Mechanistic approaches to asymmetric synthesis of aziridines from guanidinium ylides and aryl aldehydes

Article abstract of DOI:10.1016/j.tet.2005.01.088

To approach more realistic mechanisms for asymmetric aziridine synthesis from guanidinium ylides and aryl aldehydes, reactions were systematically carried out by using a variety of p-substituted benzaldehydes under modified conditions. Two kinds of reaction mechanisms controlled by the nature of the p-substituents of aryl aldehydes is proposed for the two-steps aziridine synthesis composed of a C-C bond formation by nucleophilic addition of guanidinium ylides to aryl aldehydes (step 1) and the fragmentation of intermediate adducts to aziridine products by intramolecular nucleophilic substitution (step 2). A S_Ni-like mechanism via cationic-like transition state is proposed for step 2 in the asymmetric synthesis using EDG-substituted benzaldehydes, whereas with EWG-substituted benzaldehydes, a S_N2-like mechanism is proposed. Hammett analysis, based on the diastereomeric ratio in the aziridine products, is consistent with the proposed rate-determining steps in these two mechanisms. A second Hammett analysis, based on the enantiomeric ratio of the aziridine products, clearly reveals the difference in the susceptibilities to the electronic substituents effect between step 1 and step 2.

Full text of DOI:10.1016/j.tet.2005.01.088

Tetrahedron 61 (2005) 2857–2869
Mechanistic approaches to asymmetric synthesis of aziridines
from guanidinium ylides and aryl aldehydes
Toyokazu Hagaa,b and Tsutomu Ishikawaa,
*
aGraduate School of Pharmaceutical Sciences, Chiba University, 1-33 Yayoi, Inage, Chiba 263-8522, Japan
bProduction Department, Technology Group, Eisai Co., Ltd., 22-Sunayama, Hasaki-machi, Kashima-gun, Ibaraki 314-0255, Japan
Received 20 December 2004; accepted 19 January 2005
Abstract—To approach more realistic mechanisms for asymmetric aziridine synthesis from guanidinium ylides and aryl aldehydes,
reactions were systematically carried out by using a variety of p-substituted benzaldehydes under modified conditions. Two kinds of reaction
mechanisms controlled by the nature of the p-substituents of aryl aldehydes is proposed for the two-steps aziridine synthesis composed of a
C–C bond formation by nucleophilic addition of guanidinium ylides to aryl aldehydes (step 1) and the fragmentation of intermediate adducts
to aziridine products by intramolecular nucleophilic substitution (step 2). A SNi-like mechanism via cationic-like transition state is proposed
for step 2 in the asymmetric synthesis using EDG-substituted benzaldehydes, whereas with EWG-substituted benzaldehydes, a SN2-like
mechanism is proposed. Hammett analysis, based on the diastereomeric ratio in the aziridine products, is consistent with the proposed rate-
determining steps in these two mechanisms. A second Hammett analysis, based on the enantiomeric ratio of the aziridine products, clearly
reveals the difference in the susceptibilities to the electronic substituents effect between step 1 and step 2.
q 2005 Elsevier Ltd. All rights reserved.
1. Introduction
(SiO2). Introduction of chiral centers into the guanidinium
template results in effective asymmetric induction in the
aziridine formation. In this unique cycle, aziridines are
generated with excellent to moderate stereoselectivity
depending upon the choice of the aryl aldehydes 3. In
general trans-aziridines are efficiently obtained with
satisfactory enantioselectivity when aryl aldehydes carrying
an electron-donating group (EDG) such as piperonal (3a)
are used as electrophiles. Based on stereochemical results
we postulated a mechanism for the asymmetric induction
through carbocation-alkoxide intermediates 5 and spiro
intermediates 4 (Scheme 1).
Three-membered nitrogen heterocycles, aziridines, are very
important molecules not only as key components of
biologically active natural products such as mitomycins,
but also as reactive synthetic precursors for a wide variety of
nitrogen-containing compounds.1 Among them 3-arylaziri-
dine-2-carboxylates are versatile precursors for the syn-
thesis of amino acid derivatives including unnatural type
products.2 Therefore, there have been many approaches to
aziridine preparation involving asymmetric synthesis,
which could be basically classified into three types of
reactions:1 (i) intramolecular substitution of b-aminoalco-
hols by nucleophilic nitrogens, (ii) addition of carbenes
to imines, and (iii) addition of nitrenes to olefins. Recently,
we reported a new synthetic method for the preparation of
3-arylaziridine-2-carboxylates 6 from guanidinium salts
which carry a glycine unit (e.g., 1) and aryl (including
heterocyclic) aldehydes 3.3 In the proposed cyclic mecha-
nism, guanidinium ylide 2 is formed from the guanidinium
salt 1 under basic conditions and acts as a nucleophile.
Finally, urea 7, a synthetic precursor of 1, is produced as
a co-product of aziridines after treatment with silica gel
In ylide chemistry, the Wittig reaction is a well known
process;4 a phosphonium ylide reacts with a carbonyl
compound such as aldehydes or ketones to give an alkene by
C–C bond formation together with phosphine oxide, in
which the phosphorus atom of the ylide acts as oxygen
acceptor from the carbonyl substrate (see A, Scheme 2). In
the reaction of sulfonium ylide with a carbonyl substrate, the
product is oxirane in which the oxygen atom comes from
the carbonyl substrate5 (see B, Scheme 2). However, in the
guanidinium ylides mechanism, the external nitrogen atom,
among three ylide nitrogens, is incorporated into aziridines
and the remaining amidine moiety acts as an oxygen
acceptor to be converted to urea (see C, Scheme 2). Thus,
these three types of ylide-participating reactions with aryl
Keywords: Asymmetric synthesis; Aziridine; Guanidinium ylide; Mechan-
ism; Hammett analysis.
* Corresponding author. Tel./fax: C81 43 290 2910;
e-mail: benti@p.chiba-u.ac.jp
0040–4020/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2005.01.088

2858
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
Scheme 1. The postulated mechanism for asymmetric aziridine synthesis from guanidinium salt 1 and aryl aldehydes 3.
result, we proposed more realistic mechanisms to cover
the overall aziridine preparation from guanidinium ylides by
application of Hammett relation. In this paper we present
two different reaction pathways depending upon the
presence of either EDG or electron-withdrawing group
(EWG) on the aryl aldehyde components.
2. Results and discussion
Aziridine synthesis is basically composed of two steps
(Scheme 3), as previously mentioned in a communication.3
Step one is a C–C bond formation between a guanidinium
salt 1 and an aryl aldehyde 3 under basic conditions in which
an initially-formed zwitter ionic species is in equilibria with
a non-ionic spiro compound 4. The second step is the
fragmentation of the intermediate adduct, triggered with
SiO2, to afford aziridine product 6 and urea 7. This two steps
reaction can be followed by a normal phase TLC: a first
intermediate adduct is observed as a highly polar product
then, after SiO2 treatment, a less-polar component is
generated.
Scheme 2. Schematic reaction profiles of phosphonium, sulfonium, and
guanidinium ylides with aryl aldehydes.
aldehydes afford quite comparable profiles in the product
formation.
Despite the reasonable mechanisms proposed for the
reactions of phosphonium4 and sulfonium ylides5 with
carbonyl substrates, the reactivity and the overall stereo-
selectivity observed in the guanidinium ylide chemistry
could not be rationally explained by the mechanism3
proposed in Scheme 1. The presence in some cases of cis-
aziridine as major isomer, forces us to reevaluate the role of
carbocation-alkoxide intermediates 5. Thus, reactions were
systematically carried out by using a variety of p-substituted
benzaldehydes under modified conditions and, as a
Scheme 3. C–C bond formation between a guanidinium salt 1 and an aryl
aldehyde 3 followed by fragmentation to aziridine 6 and urea 7.

