Chem. Soc. 1961, 83, 1959. (d) Stevens, C. L.; Munk, M. E.;
Ash, A. B.; Elliott, R. D. J . Am. Chem. Soc. 1963, 85, 3390. (e) Stevens,
C. L.; Dittmer, H.; Kovacs, J . J . Am. Chem. Soc. 1963, 85, 3394. (f)
Christe, K. O.; Pavlath, A. E. J . Org. Chem. 1965, 30, 4104.
(5) (a) Cram, D. J . J . Am. Chem. Soc., 1953, 75, 332. (b) Lewis, E.
S.; Boozer, C. E. J . Am. Chem. Soc. 1952, 74, 308. (c) Boozer, C. E.;
Lewis, E. S. J . Am. Chem. Soc. 1953, 75, 3182. (d) Lewis, E. S.;
Coppinger, C. E. J . Am. Chem. Soc. 1954, 76, 796. (e) Lee, C. C.;
Finlayson, A. J . Can. J . Chem. 1961, 39, 260. (f) Lee, C. C.; Clayton,
J . W.; Lee, D. G.; Finlayson, A. J . Tetrahedron 1962, 18, 1395. (g)
Lee, C. C.; Newman, D.; Thornhill, D. P. Can. J . Chem., 1963, 41, 620.
(h) White, E, H.; Ellinger, C. A. J . Am. Chem. Soc., 1965, 87, 5261.
(6) (a) Schreiner, P. R.; Schleyer, P. R.; Hill, R. K. J . Org. Chem.
1993, 58, 2822, and references cited therein. (b) Schreiner, P. R.;
Schleyer, P. R.; Hill, R. K. J . Org. Chem. 1994, 59, 1849.
(7) (a) Nishiguchi, T.; Tochio, H.; Nabeya, A.; Iwakura, Y. J . Am.
Chem. Soc. 1969, 91, 5841. (b) Nishiguchi, T.; Tochio, H.; Nabeya, A.;
Iwakura, Y. J . Am. Chem. Soc. 1969, 91, 5835. (c) We repeated the
experiment with 1, and 19F NMR spectroscopy of the derivative of 2
(2′) confirmed the 100% retention of the configuration more precisely.
See Nabeya, A.; Endo, T. J . Org. Chem. 1988, 53, 3358.
S0022-3263(96)01089-4 CCC: $14.00 © 1997 American Chemical Society

3082 J . Org. Chem., Vol. 62, No. 10, 1997
Hori et al.
Sch em e 1
(N-phenylcarbamoyl)-2,3-dimethylaziridine (3) gave cis-
2-anilino-4,5-dimethyl-2-oxazoline (4) when it was heated
in refluxing benzene with BF3.7c
ring to form the open chain intermediate 7 followed by
attack of amide oxygen atom to displace the tosylate.
These two Walden inversions proceed with a net reten-
tion of the ring carbon configuration. The last is a front
side attack of the carbonyl oxygen directly on the ring
carbon to break the C-N bond (i.e., the SNi mechanism)
where the configuration of the ring carbon is retained in
the products because the oxygen approaches from the
same side as the leaving group (nitrogen) leaves.
In order to examine the mechanism of these reactions,
we performed ab initio molecular oribital (MO) calcula-
tions of 1-formylaziridine and its derivatives as model
molecules. In this connection, we examined the reaction
mechanism of 1(R)-[R-methoxy-R-(trifluoromethyl)phen-
ylacetyl]-2(S)-methylaziridine (5). All the molecules are
aziridines with an amide fragment. It was experimen-
tally confirmed that the reaction of 5 with BF3‚OEt2 only
produced the oxazoline 6 while the reaction of 5 with
p-toluenesulfonic acid (TsOH) mainly gave a linear
product 7 together with a small amount of 6.
In the present study, we performed ab initio MO
calculations on the isomerization processes to examine
the validity of the SNi mechanism proposed before.7 We
examined in detail the acid-catalyzed isomerization of
8a (NH+) to the oxazoline 9a (NH+) using H+ as a model
acid. Chloride anion Cl- was adopted as a model nu-
cleophile instead of TsO- for the double inversion mech-
anism which involves a reaction via a linear intermediate
10a . In order to investigate the substituent effect, the
reaction of 8b with a methyl group was also investigated,
and the changes in activation energies were estimated.
