Bimetallic catalysis in the highly enantioselective ring-opening reactions of aziridines

Article abstract of DOI:10.1039/c3sc52929k

Bimetallic yttrium- and lanthanide-salen complexes, readily prepared from commercially available metal isopropoxides, 2-dimethylaminoethanol, 1,1′-binaphthyl-2,2′-diamine and 2-hydroxy-3-methoxybenzaldehyde (3 steps) catalyze highly enantioselective ring opening (～90-99% ee) reactions of meso-N-4-nitrobenzoyl aziridines by TMSCN and TMSN₃. The TMSN ₃-mediated reactions give the highest enantioselectivities reported to date for several prototypical aziridines. Selectivity in a related ring opening by silyl isothiocyanates depends on the substituents on silicon, larger ^tBuPh₂SiNCS giving the best selectivities, especially when an yttrium or ytterbium complex is used as the catalyst. The bimetallic yttrium complex also effects unprecedented regiodivergent parallel kinetic resolution of racemic monosubstitued aziridines upon reaction with TMSN₃. In these reactions, each of the enantiomers undergoes nucleophilic addition of azide at different carbons giving two products in nearly enantiomerically pure form. To explain the dramatic differences in the selectivities between the mono- and bimetallic catalysts in these reactions, a mechanism that involves activation of both the electrophile (aziridine) and the nucleophile (azide or cyanide) at two different metals of the bimetallic complex is proposed. Molecular weight determination by vapor pressure osmometry, diffusion ordered NMR spectroscopy (DOSY) and kinetic studies, which suggest first order dependence on the concentration of the catalyst, provide strong support for this proposal. The Royal Society of Chemistry 2014.

Full text of DOI:10.1039/c3sc52929k

Chemical
Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Bimetallic catalysis in the highly enantioselective
ring–opening reactions of aziridines†
Cite this: Chem. Sci., 2014, 5, 1102
*
*
Bin Wu,‡ Judith C. Gallucci, Jon R. Parquette and T. V. RajanBabu
Bimetallic yttrium- and lanthanide-salen complexes, readily prepared from commercially available
metal isopropoxides, 2-dimethylaminoethanol, 1,10-binaphthyl-2,20-diamine and 2-hydroxy-3-
methoxybenzaldehyde (3 steps) catalyze highly enantioselective ring opening (ꢀ90–99% ee)
reactions of meso-N-4-nitrobenzoyl aziridines by TMSCN and TMSN3. The TMSN3-mediated
reactions give the highest enantioselectivities reported to date for several prototypical aziridines.
Selectivity in a related ring opening by silyl isothiocyanates depends on the substituents on silicon,
t
larger BuPh2SiNCS giving the best selectivities, especially when an yttrium or ytterbium complex is
used as the catalyst. The bimetallic yttrium complex also effects unprecedented regiodivergent
parallel kinetic resolution of racemic monosubstitued aziridines upon reaction with TMSN3. In these
reactions, each of the enantiomers undergoes nucleophilic addition of azide at different carbons
giving two products in nearly enantiomerically pure form. To explain the dramatic differences in the
selectivities between the mono- and bimetallic catalysts in these reactions, a mechanism that
involves activation of both the electrophile (aziridine) and the nucleophile (azide or cyanide) at
two different metals of the bimetallic complex is proposed. Molecular weight determination by
vapor pressure osmometry, diffusion ordered NMR spectroscopy (DOSY) and kinetic studies, which
suggest first order dependence on the concentration of the catalyst, provide strong support for this
proposal.
Received 22nd October 2013
Accepted 20th December 2013
DOI: 10.1039/c3sc52929k
www.rsc.org/chemicalscience
enantioselective ring-opening reactions of epoxides3–6 and
aziridines,5,7–11 which are among the most useful asymmetric
transformations.12 Mechanistic studies of these ring opening
Introduction
Bimetallic catalysis is ubiquitous in nature and many of the
essential processes that sustain life on our planet depend on
reactions catalyzed by bimetallic enzymes.1 The scope and
efficiency of bimetallic catalysts designed to carry out highly
selective organic transformations have been steadily
improving during the past two decades.2 Bimetallic activation
has played a signicant role in the development of catalytic
reactions have provided some of the most unequivocal
evidence for a bimetallic activation mode in these catalytic
transformations.9,13–17 In 2009 we reported the discovery of a
discrete dimeric yttrium salen complex [Y] (eqn (1)) that
enable the nucleophilic ring-opening reactions of meso-N-
acylaziridines with TMSN3 and TMSCN with exceptionally high
enantioselectivities.10 In reactions with racemic chiral azir-
idines, the same bimetallic Y complex promotes unprece-
dented regio- and enantiodivergent ring opening reactions
giving two different products from each of the enantiomers in
nearly enantiomerically pure form (eqn (2)).18 Here we report
the full details of the discovery of this remarkable catalyst,
along with the scope and limitations of its applications
including new reactions of aziridines with silyl iso-
thiocyanates. Since the original studies, we have also exam-
ined the key role played by the metal through a systematic
comparison of the catalytic activities (and selectivities) of
analogous complexes across the lanthanide series. Further
we document, for the rst time, structural and kinetic
evidence for the bimetallic activation in the aziridine opening
reactions.
Department of Chemistry and Biochemistry, The Ohio State University, 100 W. 18th
Av., Columbus, OH 43210, USA. E-mail: rajanbabu.1@osu.edu; parquett@
chemistry.ohio-state.edu; Fax: +1 614-292-1685; Tel: +1 614-688-3543; +1 614-688-
5886
† Electronic supplementary information (ESI) available: Full experimental details
for the preparation of previously undisclosed catalysts and substrates, typical
protocols for desymmetrization and regiodivergent kinetic resolution, 1H and
13C NMR spectra of key compounds, CSP HPLC and CSP GC of key
ring-opening products, HR-DOSY spectra for 2b and 10b, details of VPO and in
situ IR kinetics, CIF for 20 and 21. CCDC 967684–967685. For ESI and
crystallographic data in CIF or other electronic format see DOI:
10.1039/c3sc52929k
‡ Current address: State Key Laboratory of Phytochemistry and Plant Resources
in West China, Kunming Institute of Botany, Chinese Academy of Science,
Kunming 650201, Yunnan (China), Fax: +86 871-65216960, E-mail: wubin@
mail.kib.ac.cn
1102 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Results and discussion
Monometallic yttrium complexes in ring opening reactions of
aziridines
Our studies in the area started as an expansion of the applica-
tion of a series of monomeric yttrium salen complexes (1–4),
which we had synthesized as catalysts for trans-esterication of
secondary alcohols with enol acetates19 and ring-opening reac-
tions of meso-epoxides.15 Subsequently we also prepared related
the Y(III)-bis-phenolate complexes 5 and 6 using previously
reported ligands.20,21 During these studies it was observed that
one of these catalysts, 2a, effected the ring-opening of neat
epoxycyclohexane with TMSCN at room temperature at a
loading level of 0.0001 equivalent (substrate: catalyst ¼ 10 000!)
in less than 4 hours.15 Even though the best enantioselectivity
observed with this catalyst was modest (77% ee), such catalytic
efficiency remains unprecedented in nucleophilic ring opening
reactions. We wondered whether these catalysts would be
applicable to the more challenging aziridine ring opening
reactions. Aer screening a series of N-acyl groups as activating
groups, we recognized that the N-(4-nitrobenzoyl)aziridine
originally used by Oguni et al.22 was the most optimum
substrate for this reaction. The results of the reaction of N-(4-
nitrobenzoyl)-7-aza-bicyclo[4.1.0]heptane (aziridine 7) with
TMSCN (eqn (3)) in the presence of monomeric yttrium cata-
lysts, listed in Fig. 1, are shown in Table 1. An authentic sample
of the racemic product 8 was prepared by the reaction of 7 with
TMSCN in the presence of Y(OCH2CH2NMe2)3 or a bimetallic
complex 13 (see later).
Fig. 1 Monomeric yttrium-salen and related complexes and solid-
state structure of 1a.27
In general yttrium salen complexes are excellent catalysts for
these ring-opening reactions, even though as compared to
epoxides,15 a higher loading (typically 0.05 equivalents) of the
catalyst is required to complete the reaction in reasonable time.