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
Table 1. Effect of temperature on the C–C bond formation (step 1) of 1 and 3a
2859
Run
Temperature (8C)
Time (h)
trans-6a
ee (%)b
cis-6a yield (%)a
Yield (%)a
1
2
3
4
5
6
7
8
K35
K20
K10
10
25
40
O48c
49
73
86
91
88
89
88
86
34
84
83
85
92
91
91
89
84
4
5
5
5
6
6
5
3
23
4.5
4.5
4.5
4.5
4d
60
K20
A 1.5 M equiv of TMG was used.
a Determined by quantitative HPLC.
b Determined by chiral HPLC.
c A trace of 3a was detected at 48 h.
d Unreacted 3a was observed on TLC.
In this mechanistic approach, (4S,5S)-4,5-diphenylimidazo-
line-derived guanidinium salt 1 and piperonal (3a) have
been selected as an ylide and aryl aldehyde source because
effective production of trans-aziridine trans-6a has been
observed with high diastereomeric excess (de) and enantio-
meric excess (ee). Although 1,1,3,3-tetramethylguanidine
(TMG) has not been used only as a base but also as solvent
in the previous communication,3 the reaction conditions
were modified to use a limited amount of tetrahydrofuran
(THF) as solvent in order to obtain reproducible results.†
Thus, trans-6a and cis-6a were obtained in 89 and 6%
yields, respectively, in the reaction of 1 and 3a at 25 8C for
4.5 h followed by fragmentation with SiO2 in chloroform
for 24 h, as reported previously3 (see run 5, Table 1). The ee
of the former trans-6a is 91%.
dependent upon the conditions used, even with a small
difference, suggests that no equilibration between the
intermediate and the starting piperonal occurs during the
SiO2 treatment (step 2). Thus, it is reasonable to deduce that
the overall reaction of this aziridine formation is controlled
by kinetic mode.
Next, we approached to the structure of the intermediate
adduct by spectroscopic means; however, direct NMR study
on the reaction mixture obtained in step 1 using a variety of
solvents (CDCl3, CD3CN, benzene-d6, and pyridine-d5)
resulted in a complex signal pattern in all cases examined.
Although trials for the isolation of the adduct derived from
piperonal (3a) were failed, the intermediate adduct in the
reaction using p-chlorobenzaldehyde (3g) mentioned later
was successfully isolated as a ca. 5:1 mixture of two
isomeric polar compounds (see Section 4.2.11), which was
smoothly converted into the corresponding aziridines by
SiO2 treatment [trans-6g: 22% yield (84% ee); cis-6g: 55%
yield (86% ee)]. Unfortunately, in this case, a decisive
structural information on the adduct could not be taken even
using 2D NMR techniques under various conditions. These
situations force us to attempt to chemically trap the zwitter
ionic intermediate adduct (see Scheme 3).
2.1. Examination of reaction conditions: effect of
temperature
First we examined the effect of temperature on the C–C
bond formation of step 1 to determine whether the reaction
was controlled by either kinetic or thermodynamic mode
(Table 1). The reaction was carried out at various
temperature until the starting materials disappeared on
TLC and, then, the formed adduct intermediate was treated
with SiO2 for 24 h (runs 1–7). Only small difference in
product profiles was observed in all reactions examined and,
in addition, identical enantioselectivity in aziridine products
were obtained in shorter treatment (4 h, run 8) as well as in
longer one (49 h, run 2). These facts indicate that step 1 is
basically under kinetic control, although higher asymmetric
induction at higher temperature (runs 1–3 vs 4–7) suggests
the partial participation of thermodynamic mode to the C–C
bond formation. Moreover, variable ee of trans-6a
2.2. Chemical approaches to the adduct: fragmentation
to aziridine with acid anhydrides
Trials for the derivatization of the intermediate adduct using
p-nitrobenzenesulfonyl chloride, m-bromobenzoyl chloride,
or tert-butyldimethylsilyl triflate were unsuccessful. On the
other hand, it was found that treatment with acetic
anhydride (Ac2O) in chloroform immediately convert the
intermediate adduct into trans-6a in 86% yield with 87% ee
together with cis-6a (7% yield) and urea 7 (80% yield), but
no trapped product (run 1, Table 2). The same asymmetric
induction‡ as with SiO2 suggests that both reactions
occurred under the same stereochemical controls,
† In the preliminary communication,3 we reported the effectiveness of
reaction under solvent-free conditions; however, inconstant results were
sometimes obtained in scale-up reactions. After precise examination,
homogeneous condition in limited amounts of THF solution led to
reproducible results. Other organic bases examined such as DBU did not
initiate the reaction.
‡ It was confirmed by chiral HPLC analysis. See Section 4.

2860
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
Table 2. Effect of additives on the fragmentation (step 2) of the intermediate adduct obtained from 1 and 3a
Run
Additive/solvent
M equiv/mL
Time (h)
trans-6a
ee (%)c
cis-6a yield (%)a
7 yield (%)b
Yield (%)a
1
2
3
4
5
6
7
8
9
10
Ac2O/CHCl3
2.2d/10
2.2/10
2.2/10
2.2/10
2.2/0.8
2.2/0.4
2.2/10
2.2/0.4
2.2/0.4
2.2/10
0.3
1
2
14.5
2
0.3
43
28
89
0.5
86
87
83
84
68b
77
57
53
71
68
87
85
90
91
92
84
87
90
86
88
7
7
7
8
80
e
(EtCO)2O/CHCl3
(iPrCO)2O/CHCl3
(tBuCO)2O/CHCl3
(PhCO)2O/CHCl3
Ac2O/THF
(CH2CO)2O/THF
(tBuCO)2O/THF
(Boc)2O/THF
–
e
–
e
–
e
e
–
–
e
6
3
4
6
6
–
e
–
e
–
e
–
79
Ac2S/CHCl3
A 1.1 M equiv of TMG was used because excess amount had no effect on the reaction.
a Determined by quantitative HPLC.
b Isolated yield, but not optimized.
c Determined by chiral HPLC.
d Incomplete reaction was observed when treated with 1.1 M equiv of Ac2O for 0.7 h. Further addition of Ac2O (total 2.2 M equiv) led to the conversion to
aziridines. Therefore, 2.2 M equiv of an acid anhydride was used at the beginning of all experiments.
e Not isolated.
prompting us to further examine the fragmentation reaction
with various acid anhydrides for the scope and limitation.
fragmentation proceeds rapidly to give trans-6a in 68%
yield with 88% ee together with urea 7 (79% yield) (run 10,
Table 2), no thiourea could be detected at the end of the
reaction.¶ From this experiment, we can conclude that the
urea oxygen comes from the aldehyde oxygen of piperonal
(3a) even in the acid anhydride-triggered fragmentation.
Although alkyl acid anhydrides effectively act as initiators
in the reaction, reaction rate is dependent upon the bulkiness
of the acid anhydride used and, thus, the bulkier the alkyl
group is, the longer it takes for the conversion to occur while
the enantioselectivity is higher (runs 2–4, Table 2). We as
well observed that this aziridine preparation method is
sensitive to steric demands of not only nucleophiles but also
electrophiles; either replacement of benzyl or alkoxycarbo-
nylmethylene groups on the external nitrogen of guani-
dinium salts to bulkier ones or the use of ketone substrates in
place of aldehydes led to nearly no production of aziridine
products.§ These facts suggest that the reaction site of the
intermediate adduct is fairly crowded and that the acylation
with an acid anhydride is a rate-determining step in this
fragmentation sequence.
The fragmentation reaction may be initiated by the
corresponding acid possibly generated by the acylation of
the remaining TMG with an acid anhydride. The excess
TMG in a reaction mixture should be stoichiometrically
calculated to be 0.1 M equiv because of the use of
1.1 M equiv of TMG in our experiments. Thus, the reaction
mixture containing an adduct intermediate was treated with
10 M extra of acetic acid (1.1 M equiv) under the same
conditions. However, incomplete fragmentation was
observed even after 0.5 h,s in contrast to smooth conversion
with Ac2O (0.3 h), strongly indicating that an acid
anhydride is a real initiator for the fragmentation reaction.
This deduction is supported by a comparable experiment
using pivalic acid (2.2 M equiv), in place of pivalic
anhydride (run 4 in Table 2), as an initiator, in which
longer reaction time (24 h vs 14. 5 h) is needed for the
completion of fragmentation.** Furthermore, low conver-
sion to aziridine products†† in the presence of BF3 Et2O
(2.2 M equiv) even after 41 h suggests that a Lewis acid is
not an effective initiator for the fragmentation reaction, too.
Aziridine could also be obtained when changing the solvent
from chloroform to THF (runs 1 vs 6 and 4 vs 8, Table 2);
however, the activity is slightly lowered in THF. In addition
di-t-butyl dicarbonate was found to trigger the fragmen-
tation reaction (run 9, Table 2).
The oxygen atom in the urea molecule should originate from
the aldehyde oxygen of piperonal (3a) when SiO2 is used as
an initiator of fragmentation. However, the possibility of the
urea oxygen to be delivered from the oxygen atom of Ac2O
could not be completely excluded in the treatment of the
intermediate adduct with Ac2O. Therefore we examined the
reaction using diacetyl sulfide (Ac2S) in place of Ac2O, in
which the alternative product will be thiourea. The
¶ Urea was identified by spectral data [dC 161.8 ppm (CO); m/z: 267 (MC
H)C]. Precise examination of the reaction mixture obtained in run 10 in
Table 2 using Ac2S led to the isolation of O-acetyl-b-aminoalcohol 8,
but no sulfur-containing products, even in minute amount (2%). The
same compound 8 was also isolated as a side-product (1%) in the case of
the Ac2O treatment (run 1, Table 2). The corresponding N-acetylated
product was not obtained.
s The reaction was completed when treated with an aditional amount of
AcOH (1.1 M equiv; total 2.2 M equiv) for further 0.5 h, affording
trans-6a [81% (83%ee)] and cis-6a (7%), respectively.
** trans-6a [88% (92% ee)] and cis-6a (8%) were obtained.
†† trans-6a [17% (84% ee)] and cis-6a [23% (89% ee)] were obtained.
§ For example, although p-nitroacetophenone produced an intermediate
adduct, reverse reaction to the starting ketone was observed on TLC
when treated with Ac2O for 4 days at room temperature. Aziridine was
obtained in only 4% yield and the ketone was recovered in 57% yield.