Meth od of Ca lcu la tion s
The ab initio MO calculations were carried out using
the GAUSSIAN92 program9 running on a NEC-HSP
computer at the Institute for Molecular Science and on
the IBM RS6000 computer. MOPAC Ver. 610 was also
used for the preliminary search of potential profiles of
the reaction mechanisms considered here. Grand and
transition state (TS) geometries were optimized by the
There are three feasible mechanisms for the reactions
we observed.8 The first, shown in Scheme 1, is an SN1
mechanism via a carbenium ion intermediate. However,
this mechanism would cause racemization and produce
a mixture of the products with S- and R-configurations
in the five-membered ring. Therefore, the SN1 mecha-
nism cannot explain our experimental results. The
second, the double inversion mechanism, is an addition-
elimination mechanism which involves two SN2 type
reactions. The tosylate nucleophile attacks the aziridine
(9) Gaussian 92, Revision D.2, M. J . Frish, G. W. Trucks, M. Head-
Gordon, P. M. W. Gill, M. W. Wong, J . B. Foresman, B. G. J ohnson,
H. B. Schlegel, M. A. Robb, E. S. Replogle, R. Gomperts, J . L. Andres,
K. Raghavachari, J . S. Binkley, C. Gonzalez, R. L. Martin, D. J . Fox,
D. J . Defrees, J . Baker, J . J . P. Stewart, J . A. Pople, Gaussian, Inc.,
Pittsburgh PA, 1992.
(10) (a) MOPAC Ver.6., Stewart, J . J . P. QCPE Bull. 1989, 9, 10.
(b) We performed semiempirical PM3 calculations, and then the
obtained PM3 geometries were used as the initial geometries of the
ab initio calculations. Although the PM3 geometries were not far from
the 6-31G* geometries, the semiempirical calculation did not represent
well the energy relation among reactants, intermediates, products, and
TS.
(8) It may be possible to consider another path that the reaction
proceeds via a tight ion-pair intermediate of 8a (NH+) with the broken
three-membered ring. Although we tried to optimize this intermediate
by using several initial geometries, we failed to optimize it and obtained
the product with a five-membered ring. This result suggests that the
there is no tight ion-pair intermediate with a broken three-membered
ring.
(11) Hehre, W. J .; Ditchfield, R.; Pople, J . A. J . Chem. Phys. 1972,
56, 2257.

Acid-Catalyzed Isomerization of 1-Acylaziridines
J . Org. Chem., Vol. 62, No. 10, 1997 3083
F igu r e 1. Optimized structures related to the SNi mechanism together with their geometrical parameters. Values under the
molecules are the total energies in Hartree units calculated at the RHF/6-31G* (MP2/6-31G**//RHF/6-31G*) level of theory.
energy gradient method at the RHF/6-31G*11 level of
theory. The vibration frequency calculations showed that
all the TS structures have only one imaginary frequency.
The optimized structures of the proposed mechanisms
and total energies are shown in Figure 1.
The intrinsic reaction coordinate12 (IRC) method was
used in order to obtain energy profiles of the reaction
pathway via 8a(TS), 8a(NH+,TS), 8a(OH+,TS), 8a(Cl-,TS)
and 10a (TS). The first three TS’s relate to the SNi
mechanism and the last two to the double inversion
mechanism. We adopted 0.02 amu1/2 Bohr as the step
size for the IRC calculations. In order to study the effect
of correlation, MP2/6-31G**//RHF/6-31G* energies were
also calculated. Unless otherwise noted, all energies
mentioned in the discussion are at this high level of
theory.
oxygen of the substrate. In the present calculations, we
used H+ as the model acid instead of BF3. Figure 1
displays the optimized structures of 1-formylaziridine 8a
and its N- and O-protonated forms, 8a(NH+) and 8a(OH+).
The optimized C(4)-C(5) and N(3)-C(4),C(5) lengths in
8a (1.484, 1.442, and 1.431 Å, respectively) are close to
the corresponding lengths observed in aziridine.13 The
CHO plane of the acyl group is neither perpendicular nor
periplanar to the NC2 plane of the aziridine ring so that
this geometrical feature results in the rather long O(1)-
C(5) length of 3.374 Å.
Although the O-protonation to 8a causes changes in
bond lengths, the differences between the unprotonated
and the protonated form are not very large in the three-
membered ring. For example, in 8a (OH+) the O(1)-C(2)
(1.288 Å) lengthens by 0.101 Å and the C(2)-N(3) bond
(1.258 Å) shortens by 0.123 Å in comparison with those
in 8a (1.187 and 1.381 Å) while the bond lengths in the
ring do not change at all (less than 0.01 Å). The CHO
plane is periplanar to the NC2 plane in the O-protonated
form, and then, the O(1)-C(5) length is 3.137 Å which is
shorter by 0.237 Å than that in 8a .