As expected, the chirality of the diamine backbone and the
substituents on the salicylimine have signicant effects on
catalytic efficiency and selectivity. While the salen complex
derived from (R,R)-1,2-trans-cyclohexanediamine and 3,5-di-t-
butylsalicylaldehyde (1a) gives a quantitative yield of the product
(entry 1, 52% ee, R,R-product), the corresponding 3-methoxy-
salicylimine analog, 1b showed lower reactivity, also with
formation of the S,S-product (18% ee). For the salens (2a–e)
derived from 2,20-diamino-1,1-binaphthyl, the 3-methoxy analog
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1103

View Article Online
Chemical Science
Edge Article
Table 1 Monomeric yttrium catalysts for ring–opening reaction of
structures of related tetradentate lanthanide Schiff base
complexes, when examined as function of the solvent in which
they are prepared,23 suggest that hydrocarbon solvents favor
formation of very stable dinuclear [M2(L)3] complexes held
together by bridging ligands (eqn (4)). In sharp contrast, in THF,
mononuclear complexes, which form weak dimers [XM(L)]2 that
readily dissociates at or above room temperature are formed
(eqn (4)). We reasoned that the catalytic activities of such
complexes could be different, thanks in part due to the relative
juxtaposition of the metals, and possibility of bimetallic acti-
vation in the more stable binuclear complexes.
aziridine with TMSCNa
Entry
Catalyst
Yield (%)
%ee (R,R)b
1
2
3
4
5
6
7
8
1a
1b
2a
2b
2c
2d
2e
4
>99
58
77
52
ꢁ18
<5
55
30
14
10
ꢁ6
ꢁ34
ꢁ29
47
85
>99
>99
93
59
99
99
96
38
9
3a
3b
5
10
11
12
6
<5
a
b
See eqn (3), solvent CH2Cl2. Determined by Chiral Stationary Phase
(CSP) HPLC; (ꢁ)-sign indicate (S,S)-isomer.
(2b) gave the best result (entry 4). Salen complexes derived from
(S,S)-1,2-diphenyl-1,2-diaminoethane (4, entry 8) and a carbo-
hydrate-derived diamine (3a,b) also yielded only modest enan-
tioselectivity. The bis-phenolate/bis-phosphine oxide complex 5
based on a binaphthol is an excellent ligand giving nearly
quantitative yield, albeit with only low selectivity (entry 11).
In order to probe the importance of aggregation of the
yttrium catalysis we prepared the complex 1a in varying
mixtures of hexane and THF (eqn (5)), and examined the
enantioselectivities in the aziridine-opening reactions. The
results are shown in Table 2. We note that there is a signi-
cant deterioration of selectivity as the THF content of the
solution in which the catalyst is prepared increases, eventually
reversing the conguration of the major product in pure THF
(entries 1–4).
Based on the available information on the structural varia-
tions of lanthanide complexes in different solvents,23 one could
postulate the intermediates of different nuclearity as being
responsible for the remarkable solvent effects. While at the
outset this was only based on conjecture, it was not without
some support. For example, McCleland, Nugent and Finn have
invoked a dinuclear catalyst to explain the kinetic behavior of
(alkanolamine)3Zr-mediated ring-opening of epoxides with
azides (Scheme 1).14 In this mechanism, aer an initial dimer-
ization of the Zr complex, the electrophile (epoxide) is activated
at one of the metal centers, while the nucleophile (N3) is acti-
vated at the other metal. Our previous report15 of the second
order dependence of a monometallic Y-catalyst (2a) on the
exceptionally facile ring-opening reaction of epoycyclohexane
with TMSCN in nonpolar solvents is also consistent with such a
hypothesis. Further anecdotal support comes from the ease
A bimetallic catalyst for ring–opening of aziridines
The rst indications that aggregation might play a crucial role
in the selectivities of these reactions appeared in our attempts
to optimize the reaction by changes in the solvent. For example,
the monomeric complex 1a yielded varying selectivities (52–
63% ee) of the (R,R) product (eqn (3)) in non-coordinating
solvents such as CH2Cl2, ClCH2CH2Cl, benzene or mixtures of
these solvents.
The reaction carried out in THF, on the other hand, consis-
tently yielded the corresponding (S,S)-product, albeit in lower
(ꢀ20% ee) selectivities. We had previously made similar
observations in the ring-opening reactions of epoxycyclohexane
with TMSCN in the presence of 1a.15 While there was little doubt
about the structure of the monomeric catalyst 1a, thanks to a
solid state structure (Fig. 1, bottom) derived from X-ray crys-
tallography,19 the nature of the catalyst in THF remains highly
speculative. There is considerable evidence in the literature to
suggest that structures of tetradentate complexes of lanthanides
and yttrium depend strongly on the solvent. For example,
1104 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Table 2 Effect of the solvent used in the preparation of the catalyst on
enantioselectivity
Entry
Solventa [n-hexane/THF (%)]
Yield (%)
ee (%)
1
2
3
4
100 : 0
80 : 20
20 : 80
0 : 100
>99
88
82
52 (RR)
51 (RR)
32 (RR)
ꢁ43 (SS)
>99
a
Solvent for step 1 (eqn (5)). Solvent for step 2: CH2Cl2.
with which early transition metals including yttrium and
lanthanides form anion-bridged dimers.24
We decided to test the veracity of the bimetallic activation
using stable, discrete bimetallic yttrium complexes prepared
from yttrium tris-(2-dimethylaminoethoxide) and salen ligands
according to eqn (6).25,26 A solution of the ligand and the yttrium
alkoxide was heated in toluene at 70 ꢂC for 24 hours. The
solvents were removed and the product was redissolved in
minimum amount of CH2Cl2. Filtration of this solution and
careful layering on the top with hexane resulted in crystalliza-
tion of the product, which was used as a catalyst for the ring-
opening reactions. The structures of several catalysts prepared
in this fashion along with the a 3-D-rendition of a solid state
structure of a key complex, 10b, are shown in Fig. 2.27
Ring-opening reactions of aziridines
(A). With trimethylsilyl cyanide (TMSCN). The catalysts 9a,
9b, 10a, 10b and 10c were tested for the ring opening reaction of
the prototypical substrate 7 using TMSCN under the optimized
protocol shown in eqn (3) and the results are shown in Table 3.
Fig. 2 Structures of bimetallic catalysts. An enlarged rendition of the
solid-state structure of 10b is included in the ESI.†27
Table 3 Bimetallic yttrium catalysts for asymmetric ring–opening
reaction of aziridine 7 with TMSCNa
Entry
Cat.
Cat. (mol%)
Yield (%)
%ee (R,R)b
1
2
3
4
5
9a
9b
10a
10b
10c
5
5
10
10
5
79
58
>99
>99
64
37
ꢁ6
8
88 (94)c
30
a
b
See eqn (3), solvent ClCH2CH2Cl, rt, 3 days. Determined by CSP
HPLC; (ꢁ)-sign indicates (SS) isomer. c At 0 ꢂC.
The reactions were carried out in 1,2-dichloroethane using
recrystallized bimetallic catalysts at room temperature.
Compared to the typical epoxide opening reactions, the azir-
idine opening is sluggish, and the reactions are best carried out
using 10 mol% of the catalyst. The best catalyst for this reaction
Scheme 1 Nugent–Finn mechanism of bimetallic epoxide opening
catalyzed by (triisopropanolamine)3Zr complexes.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1105

View Article Online
Chemical Science
Edge Article
was found to be 10b, carrying a 2,20-binaphthyl bis-imine scaf-
folding. The 1,2-trans-cyclohexanediamine-derived complexes
9a and 9b gave much lower selectivities. The substituents on the
salicylimine fragment also play a signicant role in the enan-
tioselectivity. In the binaphthyl series, 3-methoxy substituent
(10b) on the salicylimine gave the best selectivities (entry 4).
Complexes bearing the 3-t-butyl substituent (10a) and a 3-
methoxymethyl (10c) gave lower selectivities (entries 3 and 5).
The catalyst 10b was tested for opening of other aziridines
under the optimized conditions, and the results are shown in
Table 4, columns 3 and 4. As seen in entries 1–3, the reactions
with TMSCN proceeds in very good yields and enantioselectiv-
ities comparable to the highest reported values8,9 for relatively
‘small’ aziridines. Perhaps reecting the capsular nature of the
catalyst, as the size of the aziridine increases, the reaction
becomes increasingly slow. Thus, aziridines derived from
cycloheptene 16 (entry 4), cyclooctene 17 (entry 5), 1,4-dihy-
dronaphthalene 18 (entry 6), Z-stilbene 19b (entry 8) and
E-stilbene 19c (entry 9) failed to react, while the one from Z-4-
octene 19a (entry 7) gave a lower yield under the standard
conditions.
Table 4 Ring-opening of aziridines catalyzed by bimetallic salen
complex 10ba
TMSCN
Y
TMSN3
Y
No.