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
2861
Table 3. Reactions of 1 and various p-substituted benzaldehydes (3b–3j) under the new conditions
Run
X in 3
Time (h)
in step 2
Total 6
(%)
trans-6
(%)a (ee)b
cis-6
(%)a (ee)b
Ratio of
trans-6 (%)c
7 (%)a
3 (%)a
d
1
2
3
4
5
6
7
8
9
NMe2 (3b)
OnBu (3c)
OMe (3d)
CH3 (3e)
H (3f)
0.5
0.5
0.5
1.5
3
—
—
—
3 (–)e
—
96
95
41
28
36
35
66
59
—
81
65
90
100
f
—
69
50
67
95
29
12
None
None
None
None
None
None
67
81
76
80
92
80
53
70
64 (92)
77 (91)
31 (93)
22 (88)
33 (84)
28 (72)
35 (32)
41 (11)
4 (–)e
45 (90)
58 (86)
59 (86)
52 (79)
18 (16)
29 (10)
Cl (3g)
17
CO2Me (3h)
CN (3i)
NO2 (3j)
17.5
17.5
49
a Isolated yield.
b Determined by chiral HPLC.
c (trans-6/total 6)!100.
d No aziridine product is formed even with an excess of 3b (3.5 M equiv). The same result is obtained when p-diethylaminobenzaldehyde is used.
e Not measured.
f
Not purified.
Thus, next we investigated, under the modified conditions
using the most reactive Ac2O, the effect on the aziridine
synthesis of various substituents in para position of the
benzaldehyde.
reaction rate of step 2; slow reactions in runs 6 and 7 (about
17 h) compared to runs 4 and 5 (!3 h). On the other hand in
the cases of p-cyano- (3i) and p-nitrobenzaldehydes (3j), the
asymmetric inductions of aziridine products are greatly
lowered and, in addition, the predominant formation of
trans-6 is again observed (runs 8 and 9, Table 3).
Therefore, these aziridine formation reactions could be
categorized into four groups based on the nature of the
aldehyde substrates: (i) group A, which includes two
benzaldehydes 3c and 3d with strong EDG, is characterized
by a very slow step 1, a very rapid step 2, and excellent
diastereoselectivity and enantioselectivity of trans-aziri-
dine; (ii) group B is represented by a slow step 1 and a rapid
step 2 with moderate cis-diastereoselectivity and excellent
to good enantioselectivity in both aziridine products, and
contains benzaldehydes 3e and 3f which possess a weak
EDG or no substituents; (iii) group C shows a similar profile
as in group B but with a rapid step 1 and a slow step 2, and
comprises benzaldehydes 3g and 3h with a weak EWG, and
(iv) group D, which includes 3i and 3j with a strong EWG, is
characterized by a moderate trans-diastereoselectivity and
low enantioselectivity with a very rapid step 1 and a very
slow step 2.
2.3. Effect of a substituent of aryl aldehydes
After stirring a mixture of the (S,S)-guanidinium salt 1 and a
benzaldehyde derivative 3 in THF in the presence of TMG
at 25 8C for 5 h (step 1), a solution of the intermediate
adduct in chloroform is treated with Ac2O at room
temperature until disappearance of the polar intermediate
on TLC (step 2). The results are summarized in Table 3.
No reaction occurs when p-dimethylaminobenzaldehyde
(3b) is used, even when an excess (3.5 M equiv) of aldehyde
is added (run 1, Table 3). Although in the cases of the
alkoxy-substituted benzaldehydes 3c and 3d the starting
aldehydes remained in step 1, the intermediate adducts were
immediately transformed in step 2 (runs 2 and 3, Table 3).
Thus, aziridine products 6c and 6d were obtained in 67 and
81% yields together with the recovery of the starting
benzaldehydes 3c and 3d in 29 and 12% yields, respect-
ively, in which the trans-derivatives are produced as major
aziridine isomers with excellent stereoselectivity in both
ee and de. p-Tolualdehyde (3e) and benzaldehyde (3f)
smoothly afforded aziridine products 6e and 6f with high
asymmetric inductions (runs 4 and 5, Table 3). Interestingly,
in those cases, the major isomers are the cis-derivatives,
instead of the trans. The nearly same results are obtained in
runs 6 and 7 when p-chloro- (3g) and p-methoxycarbonyl-
benzaldehydes (3h) are used. The main difference is the
2.4. Reaction mechanisms
We have reported in a preliminary communication3 that
(S,S)-guanidinium salt 1 produced (2R,3R)- and (2R,3S)-3-
phenylaziridine-2-carboxylates as cis- and trans-aziridines
after SiO2 treatment of the intermediate adduct derived from
benzaldehyde 3f. This stereochemical result can be applied
to the Ac2O-triggered reaction because of identical asym-
metric induction in both reactions.‡ The nearly same
enantioselectivity in both cis- and trans-aziridines (see
Table 3) indicates that the stereochemistry of products
should be strictly controlled in the first C–C bond formation
stage (step 1) because of the absence of epimerization of the
stereogenic centers between the isomeric aziridines.3 As