Resu lts a n d Discu ssion s
Exp er im en ta l Resu lts. Amide (R,S)-5 was prepared
from (S)-2-methylaziridine and (S)-R-methoxy-R-(trifluo-
romethyl)phenylacetyl chloride (MTPA chloride). A solu-
tion of 5 dissolved in benzene was added to a refluxing
solution of BF3‚OEt2 (1.1 mol equiv), and the reaction was
monitored by TLC. After 15 min, the reaction was judged
complete, and the solution was cooled, washed with 2 N
NaOH and brine, dried (Na2SO4), and concentrated to
give 6 in nearly quantitative yield. The ratio between
(R,S)-6 (19F NMR: δ 3.18) and (R,R)-6 (19F NMR: δ 3.21)
was found to be 97:3.
Reaction of (R,S)-5 with TsOH was carried out in
exactly the same way as mentioned above. After 15 min,
a part of the reaction mixture was removed for NMR
analysis. The 19F NMR spectrum showed that the sample
consisted of 92% of the ring-opened addition product,
(R,R)-7 and 8 % of (R,S)-6. After 1 h of heating, the
composition of the reaction mixture was found to be the
same as before.
The N-protonation causes two changes related to the
ring geometry in 8a (NH+). The first is lengthening of
the N(3)-C(4) and N(3)-C(5) bonds by ca. 0.07 Å (i.e.,
the protonation weakens these bonds). The second
involves the angle between the CHO and the NC2 planes.
The CHO plane is almost perpendicular to the NC2 plane
so that the O(1)-C(5) length is shorter by 0.498 Å
compared to that observed in the starting material 8a .
This length is also shorter by 0.216 Å than that in
8a (OH+). This short O(1)-C(5) distance should facilitate
attack of the carbonyl oxygen at the C(5) position. The
calculations of the protonated 1-formylaziridines in the
MP2/6-31G**//RHF/6-31G* level of theory demonstrate
that 8a (NH+) is more stable by 1.8 kcal mol-1 than
8a (OH+). Therefore, we first searched the TS structure
8a (NH+,TS) which connects 8a (NH+) and the N-proto-
nated oxazoline 9a (NH+).
Th e SNi Mech a n ism . In the acid-catalyzed isomer-
ization of 1-acylaziridines, the first step is the coordina-
tion of an acid to either the ring nitrogen or the carbonyl
The IRC calculation demonstrates that 8a (NH+,TS)
connects the reactant and the product as shown in Figure
(12) (a) Fukui, K. Acc. Chem. Res. 1981, 14, 363. (b) Head-Gordon,
M.; Pople, J . A. J . Chem. Phys. 1988, 89, 5777.
(13) Kagaku Binran, Kiso-hen II, 3rd ed., Maruzen: Tokyo 1984; p
653.

3084 J . Org. Chem., Vol. 62, No. 10, 1997
Hori et al.
F igu r e 2. Geometry transformation along the IRC from 8a (NH+) to 9a (NH+) by the SNi mechanism. Values under the structures
are the distances from the TS in amu1/2 Bohr unit.14
2. The geometry at s ) -5.16 amu1/2 Bohr14 is close to
that of the reactant 8a (NH+) and the structure at s )
7.98 amu1/2 Bohr resembles the product 9a (NH+). At s
) -2.99 amu1/2 Bohr, the N(3)-C(5) bond is essentially
broken since the distance is 1.762 Å. The O(1)-C(5)
length is 2.629 Å in the TS structure (s ) 0.0 amu1/2
Bohr). After the TS, the terminal C(5)H2 plane is tilted
in the geometries with s ) 0.99-5.99 amu1/2 Bohr. Such
geometrical change allows better overlap of the empty
p-orbital of the CH2 moiety and the lone pair orbital at
the carbonyl oxygen O(1) as shown in the geometry at s
) 2.99 amu1/2 Bohr. The O(1)-C(5) bond is fully devel-
oped at s ) 7.98 amu1/2 Bohr, and its length is 1.272 Å.