1
Aziridineb
eec
94
eec
97
87
85
99
99
2
3
4
99
92
—
98
99
97
86
0
99
47
(b). With trimethylsilyl azide (TMSN3). The dimeric catalyst
10b seems to be especially suitable for the aziridine opening
with TMSN3 (Table 4, columns 5 and 6), giving the highest
enantioselectivities reported to-date for all substrates in Table 4
except for 18 (entry 6).28,29 Several substrates that either reacted
sluggishly or did not react at all with TMSCN give acceptable
yields and selectivities in reactions with TMSN3 under virtually
identical conditions. For example, cycloheptene-derived azir-
idine (16, entry 4) gave excellent enantioselectivity and a modest
yield in the TMSN3-mediated opening, even though this
substrate failed to react with TMSCN. Likewise aziridines from
1,4-dihydronaphthalene (18, entry 6), Z-4-octene (19a, entry 7)
and Z-stilbene (19b, entry 8) react, to give acceptable yields and
excellent enantioselectivities. Only aziridines derived from
cyclooctene 17 (entry 5) and E-stilbene 19c (entry 9) failed to
react with both TMSCN and TMSN3.
5
6
0
0
—
—
0
—
99
90
7
8
47
0
82
73
96
92
97
—
The stereoselectivity in the opening of 19b, a substrate prone
to SN1-type reactivity in the presence of Lewis-acid, is especially
noteworthy. The structure of the product 20, arising from a
clean inversion of conguration, was conrmed by X-ray crys-
tallography (Fig. 3).30
9
0
—
0
—
(c). With trimethylsilyl isothiocyanate (TMSNCS). For the
synthesis of racemic b-aminothioethers from aziridines and
sulfur nucleophiles, heterogeneous catalysts,31,32 lithium
perchlorate,33 aqueous cyclodextrin34 and indium chloride35
have been used as effective catalysts. Uncatalyzed reaction of
activated aziridines including 7 gives a mixture of the thiocya-
nate adduct 21 and the corresponding isothiocyanate.36 Enan-
tioselective ring-opening of meso-N-acylaziridines by aromatic
thiols using stoichiometric amount of a Zn catalyst prepared
from Et2Zn and tartrate has been known since 1995.22 Other
catalytic methods for addition of sulfur nucleophiles use chiral
alkaloid derivatives,37 and chiral biphenanthryl phosphoric
acids as organocatalysts,38,39 the latter giving outstanding
a
b
See eqn (3) for procedure (10 mol% catalyst, 3 days). Ar ¼ 4-
nitrophenyl. c %, from CSP HPLC. All (R,R)-products.
In light of the excellent selectivities observed for TMSCN and
selectivities for the addition of aryl thiols. Aliphatic thiols TMSN3 we proceeded to examine TMS-NCS (trimethylsilyl iso-
generally give lower yields and selectivities.
thiocyanate) using the dimeric yttrium complexes (eqn (7)). The
1106 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Fig. 4 Solid-state structure of 21 (eqn (7)).
Fig.
3 Solid-state structure of (S,S)-2-azido-1-(4-nitrobenzoyl)-
amino-1,2-diphenylethane (20, from aziridine 19b).
Table 6 Effect of the silyl reagent on the R3SiNCS-mediated enan-
tioselective ring opening of aziridine 7a
results of the initial scouting experiments using various cata-
lysts are shown in Table 5. Since thiocyanate is an ambidentate
nucleophile capable of bonding through S or N, the structure of
the adduct 21, initially inferred from IR spectroscopy,36 was
conrmed by X-ray crystallography (Fig. 4).30 As before, the
dimeric catalysts based on 1,2-trans-cyclohexanediamine
(entries 1 and 2) are not as efficient or selective as the ones
based on 2,20-diamino-binaphthyl. Among these the ligand 10b,
which yielded the best selectivities for TMSCN and TMSN3
additions, appears to be the best for the TMSNCS addition
(entry 4) giving excellent yields and ee's up to 60% in the
additions to the meso-aziridine 7.
(d). With larger silyl isothiocyanates. The substituents on
the silicon in the reagent have a signicant effect on the
enantioselectivity of this reaction as shown in Table 6. Highest
ee's were obtained with bulky silyl reagents (entry 2, Et3Si– and
entry 9, tBuPh2Si–).
Entry
Cat.
R3
Yield (%)
eeb (%)
1
2
3
4
5
6
7
8
9
10
10b
10b
10b
10b
10b
11c
11c
11c
11c
11c
Me3
Et3
91
99
76
96
77
77
96
99
99
84
57
69
39
61
67
60
47
63
68
55
tBuMe2
tBuPh2
PhMe2
Me3
Et3
tBuMe2
tBuPh2
PhMe2
a
See eqn (7) (10 mol% cat., 3 days). All (R,R)-products. b Determined by
CSP HPLC.
been studied, systematic studies of the effect of the central
metal are few.15,40,41 As for the effect of lanthanides, Jacobsen
reported that in the TMSCN-mediated opening of epoxides
catalyzed by (2,5-bis-oxazolinylpyridine)MCl3, there is a strong
correlation of enantioselectivity with the atomic number, with
the smaller ions giving the highest ee's.4
(e). With trimethylsilyl isocyanate (TMSNCO). Several
attempts to effect the ring opening reaction of 7 with various
yttrium catalysts including 10b and 13 were unsuccessful. The
starting material was recovered unchanged.
Effect of the central metal in ring-opening reactions of meso-
aziridines
Lanthanides. Even though many metal-catalyzed ring
opening reactions of meso-epoxides and meso-aziridines have
Table 5 Catalyst screening for Me3SiNCS-mediated ring–opening of
meso-aziridine 7a
Entry
Cat.
Time (d)
Yield (%)
eeb (%)
1
2
3
4
5
6
7
8
9a
2
2
3
3
2
1
1
1
32
75
82
91
96
89
55
87
74
96
77
87
7
8
41
9bc
10a
10b
10cc
10d
10e
10f
11a
11b
11cc
11d
57 (60)d
47
We studied the effect of substituting the yttrium in the
bimetallic complex 10b with various lanthanides on the ring
opening reactions of aziridines and the results are shown in
Table 7. While all dimeric complexes with the generic structure
22 (eqn (8)) carrying the salen ligand are excellent catalysts
giving nearly quantitative yields, the selectivity of the reaction
varies considerably. For the TMSN3 reactions, except for Pr,
early lanthanides with larger ionic radii (La–Gd) generally gave
lower selectivities. Yttrium, and late lanthanides, Dy, Er and Yb,
23
8
49
50
34
60
58
9
0.67
1.25
2
10
11
12
1
a
See eqn (7) (10 mol% catalyst, 3 days). All (R,R) products. b Determined
by CSP HPLC. c 5 mol% cat. d At ꢁ3 ꢂC, 4 days, 99% yield.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1107

View Article Online
Chemical Science
Edge Article
Table 8 Effect of the trialkylsilyl group on the Yb-catalyzed ring-
opening reactions of aziridine 7 with R3SiNCSa
all with ionic radii lower than 91 pm gave the best selectivities in
these reactions (>95% ee, Table 7, column 6, entries 1, 7–9).
Incidentally the Lewis acidity of these late lanthanide ions and
yttrium are also proportionally higher.42 In the case of TMSNCS,
a similar behavior is seen (columns 7 and 8), even though the
trend is not as readily discernable.
Finally the best results in the silyl isothiocyanate-mediated
reaction are obtained when the Yb catalyst is used in conjunc-
tion with a bulky silyl reagent (Table 8). Thus both Et3Si-NCS
and tBuPh2Si-NCS upon reaction with 7 gave very high yields of
the product 21 in ꢀ85% ee. A recrystallization of this product
from ethyl acetate/hexane gives the product in 88% ee. Under
these conditions, the aziridines 14 and 15 also gave excellent
yields; but the enantioselectivities were only modest (48% and
52% respectively). Increased background reaction in these more
strained substrates is most likely the reason for this deteriora-
tion of selectivity (Table 8).
Entry
Substrate
R3Si
Yield (%)
eeb (%)
1
2
3
4
5
6
6.
7
7
7
7
7
14
15
Me3Si
Et3Si
99
99
98
99
98
99
89
76
80 (85)c
52
tBuMe2
tBuPh2
Me2Ph
Et3Si
85 (88)c
66
48
Et3Si
52
a
b
See eqn (8), 10 mol% cat., 22h, 24 h. From CSP HPLC unless
specied. c From CSP GC.
Regiodivergent parallel kinetic resolution of racemic
aziridines
While enantioselective desymmetrization (Scheme 2, A) of
meso-epoxides and aziridines, as well as kinetic resolution
(Scheme 2, B) of racemic epoxides have been the subject of
extensive research, an alternate approach to enantio-pure
products from the latter class of substrates43–47 which involves
divergent reactions of the individual enantiomers (Scheme 2, C)
of a racemate, has received relatively lesser attention.48 Such an
approach for the production of intermediates from racemic
starting materials has at least one principal advantage over the
conventional kinetic resolution (Scheme 1, B) in that the relative
rates of reactions of the enantiomers (kf/ks) can remain close
and need not approach the rarely observed >100 mark
(assuming a rst order reaction) to realize selectivities of >98%
enantiomeric excess (ee).49–51 Further if the nucleophile (Nu) is
structurally related to X, the ring atom, and both X and Nu can
be transformed into the same functional group, both products
[P1] and [P2] (Scheme 2, C) maybe converted into a common
intermediate [P3], or a congener, useful for further synthetic
operations. An example is the conversion of the b-amidoazide to
a diamine28 using the Burk protocol.52 This avoids yet another
limitation of classical kinetic resolution, viz., the maximum
Scheme
heterocycles.