2862
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
It is reasonable to expect that Ac2O could coordinate the
oxygen atom of the oxazolidine moiety in 4 because of the
isolation of O-acetyl-b-aminoalcohol 8 as a side product.¶
The partially cationic character of the oxygen atom in the
coordinated species could initiate the fragmentation of the
intermediate adduct to trans-6 and a 2-acetoxyamidinium
salt 9,6 convertible to urea 7, by simultaneous O–C(Ar)
bond cleavage and the N(oxazolidine)–C(Ar) bond forma-
tion, in the course of which partially-generated benzyl
cationic species could be stabilized by electron-relayed
participation of the p-substituted alkoxy group. As men-
tioned in the fragmentation of the piperonal-derived
intermediate adduct using various acid anhydrides (see
Table 2), the reaction is sensitive to steric demand. Efficient
synthesis of trans-aziridines with excellent stereoselectivity
is due to a judicious combination of activation with a more
sterically demanding Ac2O and the electronic stabilization
by the EDG, the former activation acting as a rate-
determining step in the fragmentation.‡‡
Group C carrying a weak EWG shows moderate diastereo-
selectivity, but with good enantioselectivity, similar to
group B either with a weak EDG or without substituents. For
the sake of convenience on discussion, the mechanism of
group C (Scheme 5) is considered prior to group B. As
mentioned above, the major difference between group B and
C is the reaction time needed for the completion of the
fragmentation (step 2), a longer time is needed in group C.
Since steric environment is all identical around the reaction
sites, this fact suggests a different rate-determining step
during the fragmentation sequences.
In group C, rapid reaction in step 1 allows an alternative
diastereomeric Re(G)–Si(A) approach in addition to a major
Re(G)–Re(A) approach, resulting in the lowered face
selectivity. The oxygen atom of the oxazolidine unit in a
diastereomeric mixture of spiro intermediates 4 is similarly
activated with Ac2O; however, the benzyl cationic species
supposed in Scheme 4 could not be formed because of the
EWG destabilization. Therefore, in this case the cleavage of
the C(spiro)–N(oxazolidine) bond involving the nitrogen
atom of the imidazolidine unit affords an opened amidinium
intermediate 10, in which an intramolecular SN2 type
reaction occurs, thus, resulting in the predominant pro-
duction of cis-(2R,3R)-aziridine from a major trans-
(4R,5S)-axazolidine spiro species 4 with inversion of the
configuration at C-5. This substitution reaction is relatively
slow because of a secondary substrate; however, the
presence of an EWG accelerates the SN2 type reaction.
The slightly lowered ee of aziridine products could be
explained by competitively enantiofacial accesses, such as
the Si(G)–Re(A) approach, in rapid reaction of step 1.
Scheme 4. Mechanistic consideration for the aziridine production in group
A.
mentioned above, results in Tables 1 and 2 show that the
overall reaction, not only step 1 but also step 2, is controlled
by kinetic mode. Thus, we based the reaction mechanisms
for the aziridine formation on the four groups categorized
above.
Nearly exclusive formation of trans-(2R,3S)-6 with excel-
lent enantioselectivity in group A with alkoxy-substituents
could be explained by the mechanism shown in Scheme 4.
An aldehyde approaches to the less hindered Re-face
[Re(G)] of the ylide enolate because the Si-face is blocked
by the N-Me group located in the stable Z-configuration of
the CaNC double bond. Thus, R stereochemistry of the
stereogenic center in the C–C bond formation (the C-2 of
aziridines) should be induced. The step 1 is very slow due to
the lowered electrophilicity of the carbonyl function of
alkoxy p-substituted aldehydes, thus, the less hindered Re-
face [Re(A)] of the aldehyde greatly participates to the bond
formation. These situations lead to the production of a
sterically stable (4R,5S)-oxazolidine spiro species 4 with a
trans-configuration between the ester and the aryl groups.
Then, very rapid concerted, SNi type, fragmentation can
occur with retention at the C-5 benzylic carbon to give
trans-(2R,3S)-aziridine.
Excellent to good ee observed in group B are explained by
the high enantiofacial selectivity of the Re(G)–Re(A)
approach in slow reaction of step 1 as observed in group
A (see Scheme 4); however, the lowered diastereoselectivity
could not be covered by simple application of the strong
‡‡ Regeneration of Ac2O in the final step suggests the possibility of Ac2O
as a catalyst. Actual use of an excess of Ac2O may be caused by its
competitive coordination to other species containing basic nitrogen
functions.

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
2863
Scheme 5. Mechanistic consideration for the aziridine production in group C.
EDG-controlled mechanism proposed for group A. The
ratio of trans-aziridine diminished in the order of 6d
(95%)[6e (41%)O6f (28%) (see runs 3–5, Table 3),
which corresponds to a decrease in the electron-donating
effect of the p-substituents. The reduction of the electron-
donating effect and the longer reaction time of step 2 in
group B than that in group A allow us to predict the presence
of a SN2 type reaction mechanism. Thus, the contamination
of both SNi and SN2 type reactions causes the lower
diastereoselectivity in group B (Scheme 6). These specu-
lations can be supported by the fact that a change in
mechanism from SN1 to SN2 by electronic effect of
p-substituted benzyl chlorides has been previously
reported.§§ Higher ratio of trans-isomer in the case of
tolualdehyde (3e) compared to benzaldehyde (3f) can be
rationalized by the ability of the methyl substituent to
stabilize a benzyl cationic-like intermediate in the SNi type
reaction as an EDG.
In group D with a strong EWG not only the great reduction
of stereoselectivity in aziridine production but also the
unexpected formation of the trans-aziridine can be
explained by the mechanism shown in Scheme 7. In a
very rapid step 1, four competitive approaches produce all
possible isomers with slight differences. On the other hand,
in a very slow step 2 amidinium intermediates 10, substrates
of the SN2 type reaction, are in equilibrium to spiro
intermediates 4, different from the three reactions
mentioned above, and a more sterically-favored anti
conformation 10 (erythro) derived from a less stable
§§ It has been reported that a change in mechanism from SN1 to SN2 on a
benzylic carbon by electronic substituent effects; see, Swain, C. G.;
Langsdorf, W. P., Jr. J. Am. Chem. Soc. 1951, 73, 2813–2819.
Scheme 6. Mechanistic consideration for the aziridine production in
group B.

2864
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
Scheme 7. Mechanistic consideration for the aziridine formation in group D.
cis-oxazolidine preferably act as the substrate for SN2 type
reaction to give trans-aziridine.
2.5. Hammett studies
As mentioned above, it is clear that substituents on
benzaldehydes 3 play a crucial role on the overall reaction
of aziridine formation, especially on stereochemical control
of the final products. To elucidate the proposed reaction
mechanisms, we undertook a Hammett analysis of the
diastereoselectivity in aziridine products with substituent
constant (sp).7 No equilibrium between cis- and trans-
aziridines allows us to assume that the diastereomeric ratio
(drZytrans/ycis) of aziridines reflects the net ratio of their
formation rates through the SNi and SN2 type reaction
pathways. Since the trans-aziridine is mainly afforded as
major isomer when conversion rate in step 2 is rapid, the
yield of trans-aziridine (ytrans) is adopted as the numerator
of the dr. The logarithm of the dr [log(dr)] was used for
these plots because it is directly correlated to the free energy
difference between the transition states of the above
pathways.8
Figure 1. Hammett plot of log(dr) vs sp for p-substituents of benzalde-
hydes 3.

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
2865
The plot of log(dr) versus the corresponding sp reveals two
natural groupings depending upon the substituents on the
benzaldehyde substrates 3 (Fig. 1). The lines obtained show
a good correlation (R2Z0.911 for EDGs, R2Z0.800 for
EWGs) in electronically analogous categories with opposite
slopes in sign. Within the EDGs, the log(dr) decreases as the
electron-donating effect decreases (the value of sp
increases), giving a negative slope line with relatively
large magnitude (K10.820). The large negative magnitude
indicates that the EDG-substituents stabilize a sizable
positive charge at the benzylic position9 developed by
the initial coordination of Ac2O to the oxygen atom of the
oxazolidine ring in a spiro intermediate adduct 4 in the
transition state of the rate-determining step and leads to high
diastereoselectivity in the aziridine products. Thus, the SNi
type mechanism proposed for the EDG-participated aziri-
dine formation as shown in Scheme 4 can be reasonably
supported.
Figure 2. Hammett plot of log(er) in trans-6 versus sp for p-substituted
benzaldehydes 3.
latter case is more electronically controlled, as expected
from the low face-selectivity at the rapid addition reaction
with reactive aldehydes.
A line with negative slope of relatively small magnitude,
leading to a plotting curve with concave upwards,9 will be
expected for the longer reaction time needed for completion
in step 2 as the electron-withdrawing ability of the EWGs
increases (the value of sp also increases). Actually, the
log(dr) gradually increases to give a line with positive slope
of relatively small magnitude (C0.871). The positive slope
can be accounted for the SN2 type mechanism proposed for
the EWG-participated aziridine formation, as shown in
Scheme 5, in which the generation of cis- and trans-
aziridines is controlled by diastereofacial selectivity in the
formation of spiro intermediate adduct 4 (step 1).
It should be noted here that the shift of the inflection point,
(C0.401 on sp) in Figure 2 compared to (K0.123 on sp) in
Figure 1, clearly reveals that the C–C bond formation in step
1 and the intramolecular nucleophilic substitution in step
2 are independently influenced by electronic effect of the
p-substitutents of the benzaldehyde substrates.
3. Conclusions
Among the substituents tested, the increase in the electron-
withdrawing effect led a gradual diminishment of the ee in
the cases of the benzaldehydes 3e–h with either a weak
EDG or a weak EWG and then a dramatic diminishment in
the case of benzaldehydes 3i–j with a strong EWG. This
trend allows us to consider another Hammett analysis of the
enantioselectivity of the aziridine products to sp.7 The
enantiomeric ratio (er)¶¶ of trans-6 will reflect the net ratio
of formation rates between the (2R,3S)- and (2S,3R)-
aziridines through the possible four approaches in step 1.
The logarithm of er [log(er)] is used for these plots because
it relates directly to the free energy differences between the
diastereomeric transition states.8 The plot of the log(er)
versus the corresponding sp leads to two groups of data, a
left-hand and a right-hand side (Fig. 2).
Based on the Ac2O treatment of the intermediate adduct,
two kinds of reaction mechanisms controlled by the nature
of the p-substituents of aryl aldehydes are proposed for
asymmetric aziridine synthesis from guanidinium ylide and
aryl aldehydes. A SNi-like mechanism via cationic-like
transition state is proposed for the asymmetric synthesis
using EDG-substituted benzaldehydes, whereas with EWG-
substituted benzaldehydes, a SN2-like mechanism is
proposed. Since the SiO2 treatment of some intermediate
adducts show identical diastereo- and enantioselectivity that
the Ac2O treatment, we can assume that the SiO2 treatment
proceeds as well through these two mechanisms by
coordination of SiO2 to the oxazolidine-oxygen atom in
the spiro intermediate adduct. Hammett analysis, based on
the diastereomeric ratio in the aziridine products, is
consistent with the proposed rate-determining steps in
these two mechanisms. A second Hammett analysis, based
on the enantiomeric ratio of the aziridine products, clearly
reveals the difference in the susceptibilities to the electronic
substituents effect between the nucleophilic addition (step
1) and the intramolecular nucleophilic substitution (step 2).
The lines obtained show a good-to-excellent correlation
(R2Z0.923 for the left-hand side, R2Z0.991 for the right-
hand side) with slightly different negative slopes. For the
left-hand side of the plot, the log(er) gradually decreases as
the value of sp increases, giving a line with a negative slope
of relatively small magnitude (K0.764), whereas for the
right-hand side of the plot, the log(er) decreases dramati-
cally as the electron-withdrawing ability of the substituents
became stronger (the value of sp increases), giving a line
with negative slope of relatively large magnitude (K2.098).
These facts indicate that the C–C bond formation step in the
We have recognized that vinyl aldehyde derivatives can as
well act as electrophiles.ss Further studies on the scope and
limitation of this aziridine synthesis and its application to
the synthesis of bioactive nitrogen-containing compounds
are at present under way in our laboratory.
¶¶ The enantiomeric ratio (er) is defined here as the relative amount of the
major enantiomer divided by the relative amount of the minor
enantiomer (e.g., 80% ee gives an er of 9.0); see Ref. 8.
ss These works will be published elsewhere in the near future.