It is necessary to check whether or not the geometrical
transformation along the IRC produces a change of the
F igu r e 3. Potential energy profile (right) and the changes in
the O(1)-C(5), (N3)-C(5) distances (left) along the IRC using
RHF/6-31G* calculations.
chirality around the ring carbon C(5). At the TS, the
terminal carbon possesses almost sp2 hybridization since
the deihedral angle HA-C(4)-C(5)-HB, is estimated to
be -173.7°. The terminal CH2 plane tilts a little after
In the isomerization from 8a (NH+) to 9a (NH+), from s
the TS in order to form a new O(1)-C(5) bond. However,
) -5.16 amu1/2 Bohr to the TS, the SN1 type N(3)-C(5)
around the C(4)-C(5) axis we cannot see the rotation
bond cleavage occurs to form an “ion-pair” TS. From the
which leads to the chirality change at C(5) in Figure 2.
TS to the product, the carbonyl oxygen attacks the C(5)
The O(1)-C(5) and the N(3)-C(5) distances are good
center from the same side of the broken N(3)-C(5) bond
reaction coordinates for following the SNi mechanism.
as seen in Figure 2. Adopting the nomenclature proposed
Figure 3 displays the distances along the IRC together
by Schreiner et al., we may describe the process as the
with the potential energy profile. From s ) -6.0 to -4.0
SN1i reaction.
amu1/2 Bohr, neither the N(3)-C(5) length (ca. 1.5 Å) nor
The activation energy (Ea) of the N-protonated SNi
the potential energy changes although the O(1)-C(5)
mechanism is calculated to be 38.9 kcal mol-1 at the MP2/
length shortens by ca. 0.2 Å. Between s ) -3.99 and
6-31G**//RHF/6-31G* level of theory. Figure 4 displays
the energy diagram of the molecules under consideration.
0.00 amu1/2 Bohr, the O(1)-C(5) distance remains almost
constant at 2.7 Å while the N(3)-C(5) distance lengthens
On the other hand, the Ea without protonation was
by ca. 0.8 Å. The change in the potential energy almost
calculated to be 49.2 kcal mol-1, and therefore, addition
parallels that of the length of the N(3)-C(5) bond. After
the TS, the O(1)-C(5) distance shortens quickly to form
the five-membered ring. The quick descent of the po-
tential energy after the TS originates from this bond
formation. On the other hand, the N(3)-C(5) distance
fluctuates a little during this period.
of a simple Lewis acid reduces the activation barrier by
10.3 kcal mol-1
.
It is possible to consider an alternative mechanism that
the O-protonated reactant 8a (OH+) isomerizes to the
O-protonated product 9(OH+) via 8a (OH+,TS), whose
structures are also displayed in Figure 1. The O(1)-C(5)
length in 9a (OH+,TS) is shorter by 0.084 Å, and the
N(3)-C(5) one is longer by 0.061 Å than the correspond-
(14) “s” in amu1/2 Bohr units is a measure of the distance of a
molecular configuration from the TS. See ref 12a.

Acid-Catalyzed Isomerization of 1-Acylaziridines
J . Org. Chem., Vol. 62, No. 10, 1997 3085
F igu r e 4. The energy relation diagram in kcal mol-1 unit among the reactants, the transition states, and the products. The
values are relative to the total energy of 8a (NH+) or 8a +Cl- or 10a . Values in brackets and parentheses are the activation
energies for the reaction from 8a (OH+) to 9a (NH+) and for the mechanism producing 5-methyl and 4-methyl derivatives of 9b(NH+),
respectively.
ing lengths in 8a (NH+,TS). The Ea of the O-protonated
path is 48.2 kcal mol-1. The protonation at the carbonyl
oxygen reduces the nucleophilicity of O(1) and 8a (OH+)
is less stable by 1.8 kcal mol-1 than 8a (NH+). Therefore,
the activation barrier in the O-protonated mechanism is
larger by 11.1 kcal mol-1 than that in the N-protonated
mechanism. Therefore, it is likely that the SNi reaction
proceeds not via 8a (OH+,TS) but rather via 8a (NH+,TS).