2 Enantioselective ring-opening reactions of small-ring
obtainable yield of 50% in such reactions. However, an enan-
tiospecic catalyst that can accomplish this must meet
exceedingly stringent requirements with respect to its structure
and reactivity so as to promote divergent reactions of two
enantiomers by appropriate recognition events. Thus far this
strategy, termed ‘enantioconvergent’ synthesis53 has been
Table 7 Effect of the central atom on the ring-opening reactions of aziridines with Me3SiN3 and Me3SiNCSa
TMSN3
Me3SiNCS
Yield (%)
Entry
Metal
Cat.
M(III) sizeb (pm)
Yield (%)
ee (%)
eec (%)
1
2
3
4
5
6
7
8
9
Y
La
Pr
Nd
Sm
Gd
Dy
Er
10b
22a
22b
22c
22d
22e
22f
90.0
103.2
99.0
98.3
95.8
93.8
91.2
89.0
86.8
99
98
98
98
99
97
99
99
99
97
31
87
80
89
73
99
99
95
91
92
99
99
99
98
99
99
99
57 (69)d
19
66
42
50
43
61
22g
22h
65
Yb
76 (88)d
a
c
See eqn (8), 10 mol% cat. b Coordination number 6 (in picometers). From CSP HPLC or CSP GC. d Using Et3SiNCS.
1108 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
successful in a practical sense only in the realm of enzyme- Under the optimized conditions described in eqn (9), about
mediated reactions of epoxides.54,55 In an example pertinent to 11% of the starting material was recovered, and was found to
this report, racemic styrene oxide has been transformed into a have an ee of 54%, with the (R)-isomer predominating. This
single, enantiomerically pure (R)-1,2-phenylethanediol by a suggests that the more reactive (S)-enantiomer of the starting
combination of two recombinant microbial epoxide hydrolase material is consumed at ꢀ3.35 times faster than the slow
enzymes (Scheme 3) possessing complementary enantiose- reacting (R)-isomer, each of them proceeding with exceptional
lectivity and regioselectivity.56,57a
regio- and enantiospecicity. Since the selectivity in the ring
Having recognized 10b to be an excellent catalyst for opening of each of the enantiomers is nearly perfect (ꢀ99% ee
desymmetrization of meso-aziridines we turned our attention to for the product), the low kf/ks is of no consequence to the
kinetic resolution of racemic aziridines. Our studies started selectivity in the formation of the two products.
with an examination of the ring-opening reactions of a proto-
As is the case with the desymmetrization reactions, the
typical racemic aziridine 23 with trimethylsilyl azide (TMSN3) in monometallic catalyst 2b, is much less selective (also less
the presence of the dimeric yttrium-salen complex 10b. Chiral reactive) in these regiodivergent reactions (eqn (10)). Under the
stationary phase HPLC analysis of unconverted starting mate- standard conditions where the bimetallic 10b le only 11%
rial 23 from reactions done under the most optimum conditions unreacted 23, 56% of the starting material (23) was recovered in
for kinetic resolution (rt, ratio of TMSN3 : 23 ¼ 1 : 1 to 1 : 2, nearly racemic form in these reactions. The ring-opening
conversions of 50–80%) returned disappointingly low ee's for products 24 (14% yield) and 25 (22% yield) were formed in 71%
unconverted 23 even at high conversions. This surprising result, and 54% ee respectively.
along with the presence of two new, easily separated
compounds (tlc) suggested that the overall reaction appeared
not regioselective and/or the enantiomers reacted at compa-
rable rates. However, upon analysis of the two new products by
CSP HPLC and NMR we were pleasantly surprised to nd that
both compounds, now identied as two regioisomeric ring-
open products 24 and 25, were formed in ꢀ99% ee (eqn (9))!58
The congurations of these products were assigned by
comparison of specic rotations and retention times on CSP
HPLC with those of authentic samples prepared from aziridines
of known conguration (see later).
Attempts to prepare the authentic samples of the racemic
azido–amides 24 and 25 via Y-catalysis revealed yet another
fascinating aspect of these reactions. The ring opening reac-
tions of the starting aziridine 23 using an achiral salen complex
derived from ethylenediamine 13 (Fig. 2) or the Y-alkoxide
Y(OCH2CH2NMe2)3 (eqn (11)) are signicantly slower compared
to the reactions with the complex 10b (with 23, only 24% overall
conversion in 6 days and with Y(OCH2CH2NMe2)3 only 13%
conversion in 10 days) under otherwise similar conditions.
Thus there is a signicant ligand assistance59 in the catalytic
reactions of 23.
These results can be explained by regiodivergent parallel
kinetic resolution (Scheme 2, C) of the two enantiomers of 23.
Thus, the nucleophilic attack by azide occurs exclusively at the
primary position in (R)-23 leading to the azido–amide 24 as the
only product. In sharp contrast, the (S)-23 gives a product 25,
resulting from exclusive SN2-inversion at the secondary center.
A number of structurally different aziridines were subjected
to the ring-opening reactions and the results are listed in Table
9. For all substrates except for the t-butyl derivative 23h (entry
9), the reaction is completely enantiospecic in the formation of
the primary azide 24, the nucleophilic attack taking place at the
Scheme 3 Enzyme-mediated enantioconvergent synthesis of (R)-1-
phenyl-1,2-ethanediol.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1109

View Article Online
Chemical Science
Edge Article
Table 9 Regio- and enantio-divergent ring-opening reactions of chiral aziridines
Yield/%eea
Entry
Aziridine
R in 23
Cat. (mol%)
Time (d)
(R)-24
(R)-25
1
2
4
5
6
7
8
9
23a
23b
23c
23d
23eb
23f
Cyclohexyl
Cyclopentyl
Benzyl
15
15
10
10
15
15
15
15
2
2
6
6
2
2
2
6
42/>99
33/>99
42/99
32/99
15/>99
8/>99
16/99
24/96
46/99
39/98
47/97
48/90
51/93
48/96
48/92
0/—
n-Bu
Cyclohexyl-methyl
Cyclopentyl methyl
i-Propyl
23g
23hc
t-Butyl
10
—
—
10
3
0
0
11
10
3
0
0
a
Isolated, >99% ee denotes that the other enantiomer was not detected in the chromatogram. b 24% starting material (96% ee) recovered. c 58%
starting material (47% ee) recovered.
primary carbon of the (R)-aziridine, giving 99 %ee of the
The high specicity and identity of the products in the ring-
product in each case. For 23h it is slightly lower (96 %ee). opening were further conrmed by reactions of several enantio-
Selectivity in the formation of 25 (from the SN2-displacement at pure aziridines with TMSN3. An exceptionally challenging
the secondary center of the fast-reacting (S)-aziridine) is also situation is illustrated with the complementary behavior of (S)-
impressive, less so for substrates 23d,e, and 23g (R ¼ n-Bu, and (R)-2-methylaziridines (eqn (12a) and (12b)). As expected,
cyclohexylmethyl and i-Pr). There is some erosion in the selec- the (R)-enantiomer of the aziridine 26 gives exclusively the
tivity for formation of these products, showing proportionately primary azide (R)-27 (eqn (12a)), and the (S)-enantiomer of 26
higher amounts of ring-opening of the (R)-enantiomer with gives the secondary azide (R)-28 (eqn (12b)) with no trace of
displacement at the secondary position. Curiously, in the case contamination with the regioisomeric product in each case. This
of the t-butylaziridine (23h) the normally fast reacting (S)- high regioselectivity seen in the formation of the two different
isomer was recovered along with signicant amounts of products from the two enantiomers indicates that the nucleo-
unreacted (R)-isomer (entry 9). Attack at the neopentyl carbon is phile is capable of distinguishing between two electrophilic
60
˚
forbidden presumably because of steric congestion.
carbons that are separated by 1.46 A in the aziridine. In the
The exquisitely high enantioselectivity in these regiodi- formation of the primary azide (R-27), the catalyst is able to
vergent reactions come at a cost in terms of the scope of the overcome the well-established preference for nucleophilic attack
substrates that are amenable to this reaction. Two examples of at the secondary position in Lewis acid-catalyzed opening of a
racemic C2 symmetric substrates that fail to react are shown in small ring. One may speculate that such selectivity is hardly
entries 10 (23i) and 11 (19c). Recall that the cis-aziridines 19a possible if only one of the components, say, the aziridine is
and 19b do undergo the desymmetrization reaction with high ee activated. A bimetallic activation in which both the aziridine and
(entries 7 and 8, Table 4).
the nucleophile (N3ꢁ) are activated thus looks very plausible.