2866
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
4. Experimental
4.2.1. 1-Benzyl-2-tert-butoxycarbonyl-3-(3,4-methylene-
dioxyphenyl)aziridine (6a). trans-6a. 1H NMR d 7.36 (2H,
d, JZ7.1 Hz, Ar-H), 7.30–7.19 (3H, m, Ar-H), 6.79 (1H, dd,
JZ7.9, 1.5 Hz, Ar-H), 6.75 (1H, s, Ar-H), 6.73 (1H, d, JZ
7.9 Hz, Ar-H), 5.92 (2H, s, OCH2O), 4.22 (1H, d, JZ
13.9 Hz, CHaHbPh), 4.06 (1H, d, JZ13.9 Hz, CHaHbPh),
3.19 (1H, brs, C3–H), 2.64 (1H, d, JZ2.2 Hz, C2–H), 1.39
(9H, s, OC(CH3)3); 13C NMR d 167.5, 147.6, 146.8, 139.2,
132.4, 128.0, 127.9, 126.6, 119.8, 107.9, 106.2, 100.8, 81.6,
54.5, 47.9, 45.5, 28.0. Anal. Calcd for C21H22NO4 C, 71.37;
H, 6.56; N, 3.96. Found: C, 71.04; H, 6.55; N, 3.92; HPLC
analysis: Daicel Chiralcel OD, mobile phase: hexane/
2-propanolZ100:1, flow rateZ1.0 mL/min, detection
254 nm, retention time: 6.3 min (minor isomer) and
7.1 min (major isomer).
4.1. General
1H (400 MHz) and 13C NMR (100 MHz) spectra were
measured in CDCl3 on JEOL JNM-ALPHA 400 or JEOL
JNM-AL 400 spectrometers. HPLC was performed on
Shimadzu 10A instruments using Daicel Chiralcel OD or
OD-H and Chiralpak AD-H columns (4.6 mm!25 cm). For
TLC, Merck precoated TLC plates (silica gel 60 F254) were
used. The products were purified by column chromatog-
raphy on silica gel (FL-100D; Fuji Silysia Chemical Ltd.).
Low (FABMS) and high-resolution FABMS (HRFABMS)
were performed on JEOL JMS-HX 110A and JMS-700-T
spectrometers. Dry tetrahydrofuran (THF) was purchased
from Kanto Chemical Co. Inc. as ‘Dehydrated’. 1,1,3,3-
Tetramethylguanidine (TMG) was distilled under reduced
pressure prior to use.
cis-6a. 1H NMR d 7.41 (2H, d, JZ7.3 Hz, Ar-H), 7.34–7.30
(2H, m, Ar-H), 7.26–7.23 (1H, m, Ar-H), 6.92 (1H, d, JZ
1.5 Hz, Ar-H), 6.85 (1H, dd, JZ8.1, 1.5 Hz, Ar-H), 6.70
(1H, d, JZ8.1 Hz, Ar-H), 5.90 (2H, s, OCH2O), 3.94 (1H, d,
JZ13.9 Hz, CHaHbPh), 3.54 (1H, d, JZ13.9 Hz, CHaHb-
Ph), 2.92 (1H, d, JZ6.8 Hz, C3–H), 2.48 (1H, d, JZ6.8 Hz,
C2–H), 1.24 (9H, s, OC(CH3)3); HPLC analysis: Daicel
Chiralpak AD-H, mobile phase: hexane/2-propanolZ50:1,
flow rateZ1.0 mL/min, detection 254 nm, retention time:
15.5 min (major isomer) and 17.9 min (minor isomer).
4.2. General procedure for the formation of chiral
aziridines
Treatment with SiO2. A solution of TMG (25 mL,
0.202 mM) in THF (35 mL) was dropwise added to a
solution of the (S,S)-guanidinium salt 1 (101 mg,
0.184 mM) and a benzaldehyde 3 (0.180 mM) in THF
(365 mL) at 25 8C. The mixture was stirred at 25 8C for
4–5 h under nitrogen atmosphere and then SiO2 (3.0 g) was
added after dilution with CHCl3 (10 mL). The resulting
slurry was stirred at room temperature until disappearance
of a highly polar intermediate on TLC (ethyl acetate/
hexane). After removal of the SiO2 by filtration, filter cakes
were successively washed with CHCl3 (36 mL) and CH3CN
(10 mL). The filtrate and the washings were combined.
After evaporation, the residue was purified by column
chromatography on SiO2 (ethyl acetate/hexane as eluent) to
give trans- trans-6 and cis-aziridines cis-6.
4.2.2. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-butoxyphe-
nyl)aziridine (6c) (run 2 in Table 2). trans-6c. (64%
1
yield, 92% ee); H NMR d 7.37 (2H, d, JZ7.3 Hz, Ar-H),
7.30–7.18 (5H, m, Ar-H), 6.82 (2H, d, JZ8.5 Hz, Ar-H),
4.24 (1H, d, JZ13.9 Hz, CHaHbPh), 4.07 (1H, d, JZ
13.9 Hz, CHaHbPh), 3.94 (2H, t, JZ6.5 Hz, CH2), 3.20 (1H,
brs, C3–H), 2.67 (1H, d, JZ2.2 Hz, C2–H), 1.78–1.71 (2H,
m, CH2), 1.52–1.43 (2H, m, CH2), 1.39 (9H, s, OC(CH3)3),
0.97 (3H, t, JZ7.3 Hz, CH3); 13C NMR d 167.8, 158.6,
139.4, 130.3, 128.1, 127.9, 127.3, 126.6, 114.4, 81.6, 67.8,
54.6, 47.8, 45.5, 31.3, 28.0, 19.3, 13.9; HRFABMS m/z
382.2405, calcd for C24H32NO3 382.2382; HPLC analysis:
Daicel Chiralcel OD-H, mobile phase: hexane/2-propanolZ
400:1, flow rateZ1.0 mL/min, detection 254 nm, retention
time: 18.7 min (minor isomer) and 20.6 min (major isomer).
Treatment with carboxylic anhydrides. A solution of TMG
(25 mL, 0.202 mM) in THF (35 mL) was dropwise added to a
solution of 1 (101 mg, 0.184 mM) and a benzaldehyde 3
(0.180 mM) in THF (365 mL) at 25 8C. The mixture was
stirred at 25 8C for 5 h under nitrogen atmosphere. After
evaporation of THF with a stream of nitrogen, CHCl3
(10 mL) was added. After addition of carboxylic anhydride
(0.405 mM) the mixture was stirred at room temperature
until disappearance of a highly polar intermediate on TLC.
Evaporation of the solvent followed by purification by
column chromatography on SiO2 (ethyl acetate/hexane as
eluent) afforded trans-6 and cis-6.
1
cis-6c. (2.8% yield); H NMR d 7.42 (2H, d, JZ7.3 Hz,
Ar-H), 7.32–7.22 (5H, m, Ar-H), 6.79 (2H, d, JZ8.8 Hz,
Ar-H), 3.94 (1H, d, JZ13.9 Hz, CHaHbPh), 3.92 (2H, t, JZ
6.6 Hz, CH2), 3.57 (1H, d, JZ13.9 Hz, CHaHbPh), 2.94
(1H, d, JZ6.8 Hz, C3–H), 2.48 (1H, d, JZ7.1 Hz, C2–H),
1.77–1.70 (2H, m, CH2), 1.56–1.42 (2H, m, CH2), 1.20 (9H,
s, OC(CH3)3), 0.94 (3H, t, JZ7.3 Hz, CH3).
Treatment with pivalic acid. A highly polar intermediate
was prepared from 1 and piperonal (3a) in the same manner
described above. A solution of pivalic acid (41.3 mg,
0.405 mM) in CHCl3 (1.0 mL) was added to a solution of
the intermediate in CHCl3 (9.0 mL). The mixture was
stirred at room temperature for 24 h, washed with saturated
NaHCO3 (3.0 mL) and the upper aqueous layer was
re-extracted with CHCl3 (2.0 mL!4). The combined
organic layer was dried over MgSO4 and concentrated.
The residue was purified by column chromatography on
SiO2 using 3–20% ethyl acetate in hexane to give trans-6a
and cis-6a and urea 7.
4.2.3. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-methoxy-
phenyl)aziridine (6d) (run 3 in Table 2). trans-6d. (77%
yield, 91% ee); H NMR d 7.37 (2H, d, JZ7.3 Hz, Ar-H),
1
7.30–7.20 (5H, m, Ar-H), 6.84 (2H, d, JZ8.3 Hz, Ar-H),
4.24 (1H, d, JZ13.9 Hz, CHaHbPh), 4.08 (1H, d, JZ
13.9 Hz, CHaHbPh), 3.79 (3H, s, OCH3), 3.21 (1H, brs, C3–
H), 2.67 (1H, d, JZ2.2 Hz, C2–H), 1.39 (9H, s, OC(CH3)3);
13C NMR d 167.7, 159.0, 139.3, 130.5, 128.1, 127.9, 127.3,
126.7, 113.7, 81.6, 55.3, 54.6, 47.8, 45.5, 28.0; HRFABMS
m/z 340.1885, calcd for C21H26NO3 340.1913; HPLC
analysis: Daicel Chiralcel OD-H, mobile phase: hexane/
2-propanolZ100:1, flow rateZ0.5 mL/min, detection