Dou ble In ver sion Mech a n ism . Double inversion
mechanism consists of two SN2 reactions. The first
involves the ring opening caused by the attack of a
nucleophile on C(5) from the opposite side of the N(3) to
form an acyclic intermediate. We used chloride as a
model nucleophile in the present study. The MP2/6-
31G**//RHF/6-31G* energies of 8a +Cl- and 8a (Cl-,TS)
are calculated to be -706.18175 and -706.20408 Hartree,
respectively.15 As the activation barrier is estimated to
be only 14.0 kcal mol-1, the ring opening caused by
chloride attack appears very facile. This reaction ac-
companies the so-called Walden inversion seen in Figure
5 which displays the geometry transformation along the
IRC. Starting at s ) -5.98, the chloride nucleophile
approaches the ring methylene moiety, and then, the
N(3)-C(5) bond breaks at s ) -1.00 amu1/2 Bohr. The
structure at s ) 7.01 amu1/2 Bohr, where a linear anionic
intermediate has fully formed, has two H atoms directing
to the another side of Cl-. We do not see such a
configuration change at the C(5) in the SNi mechanism
mentioned above. After the formation of the anionic
intermediate, it accepts an H+ to form 10a . It is
important to point out that no TS structure for the ring
opening was obtained for 8a (NH+) and 8a (OH+). It is,
therefore, gathered that this reaction proceeds without
protonation to the reactant.
The carbonyl oxygen O(1) functions as a nucleophile
in the second SN2 reaction. This reaction begins by
attack of the carbonyl oxygen onto the C(5) center from
the opposite direction of the Cl atom and ends with the
formation of a five-membered ring. In Figure 6, we can
see another Walden inversion around C(5). At s ) -5.85
amu1/2 Bohr, O(1), C(5), and Cl make nearly a straight
alignment, which facilitates attack of O(1) on the C(5)-
H2Cl fragment. The C(5)-Cl bond disappears at s )
-2.99 amu1/2 Bohr since the length is 2.038 Å. At s )
-0.99 amu1/2 Bohr, the C(5) takes sp2 hybridization
because C(4) is almost in the plane defined by the C(5)-
H2 moiety. The five-membered ring has already formed
(O(1)-C(5): 1.512 Å) and the C(5)-Cl length is 2.886 Å
at s ) 2.54 amu1/2 Bohr so that the reaction has nearly
completed at this point on the IRC pathway.
The MP2/6-31G**//RHF/6-31G* energies for the second
SN2 reaction are -706.79937, -706.72697, and
-706.72701 Hartree for 10a , 10a (TS), and 9a (NH+,Cl-),
respectively. The activation barrier of this process is 45.4
kcal mol-1 16
The energy of the ring formation is higher
.
by 24.9 kcal mol-1 than that of the first SN2 reaction and
by 6.5 kcal mol-1 than that of the SNi mechanism.
Su bstitu en t Effect. According to our experimental
results, heating 5 in benzene with BF3‚OEt2 produced
exclusively the 5-methyloxazoline 6, and no 4-methyl
isomer is formed. In order to explain this behavior, we
optimized the methyl-substituted aziridine 8b(NH+) and
the two TS’s, 8b(5,NH+,TS) and 8b(4,NH+,TS), leading
to the formation of the 5-methyl- and 4-methyloxazolines,
respectively. Figure 7 displays their optimized struc-
tures. As the N(3)-C(5) length (1.535 Å) is longer by
0.026 Å than the N(3)-C(4) one (1.509 Å) in 8b(NH+),
the introduction of a methyl group on C(5) weakens the
N(3)-C(5) bond.
(15) Because the 8a (TS,Cl-) is an anion, we checked the effect
diffuse functions for the activation energy. MP2/6-31+G**//RHF/6-
31G* energies of Cl-, 8a , and 8a (TS,Cl-) are -459.67115, -246.54842,
and -706.21702 Hartree, respectively. The activation energy is
estimated to be 19.2 kcal mol-1 which is larger by 5.2 kcal mol-1 than
the value from the calculation without diffuse functions. This value
is still much smaller than that for the second SN2 mechanism and for
the SNi mechanism. The addition of diffuse functions does not cause
any differences in the rank ordering of relative energies.
(16) The MP2/6-3+G**//RHF/6-31G* energies for the second SN2
reaction are -706.81763, -706.74550, and -706.74669 Hartree for
10a , 10a (TS), and 9a (NH+,Cl-), respectively. The activation energy
for the second reaction is 45.3 kcal mol-1, and the energy difference
The O(1)-C(5) length in 8b(5,NH+,TS) is 2.788 Å and
the O(1)-C(4) distance in 8b(4,NH+,TS) is 2.607 Å. On
the other hand, the three-membered ring is broken more
between the last two geometries is only 0.8 kcal mol-1
functions do not cause any serious differences, either.
. The diffuse

3086 J . Org. Chem., Vol. 62, No. 10, 1997
Hori et al.