1110 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
bimetallic mechanism in the reactions catalyzed by the discrete
dimeric Y-complex 10b.
Stability of the monometallic (2b) vs. bimetallic (10b)
catalysts
NMR spectroscopy. Recrystallized samples of the mono-
metallic complex (2b) and the bridged, bimetallic complex (10b)
were dissolved in the appropriate solvent and the spectra were
recorded. Both in THF-d8 and CD2Cl2, the complex 2b gave very
1
complex H NMR spectra with broad peaks. In sharp contrast,
the catalytically useful bridged dimer 10b is an extremely well-
behaved compound exhibiting sharp signals for the various
protons and carbons. Fig. 5 shows just the aromatic region of
the two spectra. Full spectra are included in the ESI p. S28 (2b)
and S29 (10b).† With the help of a series of decoupling and 2D-
experiments most of these signals can be assigned,62 and, as in
solid state (Fig. 2), it appears to be a C2 symmetric structure in
solution with the expected 34 distinct down-eld (d > 110 ppm)
and 6 up-eld (d < 70 ppm) signals in the 13C NMR spectrum. A
CD2Cl2 solution of 10b is quite stable for days as compared to a
CD2Cl2 solution of the monomeric complex 2b, which
undergoes signicant deterioration in ꢀ24 h. While the exact
nature of this change in the monomeric complex is not easily
discernable because of the complex nature of the spectra,
diffusion ordered NMR spectra (DOSY) recorded as a function of
time gives some qualitative indication of the changes in solu-
tion. The DOSY spectrum of the monomeric species 2b in
CD2Cl2 undergoes substantial changes when recoded over
time.62 For example ꢁlog D, a parameter related to the hydro-
dynamic volume of the solute goes from 8.42 at t ¼ 0 to 8.75 at
t ¼ 12 h for the monomer, eventually returning to a value of 8.4
at 23 h. These changes could be attributed to some form of
dynamic behavior in this molecule. In sharp contrast, the DOSY
spectrum of the bimetallic complex 10b in CD2Cl2 remains
invariant (ꢁlog D ¼ 8.40–8.45) over 40 h and beyond.
Ring-opening reactions of two other sets of enantio-pure
substrates (23a, and 23c, Table 10) conrm the generality of
these highly selective reactions. As was inferred from the
previous studies with racemic substrates (Table 9), the (S)-
aziridines react faster than the (R)-enantiomers in all cases
under identical conditions.
Regiodivergent parallel kinetic ring opening of racemic
aziridines is the rst example of a single, small molecular-
weight catalyst exhibiting such enzyme-like complementary
selectivities in the ring-opening reactions of small ring
compounds.61 High yields are obtained for several 1,2-diamine
derivatives from racemic aziridines in nearly enantiomerically
pure (>97% ee) form. Subsequent high-yielding reactions
(amide hydrolysis and azide hydrogenation) should produce the
same 1,2-diamine from each of the ring-opening products28,57b
even though from a synthetic perspective the differentially
protected primary products (vicinal azido-amides) are likely to
be more valuable.
Molecular weights by osmometry. Measure of the relative
stability of these complexes and the state of aggregation in
solution can also be gleaned by molecular weight measure-
ments that depend on colligative properties. Among these vapor
Kinetics and mechanism of the ring opening reactions
The exceptionally high enantioselectivities, especially in the pressure osmometry (VPO) is a simple yet powerful technique.63
regiodivergent parallel kinetic resolutions, clearly suggest a The molecular weights of the monomeric and yttrium catalysts
central role for both metallic centers of the catalyst in these were determined using vapor pressure osmometry (VPO) using a
reactions. Since anecdotal evidence in the epoxide-15 and azir-
idine- (for example, see Table 2) openings (and a host of other
reactions) implicate aggregation of the metallic catalyst even
Knauer K-7000 apparatus operated at 306 K in a rigorously inert
atmosphere. The results are shown in Table 11. The monomeric
complex 2b (calculated molecular weight 843.9 g molꢁ1) was
when a monomeric complex is used, we sought to provide dissolved in 1,2-dichoroethane, and, using sucrose octaacetate
further direct structural and kinetic evidence in support of a (calculated MW ¼ 678.6, observed 684) as a standard, its
molecular weight was determined.30 The observed value of the
molecular weight of this compound is highly dependent on the
Table 10 Regiospecific ring–opening reactions of enantiopure age of the solution, as was suspected from the DOSY experi-
ments and other evidence reported earlier.15 Aer a solution of
aziridines
2b was stored under inert atmosphere for 24 hours, a value of
Entry
Aziridine (%ee)
Product
Yield (%)
ee (%)
1430 was recorded for the molecular weight, which is a clear
indication of some form of aggregation in solution. The
molecular weight measurements on the dimeric complex 10b
(calculated MW ¼ 1455.3 g molꢁ1), on the other hand, showed
little tendency for change as a function of resident time in
solution. Aer 24 hours, a value of 1483 was recorded.
1
2
3
4
(R)-23a (95)
(S)-23a (97)
(R)-23c (99)
(S)-23c (99)
(R)-24a
(R)-25a
(R)-24c
(R)-25c
56a
79
>99
96
>99
97
46b
87
a
25% (R)-23a recovered (>99% ee). b 20% (R)-23c recovered (92% ee).
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1111

View Article Online
Chemical Science
Edge Article
Fig. 5 1H NMR spectra of recrystallized monometallic (2b) and bimetallic (10b) catalysts. For clarity only aromatic portion is shown. Full spectra
are included in the ESI† (pages S28 and S29).
Table 11 Molecular weights from VPO
in 1.0 mL of 1,2-dichloroethane inside a nitrogen-lled glove
Observed MW
Calculated MW
box. The reactor was brought outside the box and, the apparatus
was placed in a water bath maintained at 16 ꢃ 0.2 ꢂC. With
rigorous exclusion of moisture, and under copious ow of
nitrogen, the IR probe was introduced through the large
opening of the reactor. Aer equilibration, the aziridine 7 (61.5
mg, 0.25 mmol) dissolved in 1.0 mL of 1,2-dichloroethane was
introduced through a serum stopper on the side-arm. IR spectra
were collected at 10-minute intervals.
A stack of spectra [6 mol% of catalyst (0.006 M) with respect
to TMSN3], showing the relevant peaks is shown in Fig. 6. Fig. 7
shows the proles of peaks as a function of time. Individual
spectra at t ¼ 1 min, 90 min and 180 min and at the end of the
reaction are shown in Fig. 8.
Compound
(g molꢁ1
)
(g molꢁ1
)
b-Lactose octaacetate
(standard)
2b
2b
684
678.6
920a
1430b
1483
843.9
843.9
1455.3
10b
a
VPO measurements recorded immediately aer dissolving the
compound in 1,2-DCE. Measurements aer 24 h at rt in an inert
b
atmosphere.
Molecular weight of b-lactose octaacetate (calculated MW ¼
677) was determined just as a test case, and the experiment
gives acceptable value (observed MW ¼ 684) within experi-
mental error.
The assignment of the peak at 2079 cmꢁ1 to a metal azide
follows from the in situ experiment in which stoichiometric
amount of TMSN3 is added to the catalyst. As shown in Fig. 9,
rapid disappearance of the peak at 2143 cmꢁ1 and the corre-
sponding appearance of the peak at 2079 cmꢁ1 strongly suggest
Kinetic studies
The Y-catalyzed ring opening of aziridine by TMSN3 is a reaction the formation of a metal azide. Note that the intensity of this
ideally suited for a kinetic study by in situ IR spectroscopy. The peak remains virtually unchanged through out a typical kinetic
characteristic absorption signals due to TMSN3, the putative experiment (Fig. 7), conrming its presence as the likely steady-
metal azide,64 and the product azide appear at 2143 cmꢁ1, 2079 state intermediate.
cmꢁ1 and 2097 cmꢁ1 respectively, and each are well separated
Kinetic runs were carried out in a 3-fold concentration range
from other signals. In addition a peak at 1648 cmꢁ1 whose of the catalyst with the following equivalents: 0.04, 0.06, 0.10
change in intensity parallels that of the product azide peak and 0.12. The absorbance data was plotted with ln[(a0)/(a0 ꢁ x)]
(2097 cmꢁ1) has been tentatively assigned as belonging to an vs. time (min) where a0 is the initial absorbance of TMSN3 and
iminoether65 intermediate.66
(a0 ꢁ x) is the absorbance of remaining TMSN3 at time t (Fig. 10).