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
2867
254 nm, retention time: 14.1 min (minor isomer) and
15.2 min (major isomer).
4.2.6. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-chlorophenyl)-
aziridine (6g) (run 6 in Table 2). trans-6g. (33% yield,
84% ee); H NMR d 7.36 (2H, d, JZ7.3 Hz, Ar-H), 7.31–
1
1
cis-6d. (3.8% yield); H NMR d 7.43 (2H, d, JZ7.6 Hz,
7.21 (7H, m, Ar-H), 4.23 (1H, d, JZ14.0 Hz, CHaHbPh),
4.08 (1H, d, JZ14.0 Hz, CHaHbPh), 3.22 (1H, brs, C3–H),
Ar-H), 7.33–7.21 (5H, m, Ar-H), 6.80 (2H, d, JZ8.8 Hz,
Ar-H), 3.96 (1H, d, JZ13.9 Hz, CHaHbPh), 3.77 (3H, s,
OCH3), 3.56 (1H, d, JZ13.9 Hz, CHaHbPh), 2.95 (1H, d,
JZ6.8 Hz, C3–H), 2.49 (1H, d, JZ6.8 Hz, C2–H), 1.20 (9H,
s, OC(CH3)3).
2.67 (1H, d, JZ2.7 Hz, C2–H), 1.40 (9H, s, OC(CH3)3); 13
C
NMR d 167.3, 139.1, 137.0, 133.1, 128.4, 128.2, 127.9,
127.6, 126.8, 81.9, 54.6, 47.2, 45.9, 28.0; HRFABMS m/z
344.1406, calcd for C20H23ClNO2: 344.1417; HPLC
analysis: Daicel Chiralcel OD-H, mobile phase: hexane/
2-propanolZ300:1, flow rateZ0.3 mL/min, detection
254 nm, retention time: 24.6 min (minor isomer) and
26.4 min (major isomer).
4.2.4. 1-Benzyl-2-tert-butoxycarbonyl-3-(p-tolyl)aziri-
dine (6e) (run 4 in Table 2). trans-6e. (31% yield, 93%
1
ee); H NMR d 7.37 (2H, d, JZ7.3 Hz, Ar-H), 7.31–7.18
(5H, m, Ar-H), 7.11 (2H, d, JZ7.6 Hz, Ar-H), 4.24 (1H, d,
JZ13.9 Hz, CHaHbPh), 4.08 (1H, d, JZ13.9 Hz, CHaHb-
Ph), 3.22 (1H, brs, C3–H), 2.69 (1H, d, JZ2.7 Hz, C2–H),
2.32 (3H, s, CH3), 1.39 (9H, s, OC(CH3)3); 13C NMR d
167.7, 139.3, 137.0, 135.4, 128.9, 128.1, 127.9, 126.7,
126.1, 81.6, 54.7, 48.0, 45.7, 28.0, 21.2. HRFABMS m/z
324.1945, calcd for C21H26NO2 324.1964. HPLC analysis:
Daicel Chiralcel OD-H, mobile phase: hexane/2-propanolZ
1000:0.5, flow rateZ1.0 mL/min, detection 254 nm, reten-
tion time: 22.8 min (minor isomer) and 25.3 min (major
isomer).
1
cis-6g. (59% yield, 86% ee); H NMR d 7.41 (2H, d, JZ
7.3 Hz, Ar-H), 7.35–7.22 (7H, m, Ar-H), 3.97 (1H, d, JZ
13.9 Hz, CHaHbPh), 3.56 (1H, d, JZ13.9 Hz, CHaHbPh),
2.95 (1H, d, JZ6.8 Hz, C3–H), 2.54 (1H, d, JZ6.8 Hz, C2–
H), 1.20 (9H, s, OC(CH3)3); 13C NMR d 166.7, 137.7, 133.8,
132.9, 129.2, 128.3, 127.8, 127.7, 127.1, 81.3, 63.2, 46.9,
46.5, 27.8. Anal. Calcd for C20H22ClNO2: C, 69.86; H, 6.45;
N, 4.07. Found: C, 69.52; H, 6.44; N, 4.00; HPLC analysis:
Daicel Chiralcel OD-H, mobile phase: hexane/2-propanolZ
500:1, flow rateZ1.0 mL/min, detection 254 nm, retention
time: 27.7 min (major isomer) and 30.4 min (minor isomer).
1
cis-6e. (45% yield, 90% ee); H NMR d 7.43 (2H, d, JZ
7.6 Hz, Ar-H), 7.43–7.24 (5H, m, Ar-H), 7.06 (2H, d, JZ
7.8 Hz, Ar-H), 3.96 (1H, d, JZ13.9 Hz, CHaHbPh), 3.56
(1H, d, JZ13.9 Hz, CHaHbPh), 2.96 (1H, d, JZ7.0 Hz,
C3–H), 2.50 (1H, d, JZ7.0 Hz, C2–H), 2.30 (3H, s, CH3),
1.19 (9H, s, OC(CH3)3); 13C NMR d 167.1, 137.9, 136.6,
132.0, 128.2, 128.1, 127.6, 126.9, 81.0, 63.3, 47.1, 46.7,
27.7, 21.1. HRFABMS m/z 324.1970, calcd for C21H26NO2
324.1964. HPLC analysis: Daicel Chiralcel OD-H, mobile
phase: hexane/2-propanolZ100:1, flow rateZ1.0 mL/min,
detection 254 nm, retention time: 8.7 min (minor isomer)
and 11.0 min (major isomer).
4.2.7. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-methoxycar-
bonylphenyl)aziridine (6h) (run 7 in Table 2). trans-6h.
(28% yield, 72% ee); H NMR d 7.97 (2H, d, JZ8.1 Hz,
1
Ar-H), 7.37 (2H, d, JZ8.1 Hz, Ar-H), 7.38–7.21 (5H, m,
Ar-H), 4.25 (1H, d, JZ13.8 Hz, CHaHbPh), 4.10 (1H, d, JZ
13.8 Hz, CHaHbPh), 3.90 (3H, s, COOCH3), 3.29 (1H, brs,
C3–H), 2.73 (1H, d, JZ2.2 Hz, C2–H), 1.40 (9H, s,
OC(CH3)3); 13C NMR d 167.2, 166.8, 143.8, 139.0, 129.6,
129.2, 128.2, 127.9, 126.8, 126.2, 82.0, 54.6, 52.1, 47.5,
46.2, 28.0; HRFABMS m/z 368.1852, calcd for C22H26NO4:
368.1862; HPLC analysis: Daicel Chiralcel OD-H, mobile
phase: hexane/2-propanolZ100:1, flow rateZ1.0 mL/min,
detection 254 nm, retention time: 7.6 min (minor isomer)
and 8.9 min (major isomer).
4.2.5. 1-Benzyl-2-tert-butoxycarbonyl-3-phenylaziridine
(6f) (run 5 in Table 2). trans-6f. (22% yield, 88% ee); H
1
NMR d 7.38 (2H, d, JZ7.3 Hz, Ar-H), 7.31–7.20 (9H, m,
Ar-H), 4.26 (1H, d, JZ14.2 Hz, CHaHbPh), 4.10 (1H, d, JZ
14.2 Hz, CHaHbPh), 3.26 (1H, brs, C3–H), 2.72 (1H, d, JZ
2.7 Hz, C2–H), 1.39 (9H, s, OC(CH3)3); 13C NMR d 167.6,
139.3, 138.4, 128.3, 128.1, 127.9, 127.4, 126.7, 126.2, 81.7,
54.6, 48.1, 45.8, 28.0. Anal. Calcd for C20H23NO2: C, 77.64;
H, 7.49; N, 4.53. Found: C, 77.23; H, 7.51; N, 4.45; HPLC
analysis: Daicel Chiralcel OD-H, mobile phase: hexane/
2-propanolZ250:1, flow rateZ0.5 mL/min, detection
254 nm, retention time: 13.7 min (minor isomer) and
14.5 min (major isomer).
1
cis-6h. (52% yield, 79% ee); H NMR d 7.94 (2H, d, JZ
8.5 Hz, Ar-H), 7.48 (2H, d, JZ8.1 Hz, Ar-H), 7.43 (2H, d,
JZ7.3 Hz, Ar-H), 7.35–7.24 (3H, m, Ar-H), 3.98 (1H, d,
JZ13.8 Hz, CHaHbPh), 3.89 (3H, s, COOCH3), 3.60 (1H, d,
JZ13.8 Hz, CHaHbPh), 3.02 (1H, d, JZ7.1 Hz, C3–H),
2.59 (1H, d, JZ7.1 Hz, C2–H), 1.16 (9H, s, OC(CH3)3); 13
C
NMR d 166.9, 166.6, 140.6, 137.7, 129.0, 128.9, 128.3,
127.9, 127.8, 127.2, 81.4, 63.2, 52.0, 47.1, 46.8, 27.8;
HRFABMS m/z 368.1870, calcd for C22H26NO4: 368.1862;
HPLC analysis: Daicel Chiralcel OD-H, mobile phase:
hexane/2-propanolZ100:1, flow rateZ1.0 mL/min, detec-
tion 254 nm, retention time: 21.4 min (major isomer) and
25.3 min (minor isomer).
1
cis-6f. (58% yield, 86% ee); H NMR d 7.46–7.39 (4H, m,
Ar-H), 7.34–7.19 (6H, m, Ar-H), 3.97 (1H, d, JZ13.8 Hz,
CHaHbPh), 3.59 (1H, d, JZ13.8 Hz, CHaHbPh), 3.00 (1H,
d, JZ7.0 Hz, C3–H), 2.53 (1H, d, JZ7.0 Hz, C2–H), 1.16
(9H, s, OC(CH3)3); 13C NMR d 167.0, 137.9, 135.2, 128.3,
127.8, 127.6, 127.1, 127.0, 81.1, 63.3, 47.2, 46.8, 27.8.
Anal. Calcd for C20H23NO2: C, 77.64; H, 7.49; N, 4.53.
Found: C, 77.69; H, 7.57; N, 4.62; HPLC analysis: Daicel
Chiralcel OD-H, mobile phase: hexane/2-propanolZ100:1,
flow rateZ1.0 mL/min, detection 254 nm, retention time:
10.1 min (minor isomer) and 11.9 min (major isomer).
4.2.8. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-cyanophenyl)-
aziridine (6i) (run 8 in Table 2). trans-6i. (35% yield, 32%
ee); 1H NMR d 7.58 (2H, d, JZ8.3 Hz, Ar-H), 7.40 (2H, d,
JZ8.3 Hz, Ar-H), 7.37–7.21 (5H, m, Ar-H), 4.23 (1H, d,
JZ13.9 Hz, CHaHbPh), 4.09 (1H, d, JZ13.9 Hz, CHaHb-
Ph), 3.29 (1H, brs, C3–H), 2.70 (1H, d, JZ2.4 Hz, C2–H),
1.40 (9H, s, OC(CH3)3); 13C NMR d 166.8, 144.1, 138.7,
132.1, 128.2, 127.9, 127.0, 118.8, 111.1, 82.3, 54.5, 47.0,