F igu r e 5. Geometry transformation along the IRC of the first SN2 reaction from 8a +Cl- to 10a (N-). Values under the structures
are the distances from the TS in amu1/2 Bohr unit.12
F igu r e 6. Geometry transformation along the IRC of the second SN2 reaction from 10a to 9a (NH+)+Cl-. Values under the
structures are the distances from the TS in amu1/2 Bohr unit.12
in 8b(5,NH+,TS) than that in 8b(4,NH+,TS) because the
N(3)-C(5) length in the former (2.253 Å) is longer by
0.096 Å than the N(3)-C(4) one in the latter (2.157 Å).
It is, therefore, concluded that the reaction attacking at
C(5) reaches the TS in the earlier stage than that at C(4).
to explain the experimental result that the only 5-methyl
isomer was obtained for the reaction of 5 proceeding by
the SNi mechanism. However, it must be pointed out that
Cl- and H+ were adopted as a nucleophile and an acid,
respectively, in the MO calculations while TsO- and BF3
were employed in the experiments.
The activation energies are estimated to be 28.2 and
39.8 kcal mol-1 for the mechanisms producing the
5-methyl product via 8b(5,NH+,TS) and the 4-methyl
product via 8b(4,NH+,TS). A simple methyl substitution
differentiates the activation energy for the SNi reaction
by 11.6 kcal mol-1. This energy difference is large enough
Figure 7 also includes the transition state geometries
obtained for the ring closure reaction which is the rate-
determining step of the addition-elimination mecha-
nism. While the O(1)-C(5) and C(5)-Cl lengths in
10b(5,NH+,TS) are 1.909 and 2.635 Å, the O(1)-C(4) and

Acid-Catalyzed Isomerization of 1-Acylaziridines
J . Org. Chem., Vol. 62, No. 10, 1997 3087
F igu r e 7. Optimized and TS structures for the methyl-substituted aziridines. Values under the molecules are the total energies
in Hartree units calculated at the RHF/6-31G* (MP2/6-31G**//RHF/6-31G*) level of theory.
C(4)-Cl lengths in 10b(4,NH+,TS) are 1.823 and 2.533
Å, respectively. The former TS leads to the 5-methyl
product and the latter TS to the 4-methyl product.
Although the carbonyl oxygen O(1) as the nucleophile is
located at a more distant position from O(1) in 10b(5,NH+,
TS) than that in 10b(4,NH+,TS), the C-Cl bond in the
former TS is weakened more in the latter TS. The
activation energies are calculated to be 34.9 and 44.8 kcal
mol-1, respectively. These values are larger by more than
5-15 kcal mol-1 than those for the SNi mechanism.
While the Ea to 4-methyl product is similar to the
corresponding value for 10a (45.4 kcal mol-1), that to the
5-methyl product is smaller by 10.4 kcal mol-1 than that
value.
mol-1 for the subsequent ring closure reaction, the rate
for the formation of the final product is expected to be
very slow. However, a tosylate group in the ring-opened
intermediate, which was used in the experiment, is
apparently sufficiently better leaving group as compare
to the chloride18 so that the second SN2 reaction proceeds
on a reasonable time scale. The introduction of a methyl
group at either C(4) or C(5) differentiate the activation
barrier of this process.
In the absence of strong nucleophiles, the oxazoline
formation proceeds by the SNi mechanism which has a
smaller activation barrier than the second SN2 reaction
of the addition-elimination mechanism. While the
methyl substituent at C(5) reduces the activation energy
by more than 10 kcal mol-1, that at C(4) increases that
value by 0.9 kcal mol-1
. This is consistent with the
Con clu d in g Rem a r k s
exclusive formation of 5-methyl product in the reaction
of 5.
For the acid-catalyzed transformation of 1-acylaziri-
dines to the oxazolines, we have shown both theoreti-
cally17 and experimentally that there are two plausible
mechanisms, which preserve the configuration of the
carbons in the ring. The first is the path termed the SNi
mechanism (the reaction catalyzed by H+ or BF3‚OEt2),
and the second is the addition-elimination involving the
two SN2 reactions (the double inversion concerned with
Cl- or TsO-).
Exp er im en ta l Section
Gen er a l In for m a tion . Elemental analyses were per-
formed by The Institute of Physical and Chemical Research,
Wako, Saitama. (S)- or (R)-MTPA chloride was purchased
from Aldrich Chemical Company (Milwaukee, WI). NMR
spectra (1H and 19F) were recorded on a J EOL GSX-270 (270
MHZ for 1H and 254 MHz for 19F) spectrometer using CDCl3
as a solvent, and TFA as a standard for 19F NMR.