A typical kinetic experiment was run as follows with TMSN3
As can be seen in Fig. 10, this is an exceptionally well-
as the limiting reagent with excess of the aziridine 7. A two- behaved system, and gives highly reliable data from which kobs
necked glass reactor was charged with a mixture of TMSN3 values can be calculated from the slope of the ln[a0/(a0 ꢁ x)] vs. t
(11.5 mg, 0.1 mmol) and the catalyst 10b (17.5 mg, 0.012 mmol) graph. These values for 4 different concentrations are listed in
1112 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
Fig. 6 Typical kinetic run. Stacked IR spectra shows disappearance of
TMSN3 (2143 cmꢁ1) and appearance of the product (2097 cmꢁ1) and
active species [Y]–N3 (2079 cmꢁ1).
Fig. 9 Stacked IR spectra showing disappearance of TMSN3 (2143
cmꢁ1) and appearance of the active species, [Y]–N3 (2079 cmꢁ1).
Fig. 7 Profiles of absorbances for TMSN3 and the product and active
species [L]Y–N3 as a function of time.
Fig. 10 Pseudo-first order plot of log[(TMSN3)0/(TMSN3)t] vs. t for
aziridine opening reaction (see ESI for similar plots with other catalyst
concentrations†).
suggesting a dissociation of the dimer followed by catalysis by
monometallic species showed relatively poorer t.30 This data
taken together with the low enantioselectivity observed (Table 1,
entry 4) upon use of structurally related monomeric catalysts
(e.g., 2b) provide strong support for the bimetallic mechanism
in these reactions.
A Mechanistic proposal
Fig. 8 The azide peaks as a function of time.
One possible mechanism consistent with all the observations is
shown in Scheme 4. In this mechanism, the reaction is initiated
Table 12. A plot of catalyst concentration vs. kobs (Fig. 11), which with the formation of the Y-azide by a metathesis reaction of [L]
gives a straight line with excellent t (R2 ¼ 0.99988), is consis- Y–O bond in the dimeric complex with TMSN3 to form the
tent with a rst order dependence of the catalyst concentration intermediate 29, tentatively identied as the species respon-
on the ring-opening reaction. A similar plot of [cat]0.5 vs. kobs
,
sible for the IR absorption at 2079 cmꢁ1. The nature of the
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1113

View Article Online
Chemical Science
Edge Article
Table 12 Data table kobs, [cat]
bridging interaction between the azide and the two Y atoms is
purposely le vague, since both linear and terminal bridging
are well known in lanthanides.67–69 Activation of the aziridine by
the second Y and an internal nucleophilic reaction results in the
formation of yet another intermediate, one possible structure of
which is depicted in 31. Oxygen or nitrogen silylation of 31 leads
to the silylether derivative of the product (32), returning the
catalyst as the steady state species 29. Lack of reactivity of the
bulky aziridines can be rationalized by the reluctance of a bulky
substrate to enter the cavity at the bottom of which sits the
Y2O2-core (see the 3D-representation of the solid-state structure
of 10b in Fig. 2), or any structurally related putative interme-
diates generated from 10b in the initial steps.
[Cat.] M
kobs (minꢁ1
)
0.002
0.003
0.005
0.006
0.00060
0.00215
0.00508
0.00654
Summary and conclusions
Based on our initial work on the use of monometallic yttrium-
salen complexes for TMSX-mediated ring opening reactions of
epoxides and aziridines, the structures of these complexes, and,
their propensity to form dimers, a set of binuclear complexes
(e.g., 10b) were synthesized and tested for desymmetrization of
meso-aziridines and related reactions. Enantioselectivities
observed for the TMSN3 and TMSCN mediated ring opening
reactions using 10b as a catalyst are among the highest reported
to-date for several aziridines. The reaction has been extended to
R3Si-NCS additions, which gave excellent yields and surprisingly
high ee's for the b-amino-alkyl thiocyanate, given the prevalence
of an uncatalyzed background reaction. The rate of this thermal
background reaction, especially with Me3SiNCS, can be sup-
pressed with the use of bulky (Et3Si and t-BuPh2Si) iso-
thiocyanate reagents. Use of Yb instead of Y also improves the
selectivity in selected cases.
Fig. 11 Linear kobs vs. [cat]1 Indicating a unimolecular reaction in the
bimetallic catalyst.
The ligands developed for yttrium for these reactions can be
used to prepare the corresponding bimetallic lanthanide
complexes. A systematic examination of these complexes across
the lanthanide series suggests that the smaller ions (e.g., Dy, Er
and Yb) are just as effective as yttrium in many of these ring-
opening reactions.
The bimetallic catalyst 10b upon reaction with chiral mon-
osubstitued aziridines and TMSN3 brings about unprecedented
regiodivergent parallel kinetic resolution. The catalyst acceler-
ates the reaction of both members of the racemic mixture, but
does so along divergent pathways, inducing opposing regiose-
lectivity in its respective encounters with the Y-reagent. An
advantage of this approach is that when azide is used as the
nucleophile, the full racemic mixture of an aziridine can be
transformed into a set of desirable resolved products, in this
case, 1,2-diamines.
The exact origin of the specicity of the bimetallic catalysts
remains speculative. It is conceivable that the capsular nature of
the catalyst, as revealed by a solid-state structure (Fig. 2), helps
to coordinate the electrophile (the aziridine) and the nucleo-
phile (the azide) in specic orientations during the activation
process. Each of the diastereomeric complexes has a different
electrophilic carbon favorably juxtaposed for attack by the
nucleophile, which is linked to the second metal within the
connes of the cavity. Some support for this conjecture comes
Scheme 4 One possible mechanism consistent with the available
data.
1114 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
from the lack of selectivity when a structurally related mono-
metallic complex is used, and the strict rst order dependence
of the ring-opening reaction on the concentration of the bime-
tallic reagent. Lack of reactivity when sterically demanding
aziridines such as the ones derived from cyclooctene or trans-
stilbene, also supports this notion. Vapor pressure osmometry
and NMR studies suggest that while the C2-symmetric bime-
tallic catalyst is quite robust in solution for the duration of the
reaction, the monometallic species undergo signicant
aggregation.
The results disclosed in this paper add to a growing list of
impressive applications of homo- and heteronuclear bimetallic
complexes for highly selective carbon–carbon and carbon–
heteroatom bond forming reactions. By all indications interest
in this area of research continue to burgeon with applications
being sought in small molecule as well as polymer16,70,71
chemistry.
1973; (i) M. Mazet and E. N. Jacobsen, Angew. Chem., Int.
Ed., 2008, 47, 1762; (j) Q.-X. Guo, Z.-J. Wu, Z.-B. Luo,
Q.-Z. Liu, J.-L. Ye, S.-W. Luo, L.-F. Cun and L. Z. Gong, J.
Am. Chem. Soc., 2007, 129, 13927; (k) M. H. Yang, C. J. Zhu,
F. Yuan, Y. J. Huang and Y. Pan, Org. Lett., 2005, 7, 1927;
(l) S. Mouri, Z. Chen, H. Mitsunuma, M. Furutachi,
S. Matsunaga and M. Shibasaki, J. Am. Chem. Soc., 2010,
132, 1255; (m) Z. Zhang, Z. Wang, R. Zhang and K. Ding,
Angew. Chem., Int. Ed., 2010, 49, 6746; (n) A. Matsuzawa,
T. Mashiko, N. Kumagai and M. Shibasaki, Angew. Chem.,
Int. Ed., 2011, 50, 7616; (o) J. A. Kalow and A. G. Doyle, J.
Am. Chem. Soc., 2011, 133, 16001; (p) P. Gopinath,
T. Watanabe and M. Shibasaki, Org. Lett., 2012, 14,
1358.
3 K. B. Hansen, J. L. Leighton and E. N. Jacobsen, J. Am. Chem.
Soc., 1996, 118, 10924.
4 S. E. Schaus and E. N. Jacobsen, Org. Lett., 2000, 2, 1001.
5 K. Arai, S. Lucarini, M. M. Salter, K. Ohta, Y. Yamashita and
K. Kobayashi, J. Am. Chem. Soc., 2007, 129, 8103.
6 R. M. Thomas, P. C. B. Widger, S. M. Ahmed, R. C. Jeske,
W. Hirahata, E. B. Lobkovsky and G. W. Coates, J. Am.
Chem. Soc., 2010, 132, 16520.
Acknowledgements
This work was supported by US National Science Foundation
(CHE-0526864 and CHE-1057818). This paper is dedicated to Dr
William A. Nugent in recognition of his seminal contributions
to early transition metal chemistry as applied to organic
synthesis.
7 Z. Li, M. Fernandez and E. N. Jacobsen, Org. Lett., 1999, 1,
1611.
8 T. Mita, I. Fujimori, I. R. Wada, J. Wen, M. Kanai and
M. Shibasaki, J. Am. Chem. Soc., 2005, 127, 11252.
9 I. Fujimori, T. Mita, K. Maki, M. Shiro, A. Sato, S. Furusho,
M. Kanai and M. Shibasaki, J. Am. Chem. Soc., 2006, 128,
16438.