2868
T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
46.5, 28.0; HRFABMS m/z 335.1729, calcd for
C21H23N2O2: 335.1760; HPLC analysis: Daicel Chiralcel
OD-H, mobile phase: hexane/2-propanolZ400:1, flow
rateZ1.0 mL/min, detection 254 nm, retention time:
41.4 min (minor isomer) and 48.6 min (major isomer).
sulfide (run 11 in Table 1). A highly polar intermediate was
prepared from 1 and piperonal (3a) in the same manner
described above. The solvent was evaporated with a stream
of nitrogen and diacetyl sulfide (0.405 mM) was added after
dilution with CHCl3 (10 mL). The mixture was stirred at
room temperature for 30 min, washed with saturated
NaHCO3 (2.5 mL) and brine (2.0 mL), dried over MgSO4,
and concentrated. The residue was purified by column
chromatography on SiO2 using 3–20% ethyl acetate in
hexane followed by preparative TLC (ethyl acetate/
hexaneZ1:4 as eluent) to give 9 (1.6 mg): 1H NMR d
7.30–7.19 (5H, m, Ph), 6.83 (1H, d, JZ1.5 Hz, Ar-H), 6.80
(1H, dd, JZ7.9, 1.5 Hz, Ar-H), 6.74 (1H, d, JZ7.9 Hz,
Ar-H), 5.96 (2H, s, OCH2O), 5.78 (1H, d, JZ7.2 Hz,
CHOCO), 3.82 (1H, d, JZ13.4 Hz, CHaHbPh), 3.63 (1H, d,
JZ13.4 Hz, CHaHbPh), 3.48 (1H, d, JZ7.2 Hz, CHCOO),
2.03 (1H, s, COCH3), 1.45 (9H, s, OC(CH3)3); 13C NMR d
171.3, 169.5, 147.5, 139.5, 131.0, 128.3, 128.2, 127.0,
121.4, 107.9, 107.8, 101.1, 81.8, 75.9, 65.0, 51.9, 28.0, 21.0.
FABMS m/z: 414 (MHC).
1
cis-6i. (18% yield, 16% ee); H NMR d 7.56 (2H, d, JZ
8.3 Hz, Ar-H), 7.52 (2H, d, JZ8.3 Hz, Ar-H), 7.41 (2H, d,
JZ7.3 Hz, Ar-H), 7.35–7.25 (3H, m, Ar-H), 4.01 (1H, d,
JZ13.7 Hz, CHaHbPh), 3.56 (1H, d, JZ13.7 Hz, CHaHb-
Ph), 3.01 (1H, d, JZ7.1 Hz, C3–H), 2.63 (1H, d, JZ7.1 Hz,
C2–H), 1.18 (9H, s, OC(CH3)3); 13C NMR d 166.3, 140.9,
137.4, 131.5, 128.7, 128.4, 127.8, 127.3, 118.9, 110.9, 81.6,
63.1, 47.2, 46.5, 27.8; HRFABMS m/z 335.1758, calcd for
C21H23N2O2: 335.1760; HPLC analysis: Daicel Chiralcel
OD-H, mobile phase: hexane/2-propanolZ100:1, flow
rateZ1.0 mL/min, detection 254 nm, retention time:
26.8 min (major isomer) and 30.8 min (minor isomer).
4.2.9. 1-Benzyl-2-tert-butoxycarbonyl-3-(4-nitrophenyl)-
aziridine (6j) (run 9 in Table 2).. trans-6j. (41% yield,
1
11% ee); H NMR d 8.15 (2H, d, JZ8.8 Hz, Ar-H), 7.45
4.2.11. Isolation of a mixture of adduct intermediates
in the reaction of 1 and p-chlorobenzaldehyde (3g).
The highly polar intermediate was prepared from 1 and
p-chlorobenzaldehyde (3g) in the same manner described
above. Evaporation of solvent and purification of the residue
(214 mg) by short column chromatography on SiO2 (ethyl
acetate/hexaneZ1:5) gave a ca. 5:1 mixture of adduct
intermediates (104 mg) as an oil.
(2H, d, JZ8.8 Hz, Ar-H), 7.36 (2H, d, JZ7.6 Hz, Ar-H),
7.30 (2H, dd, JZ7.6, 7.6 Hz, Ar-H), 7.25–7.22 (1H, m,
Ar-H), 4.24 (1H, d, JZ13.8 Hz, CHaHbPh), 4.10 (1H, d, JZ
13.8 Hz, CHaHbPh), 3.34 (1H, brs, C3–H), 2.73 (1H, d, JZ
2.4 Hz, C2–H), 1.41 (9H, s, OC(CH3)3); 13C NMR d 166.7,
147.2, 146.1, 138.7, 128.3, 127.9, 127.1, 127.0, 123.5, 82.4,
54.5, 46.8, 46.7, 28.0; HRFABMS m/z 355.1627, calcd for
C20H23N2O4: 355.1658; HPLC analysis: Daicel Chiralcel
OD-H, mobile phase: hexane/2-propanolZ400:1, flow
rateZ1.0 mL/min, detection 254 nm, retention time:
37.4 min (minor isomer) and 43.0 min (major isomer).
The major isomer. 1H NMR d 5.18 (1H, d, JZ8.5 Hz), 4.52
(1H, d, JZ15.0 Hz), 4.06 (1H, d, JZ8.1 Hz), 3.90 (1H, d,
JZ15.0 Hz), 3.83 (1H, d, JZ8.1 Hz), 3.75 (1H, d, JZ
8.5 Hz), 2.52 (3H, s), 2.33 (3H, s), 1.16 (9H, s).
1
cis-6j. (29% yield, 10% ee); H NMR d 8.13 (2H, d, JZ
8.8 Hz, Ar-H), 7.57 (2H, d, JZ8.8 Hz, Ar-H), 7.41 (2H, d,
JZ7.3 Hz, Ar-H), 7.35–7.25 (3H, m, Ar-H), 4.03 (1H, d,
JZ13.7 Hz, CHaHbPh), 3.57 (1H, d, JZ13.7 Hz, CHaHb-
Ph), 3.05 (1H, d, JZ6.8 Hz, C3–H), 2.66 (1H, d, JZ6.8 Hz,
C2–H), 1.18 (9H, s, OC(CH3)3); 13C NMR d 166.2, 147.1,
142.9, 137.3, 128.8, 128.4, 127.8, 127.4, 122.9, 81.7, 63.1,
47.3, 46.3, 27.9; HRFABMS m/z 355.1667, calcd for
C20H23N2O4: 355.1658; HPLC analysis: Daicel Chiralcel
OD-H, mobile phase: hexane/2-propanolZ100:1, flow
rateZ1.0 mL/min, detection 254 nm, retention time:
23.2 min (minor isomer) and 25.0 min (major isomer).
1
The minor isomer. H NMR d 5.61 (d, JZ8.4 Hz), 4.46
(d, JZ13.9 Hz), 4.14 (d, JZ7.3 Hz), 4.10 (d, JZ
13.8 Hz), 4.05 (d, JZ7.3 Hz), 3.96 (d, JZ8.4 Hz),
2.72 (3H, s), 2.49 (3H, s), 0.79 (9H, s). Aromatic protons
of both isomers were observed at d 7.46–7.00 as complex
signals.
4.3. Hammett analyses
Diastereomeric (dr) and enantiomeric ratios (er) calculated
based on the data in Table 3 and substituent constants (sp)
are summarized in the following Table 4.
4.2.10. tert-Butyl 3-acetoxy-3-benzo[1,3]dioxol-5-yl-2-
benzylaminopropionate (9): In treatment with diacetyl
Table 4. The dr, er and sp values for Hammett analyses
dr (trans/cis)a
log(dr)
er (major/minor)b
log(er)
sp
c
No.
X in 3
1
2
3
4
5
6
7
8
OnBu
OMe
Me
H
Cl
CO2Me
CN
NO2
22.857
20.184
0.682
0.375
0.555
0.541
1.966
1.443
1.359
1.305
23.415
21.222
26.027
15.393
11.060
6.049
1.369
1.327
1.415
1.187
1.044
0.782
0.290
0.099
K0.32
K0.27
K0.17
0
0.23
0.45
K0.167
K0.425
K0.256
K0.267
0.294
1.950
1.257
0.66
0.78
0.159
a drZ(yield of trans-6)/(yield of cis-6).
b erZ[(the relative amount of the major enantiomer)/(the relative amount of the minor enantiomer)] in trans-6.8
c All substituent constants were taken from Ref. 7.