P r ep a r a t ion of (R)-[r-Met h oxy-r-(t r iflu or om et h yl)-
p h en yla cetyl]-2(S)-m eth yla zir id in e (5). Into an ice-cooled
solution of 63 mg (1.1 mmol) of (S)-2-methylaziridine and 110
mg (1.1 mmol) of Et3N in ether was added a solution of 253
mg (1.0 mmol) of (S)-MTPA chloride in ether dropwise. After
2 h of stirring, the precipitated Et3NHCl was removed by
filtration, and the filtrate was washed with water, dried (Na2-
SO4), and concentrated on a rotary evaporator to give practi-
cally pure (R,S)-5 in quantitative yield. Purification of 5 by
column chromatography (n-hexane-ethyl acetate) gave ana-
lytical sample as an oil: 1H NMR: δ 1.25 (d, 3H, J ) 5.5,
In the presence of the strong nucleophile Cl-, the ring-
opening reaction easily proceeds because of a small
activation barrier of 14.0 kcal mol-1
. This value is
smaller by 24.9 kcal mol-1 than that for the SNi mecha-
nism. As the activation energy is as large as 45.4 kcal
(17) We have done the SCRF calculations of the IPCM method (ꢀ )
2.0) at the MP2/6-31G**//RHF/6-31G* level of theory in order to
estimate solvent effect, although we used the solvent in experiments
with low a dielectric constant such as benzene. The activation energy
for 8a (NH+) was estimated to be 38.8 kcal mol-1 which differs only by
0.1 kcal mol-1 from the value in the gas phase. The Ea of the double
inversion mechanism were estimated to be 13.6 and 37.1 kcal mol-1
for the first and the second reactions, respectively. These values are
differe by 0.4 and 8.3 kcal mol-1 from the gas phase value. Therefore,
the potential profiles including solvent effect is not largely different
from those without the solvent effect.
(18) March, J . Advanced Organic Chemistry: Reactions, Mechanisms,
and Structure; 4th ed.; Wiley-Interscience: New York, 1992; p 357.

3088 J . Org. Chem., Vol. 62, No. 10, 1997
Hori et al.
CCH3), 1.99 (d, 1H, J ) 3.7, Ha of 3-CH2), 2.17 (d, 1H, J )
5.7, Hb of 3-CH2), ∼2.6 (m, 1H, 2-CH), 3.55 (d, 3H, J ) 1.5,
O-CH3). 19F NMR: δ 5.86. Anal. Calcd for C13H14NO2F3: C,
57.14; H, 5.16; N, 5.13. Found: C, 57.07, 56.97; H, 5.15, 5.14;
N, 4.98, 5.14.
of (R,R)-7 (δ 6.92, 92%) and (R,S)-6 (δ 3.43,19 8%). After 1 h,
the remaining reaction mixture was treated in the same
manner. Inspection by 19F NMR showed that the ratio of 7/6
was practically unchanged.
P r ep a r a tion of Au th en tic Sa m p le of (R,R)-7. From (S)-
MTPA chloride and (R)-2-aminopropanol was obtained (R)-2-
hydroxypropyl-(R)-MTPA amide (7′) as a syrup. Treatment
of the crude 7′ with p-toluenesulfonyl chloride and pyridine
gave (R,R)-7 in crystalline form, which was recrystallized from
benzene and n-hexane (50:50) to give analytical sample: mp
Compound (S,S)-5 was prepared from (S)-2-methylaziridine
and (R)-MTPA chloride in the same way as mentioned above.
NMR spectral data of (S,S)-5: 1H NMR: δ 1.20 (d, 3H, J )
5.5, C-CH3), 1.91 (d, 1H, J ) 3.7, Ha of 3-CH2) 2.00 (d, 1H, J
) 2.0, Hb of 3-CH2), ∼2.0 (m, 1H, 2CH), 3.71 (d, 3H, J ) 1.8,
O-CH3). 19F NMR: δ 6.05.
1
80-82 °C; H NMR: δ 1.19 (d, 3H, J ) 6.4, CCH3), 2.45 (s,
3H, Ts-CH3), 3.39 (d, J ) 1.5, OCH3), ∼3.5 (m, 1H, Ha of CH2),
∼3.6 (m, 1H, Hb of CH2), 4.74 (m, 1H, CH). 19F NMR: δ 6.92.