Notes and references
1 (a) R. H. Holm, P. Kennepohl and E. I. Solomon, Chem. Rev.,
1996, 96, 2239. This article appears in an issue of Chemical 10 B. Wu, J. R. Parquette and T. V. RajanBabu, Angew. Chem.,
Reviews entirely devoted to ‘Bioinorganic Enzymology’. Int. Ed., 2009, 48, 1126.
Several other articles in this issue (1996, 96(7), 2239–3042) 11 Y. J. Xu, L. Q. Lin, M. Kanai, S. Matsunaga and M. Shibasaki,
also deal with bimetallic enzymes; (b) H. Steinhagen and J. Am. Chem. Soc., 2011, 133, 5791.
G. Helmchen, Angew. Chem., Int. Ed., 1996, 35, 2339; (c) 12 Other reports on stereoselective opening of aziridines, see:
J. A. Cricco, E. G. Orellano, R. M. Rasia, E. A. Ceccarelli
and A. J. Vila, Coord. Chem. Rev., 1999, 190–192, 519; (d)
C. Belle and J.-L. Pierre, Eur. J. Inorg. Chem., 2003, 4137; (e)
E. L. Hegg, Acc. Chem. Res., 2004, 37, 775; (f) M. J. Hawk,
R. M. Breece, C. E. Hajdin, K. M. Bender, Z. Hu,
A. L. Costello, B. Bennett, D. L. Tierney and
M. W. Crowder, J. Am. Chem. Soc., 2009, 131, 10753.
2 For reviews, see: (a) M. Shibasaki, M. Kanai, S. Matsunaga
and N. Kumagai, Multimetallic Multifunctional Catalysts
for Asymmetric Reactions, in Bifunctional Molecular
Catalysis, ed. T. Ikariya and M. Shibasaki, 2011, vol. 37, pp.
1–30; (b) B. G. Cooper, J. W. Napoline and C. M. Thomas,
Catal. Rev.: Sci. Eng., 2012, 54, 1; (c) M. Cokoja,
(a) K. Ohmatsu, Y. Hamajima and T. Ooi, J. Am. Chem.
Soc., 2012, 134, 8794; (b) J. Cockrell, C. Wilhelmsen,
H. Rubin, A. Martin and J. B. Morgan, Angew. Chem., Int.
Ed., 2012, 51, 9842; (c) A. Marti, S. Peruncheralathan and
C. Schneider, Synthesis, 2012, 44, 27; (d) A. Marti,
L. Richter and C. Schneider, Synlett, 2011, 2513; (e)
S. Nakamura, M. Hayashi, Y. Kamada, R. Sasaki,
Y. Hiramatsu, N. Shibata and T. Toru, Tetrahedron Lett.,
2010, 51, 3820; (f) S. Peruncheralathan, H. Teller and
C. Schneider, Angew. Chem., Int. Ed., 2009, 48, 4849; (g)
G. D. Sala and A. Lattanzi, Org. Lett., 2009, 11, 3330; (h)
S. Minakata, Y. Murakami, M. Satake, I. Hidaka, Y. Okada
and M. M. Komatsu, Org. Biomol. Chem., 2009, 7, 641; (i)
S. E. Larson, J. C. Baso, G. Li and J. C. Antilla, Org. Lett.,
2009, 11, 5186; (j) E. B. Rowland, G. B. Rowland, E. Rivera-
Otero and J. C. Antilla, J. Am. Chem. Soc., 2007, 129, 12084;
(k) W. H. Leung, W. L. Mak, E. Y. Y. Chan, T. C. H. Lam,
W. S. Lee, H. L. Kwong and L. L. Yeung, Synlett, 2002,
1688; (l) P. Muller and P. Nury, Helv. Chim. Acta, 2001, 84,
662; (m) M. Hayashi, K. Ohno, H. Hoshimi and N. Oguni,
Tetrahedron, 1996, 52, 7817.
¨
C. Bruckmeier, B. Rieger, W. A. Herrmann and F. E. Kuhn,
Angew. Chem., Int. Ed., 2011, 50, 8510; (d) R. M. Haak,
S. J. Wezenberg and A. W. Kleij, Chem. Commun., 2010, 46,
2713; (e) M. Shibasaki, M. Kanai, S. Matsunaga and
N. Kumagai, Acc. Chem. Res., 2009, 42, 1117; (f)
E. Jacobsen, Acc. Chem. Res., 2000, 33, 421; (g) J.-A. Ma and
D. Cahard, Angew. Chem., Int. Ed., 2004, 43, 4566, For a few
notable recent contributions, see the following references
and those cited in them: (h) V. Annamalai, E. F. DiMauro, 13 R. G. Konsler, J. Karl and E. N. Jacobsen, J. Am. Chem. Soc.,
P. J. Carroll and M. C. Kozlowski, J. Org. Chem., 2003, 68,
1998, 120, 10780.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1115

View Article Online
Chemical Science
Edge Article
14 B. W. McCleland, W. A. Nugent and M. G. Finn, J. Org. Chem., 30 See ESI for details†
1998, 63, 6656. Similar models have been invoked to 31 B. Das, R. Ramu, B. Ravikanth and K. R. Reddy, Tetrahedron
rationalize reactivity and selectivity in other nucleophile– Lett., 2006, 47, 779.
electrophile combinations, although rigorous evidence for 32 B. Das, V. S. Reddy, R. Ramu and D. N. Kumar, Catal.
the operation of bimetallic mechanisms is seldom offered, Commun., 2006, 7, 997.
see ref. 2. For some notable exceptions, see: (a) R. G. Konsler, 33 J. S. Yadav, B. V. S. Reddy, B. Jyothirmai and M. S. R. Murty,
J. Karl and E. N. Jacobsen, J. Am. Chem. Soc., 1998, 120, Tetrahedron Lett., 2005, 46, 6385.
10780; (b) Y. N. Belokon, B. Green, N. S. Ikonnikov, 34 M. S. Reddy, M. Narender, Y. V. D. Nageswar and K. R. Rao,
V. S. Larichev, B. V. Lokshin, M. A. Moscalenko, M. North, Tetrahedron Lett., 2005, 46, 6437.
C. Orizu, A. S. Peregudov and G. I. Timofeeva, Eur. J. Org. 35 S. Peruncheralathan, M. Henze and C. Schneider,
Chem., 2000, 2655; (c) Q.-X. Guo, Z.-J. Wu, Z.-B. Luo, Tetrahedron Lett., 2007, 48, 6743.
Q.-Z. Liu, J.-L. Ye, S.-W. Luo, L.-F. Cun and L.-Z. Gong, J. Am. 36 N. Prusinowska and J. Gawronski, Synth. Commun., 2009, 39,
Chem. Soc., 2007, 129, 13927; (d) C. Mazet and
E. N. Jacobsen, Angew. Chem., Int. Ed., 2008, 47, 1762.
15 B. Saha, M.-H. Lin and T. V. RajanBabu, J. Org. Chem., 2007,
72, 8648.
2795.
37 Z. Wang, X. Sun, S. Ye, W. Wang, B. Wang and J. Wu,
Tetrahedron: Asymmetry, 2008, 19, 964.
38 G. D. Sala and A. Lattanzi, Org. Lett., 2009, 11, 3330.
16 R. M. Thomas, P. C. B. Widger, S. M. Ahmed, R. C. Jeske, 39 S. E. Larson, J. C. Baso, G. Li and J. C. Antilla, Org. Lett., 2009,
W. Hirahata, E. B. Lobkovsky and G. W. Coates, J. Am.
Chem. Soc., 2010, 132, 16520.
11, 5186.
40 E. N. Jacobsen, Acc. Chem. Res., 2000, 33, 421.
17 J. A. Kalow and A. G. Doyle, J. Am. Chem. Soc., 2011, 133, 41 M. Shibasaki, S. Matsunaga and N. Kumagai, Synlett, 2008,
16001. 1583.
18 B. Wu, J. R. Parquette and T. V. RajanBabu, Science, 2009, 42 R. Anwander, Principles of Organolanthanide Chemistry, in
326, 1662.
Lanthanides: Chemistry and Use in Organic Synthesis, ed. S.
Kobayashi, Springer, Berlin, 1999, vol. 2, pp. 1–61.
19 M.-H. Lin and T. V. RajanBabu, Org. Lett., 2002, 4, 1607.
20 Y. Hamashima, D. Sawada, M. Kanai and M. Shibasaki, 43 L. C. Miller and R. Sarpong, Chem. Soc. Rev., 2011, 40, 4550.
J. Am. Chem. Soc., 1999, 121, 2641.
21 H. Kitajima, K. Ito, Y. Aoki and T. Katsuki, Bull. Chem. Soc.
Jpn., 1997, 70, 207.