T. Haga, T. Ishikawa / Tetrahedron 61 (2005) 2857–2869
2869
Asymmetry 1997, 18, 1693–1715. (f) Atkinson, R. S.
Tetrahedron 1999, 55, 1519–1559. (g) Mitchinson, A.; Nadin,
A. J. Chem. Soc., Perkin Trans. 1 2000, 2862–2892.
2. Cardillo, G.; Gentilucci, L.; Tolomelli, A. Aldrichim. Acta
2003, 36, 39–50.
Acknowledgements
This work is supported by a Grant-in-Aid for Scientific
Research (14370717) from the Ministry of Education,
Culture, Sports, Science and Technology, Japan.
3. Hada, K.; Watanabe, T.; Isobe, T.; Ishikawa, T. J. Am. Chem.
Soc. 2001, 123, 7705–7706.
4. Isobe, T.; March, J. March’s Advanced Organic Chemistry 3rd
ed.; Wiley: New York, 2001; pp 1231–1237.
References and notes
5. Smith, M. B.; March, J. March’s Advanced Organic Chemistry
3rd ed.; Wiley: New York, 2001; pp 1247–1248.
6. Smith, M. B.; Ishikawa, T. J. Org. Chem. 1999, 64, 5832–5835
see also pp 6984–6988 and 6989–6992.
1. (a) Aube, J. In Trost, B. M., Fleming, I., Eds.; Comprehensive
Organic Synthesis; Pergamon: Oxford, 1991; Vol. 1, p 835.
(b) Rosen, T. In Trost, B. M., Fleming, I., Eds.; Comprehensive
Organic Synthesis; Pergamon: Oxford, 1991; Vol. 2, p 428.
(c) Kemp, J. E. G. In Trost, B. M., Fleming, I., Eds.;
Comprehensive Organic Synthesis; Pergamon: Oxford, 1991;
Vol. 7, p 469. (d) Tanner, D. Angew. Chem., Int. Ed. Engl. 1994,
33, 599–619. (e) Osborn, H. M. I.; Sweeney, J. Tetrahedron:
7. Hansch, C.; Leo, A.; Taft, R. W. Chem. Rev. 1991, 91, 165–195.
8. Constantine, R. N.; Kim, N.; Bunt, R. C. Org. Lett. 2003, 5,
2279–2282.
9. Sykes, P. A Guidebook to Mechanism in Organic Chemistry, 5th
ed.; Longman: London, 1981; Chapter 13, pp 351–369.