Anal. Calcd for C20H22NO5F3S: C, 53.92; H, 4.98; N, 3.15.
Found: C, 54.27; H, 4.91; N, 3.07.
Rea ction of 5 w ith BF 3‚OEt2. Into a refluxing solution
of 80 mg (0.56 mmol) of BF3‚OEt2 in benzene (dried over CaH2,
0.5 mL) was added a solution of 0.5 mmol of (R,S)-5 in benzene
(0.5 mL) dropwise. After 15 min, the reaction mixture was
cooled, washed with 2 N NaOH and brine, and then dried (Na2-
SO4). Evaporation of the solution on a rotary evaporator gave
practically pure 6 as judged by 1H and 19F NMR in nearly
quantitative yield. 19F NMR analyses of the crude 6 showed
that it was composed of 97% of (R,S) (δ 3.18) and 3 % of (R,R)
(δ 3.21). Purification by column chromatography (benzene-
ethyl acetate) gave analytical sample (oil): 1H NMR: δ 1.32
(d, 3H, J ) 6.2, CCH3), 3.57 (d, 3H, J ) 1.0, OCH3), 3.65 (dd,
P r ep a r a tion of Au th en tic Sa m p le of (R,S)-6. To a
solution of 150 mg of KOH in 3 mL of EtOH was added 150
mg of (R,R)-7, and the mixture was stirred at room temper-
ature. As the spot of 7 disappeared on the TLC after 30 min,
water was added to the mixture, and the mixture was
concentrated on a rotary evaporator to remove EtOH. Benzene
was added to the residue, and the organic layer was washed
with brine, dried (Na2SO4), and concentrated to give 100 mg
of crude (R,S)-6, which was purified by column chromatogra-
phy using benzene and ethyl acetate. Both 1H and 19F NMR
spectra coincided with those of isomerization product from
(R,S)-5 with BF3‚OEt2.
1H, J gem ) 14.7, J vic ) 7.3, Ha of 4-CH2), 4.19 (dd, 1H, J gem
)
14.8, J vic ) 9.4, Hb of 4-CH2), ∼4.8 (m, 1H, 5-CH). Anal. Calcd
for C13H14NO2F3: C, 57.14; H, 5.16; N, 5.13. Found: C, 57.37;
H, 5.21; N, 5.09.
Reaction of (S,S)-5 with BF3‚OEt2 gave (S,S)-6: 1H NMR:
δ 1.40 (d, 3H, J ) 6.2, C-CH3), 3.35 (d, 3H, J ) 1.3, OCH3).
19F NMR: δ 3.21.
Ack n ow led gm en t. The authors owe thanks to the
Computer Center, Institute for Molecular Science at the
Okazaki National Research Institutes, for the use of the
NEC HSP computer and the Library Program GAUSS-
IAN92. This work is supported by the Grant-in-Aid for
Scientific Research from the Ministry of Education of
J apan.
Rea ction of (R,S)-5 w ith p-Tolu en esu lfon ic Acid . p-
Toluenesulfonic acid monohydrate (102 mg, 0.54 mmol) was
placed in a flask and heated at 110 °C under reduced pressure
for 1 h to remove the water. Benzene (0.5 mL) was added to
the flask, and the solution was heated to reflux. To the
refluxing solution was added a solution of 0.5 mmol of (R,S)-5
in benzene (0.5 mL), and the resulting solution was heated at
reflux. After 15 min, about a half of the reaction mixture was
removed, washed with 1 N NaOH followed by brine, and dried
over anhydrous sodium sulfate. After evaporation of benzene
on a rotary evaporator, the residue was submitted to NMR
Su p p or tin g In for m a tion Ava ila ble: Z-matrices and ener-
gies of all structures (21 pages). This material is contained
in libraries on microfiche, immediately follows this article in
the microfilm version of the journal, and can be ordered from
the ACS; see any current masthead page for ordering
information.
J O961089N
(19) It was found that the 19F NMR signal of CF3 of 6 was shifted
to lower field in the presence of 7. Such a tendency was found to be
pronounced when the ratio of 7/6 is large and the concentration of the
solution is high.
1
inspection. The H NMR spectrum was practically the same
as that of (R,R)-7 except that it had a doublet at 1.34 (d, CCH3
of (R,S)-6). By 19F NMR analysis, it was found to be a mixture