44 R. R. Kumar and H. B. Kagan, Adv. Synth. Catal., 2010, 352,
231.
45 E. Vedejs and M. Jure, Angew. Chem., Int. Ed., 2005, 44, 3974.
22 M. Hayashi, K. Ohno, H. Hoshimi and N. Oguni, 46 J. R. Dehli and V. Gotor, Chem. Soc. Rev., 2002, 31, 365.
Tetrahedron, 1996, 52, 7817.
23 S. A. Schuetz, M. A. Erdmann, V. W. Day, J. L. Clark and
J. A. Belot, Inorg. Chim. Acta, 2004, 357, 4045.
47 A. Gansauer, C. A. Fan, F. Keller and P. Karbaum, Chem.–Eur.
J., 2007, 13, 8084.
48 Examples of chemo- and regiodivergent kinetic resolutions
in which constitutionally different products are formed are
still relatively rare. For notable examples, see: (a)
R. Webster, A. Boyer, M. J. Fleming and M. Lautens, Org.
¨
24 (a) Ti: G. Boche, K. Mobus, K. Harms and M. Marsch, J. Am.
Chem. Soc., 1996, 118, 2770, Y: (b) (Ln)Y(m-Cl)]2 (several
chloride bridged dimers): W. J. Evans, C. H. Fujimoto and
J. W. Ziller, Chem. Commun., 1999, 311; (c)
C. M. Mascarenhas, S. P. Miller, P. S. White and
J. P. Morken, Angew. Chem., Int. Ed., 2001, 40, 601(L)(m-
OH)]2 (a hydroxide bridged dimer): ; (d) T. M. Ovitt and
G. W. Coates, J. Am. Chem. Soc., 2002, 124, 1316(L)(m-OR)]2
(an alkoxide bridged dimer): . CN-bridged complexes: ; (e)
C. E. Plecnik, S. Liu and S. S. Shore, Acc. Chem. Res., 2003,
36, 499; (f) S. Tanase, Coord. Chem. Rev., 2006, 250, 2501.
Azide bridged complexes:; (g) H. Schumann, C. Janiak and
J. Pickardt, J. Organomet. Chem., 1988, 349, 117; (h)
H.-C. Wang, M. Xue, Q. Guo, J.-P. Zhao, F.-C. Liu and
J. Ribas, J. Solid State Chem., 2012, 187, 143; (i)
X.-C. Huang, C. Zhou, H.-Y. Wei and X.-Y. Wang, Inorg.
Chem., 2013, 52, 7314.
¨
Lett., 2010, 12, 5418; (b) R. Webster, C. Boing and
M. Lautens, J. Am. Chem. Soc., 2009, 131, 444; (c) C. K. Jana
and A. Studer, Angew. Chem., Int. Ed., 2007, 46, 6542; (d)
A. Gansauer, C. A. Fan, F. Keller and J. Keil, J. Am. Chem.
Soc., 2007, 129, 3484; (e) H. M. L. Davies, X. Dai and
M. S. Long, J. Am. Chem. Soc., 2006, 128, 2485; (f)
M. Pineschi, F. Del Moro, P. Crottti, V. Di Bussolo and
F. Macchia, J. Org. Chem., 2004, 69, 2099; (g) K. Tanaka
and K. F. C. Fu, J. Am. Chem. Soc., 2003, 125, 8078; (h)
F. Bertozzi, P. Crotti, F. Macchia, M. Pineschi and
B. L. Feringa, Angew. Chem., Int. Ed., 2001, 40, 930; (i)
G. R. Cook and S. Sankaranarayanan, Org. Lett., 2001, 3,
3531; (j) B. D. Brandes and E. N. Jacobsen, Synlett, 2001,
1013; (k) Y. Chen and L. Deng, J. Am. Chem. Soc., 2001,
123, 11302; (l) C. Bolm, O. Beckmann and C. Palazzi, Can.
J. Chem., 2001, 79, 1593; (m) M. S. Visser and
A. H. Hoveyda, Tetrahedron, 1995, 51, 4383; (n) M. P. Doyle,
A. B. Dyatkin, A. V. Kalinin, D. A. Ruppar, S. F. Martin,
M. R. Spaller and S. Liras, J. Am. Chem. Soc., 1995, 117,
11021; (o) C. Bolm and G. Schlingloff, J. Chem. Soc., Chem.
Commun., 1995, 1247; (p) S. F. Martin, M. R. Spaller,
S. Liras and B. Hartmann, J. Am. Chem. Soc., 1994, 116, 4493.
25 S. L. McLain, T. M. Ford and N. E. Drysdale, Polym. Prepr.
(Am. Chem. Soc., Div. Polym. Chem.), 1992, 33, 463.
26 Ref. 24(d).
27 See ESI.†
28 Y. Fukuta, T. Mita, N. Fukuda, M. Kanai and M. Shibasaki,
J. Am. Chem. Soc., 2006, 128, 6312.
29 E. B. Rowland, G. B. Rowland, E. Rivera-Otero and
J. C. Antilla, J. Am. Chem. Soc., 2007, 129, 12084.
1116 | Chem. Sci., 2014, 5, 1102–1117
This journal is © The Royal Society of Chemistry 2014

View Article Online
Edge Article
Chemical Science
49 H. B. Kagan and J. C. Fiaud, Kinetic Resolution, in Topics in
Stereochemistry, ed. E. L. Eliel and S. H. Wilen, John Wiley,
1988, vol. 18, pp. 249–330.
4-nitrobenzoyl chloride. The enantiopure N-acyl aziridines
were prepared from the corresponding a-amino acids via
LiAlH4 reduction, Mitsunobu reaction and benzoylation.
See ref. 18 for details.
50 S. E. Schaus, B. D. Brandes, J. F. Larrow, M. Tokunaga,
K. B. Hansen, A. E. Gould, M. E. Furrow and 59 For other examples, see: (a) D. J. Berrisford, C. Bolm and
E. N. Jacobsen, J. Am. Chem. Soc., 2002, 124, 1307.
51 M. Tokunaga, J. Kiyosu, Y. Obora and Y. Tsuji, J. Am. Chem.
Soc., 2006, 128, 4481.
K. B. Sharpless, Angew. Chem., Int. Ed., 1995, 34, 1059; (b)
In yttrium chemistry: M.-H. Lin and T. V. RajanBabu, Org.
Lett., 2002, 4, 1607.
52 M. J. Burk and J. G. Allen, J. Org. Chem., 1997, 62, 7054–7057. 60 N. W. Mitzel, J. Riede and C. Kiener, Angew. Chem., Int. Ed.,
The b-amido–azide from 7 has been converted into trans-1,2- 1997, 36, 2215.
diaminocyclohexane by trans-amidation with (boc)2O/ 61 (a) For another recent example in the Ti(III)-mediated
DMAP, acid-mediated deprotection and hydrogenation of
the resulting azide.28
epoxide-opening, see: A. Gansauer, L. Shi and M. Otte, J.
Am. Chem. Soc., 2010, 132, 11858; (b) Reference to the
original Ti(III)-mediated epoxide opening reaction:
T. V. RajanBabu and W. A. Nugent, J. Am. Chem. Soc., 1994,
116, 986.
53 K. Faber, Chem.–Eur. J., 2001, 7, 5004.
54 U. T. Strauss, U. Felfer and K. Faber, Tetrahedron: Asymmetry,
1999, 10, 107.
55 Yields of more than 50% can be obtained in special cases 62 See ESI for the 1H, 13C, 1H–1H COSY, HMQC, time-
where the epoxide can undergo racemization under
hydrolytic conditions. This is an example of the well-
known dynamic kinetic resolution, and is distinctly
dependent DOSY spectra of 10b. DOSY spectra of 2b
recorded at time intervals t ¼ 0, 7.5, 12 and 23 h are also
included there.†
different from the strategy outlined here. See, ref. 50. This 63 Experimental Organometallic Chemistry, ACS Symposium
paper also illustrates a rare instance in which the slow
reacting enantiomer from a classical kinetic resolution
Series, ed. A. L. Wayda and M. Y. Darensbourg, American
Chemical Society, Washington, DC, 1987.
with exceptionally high kf/ks is transformed into the same 64 S. Tosaki, R. Tsuji, T. Ohshima and M. Shibasaki, J. Am.
initial product in a subsequent chemical reaction. Chem. Soc., 2005, 127, 2147.
56 S. Hwang, C. Y. Choi and E. Y. Lee, Biotechnology and 65 B. Baccar, J. Mol. Struct., 1971, 7, 369.
Bioprocess Engineering, 2008, vol. 13, p.453.
57 (a) W. Kroutil, M. Mischitz and K. Faber, J. Chem. Soc., Perkin
66 See ESI† for the IR spectrum in the range 1560–1720 cmꢁ1
and the time-dependent prole of the peak at 1648 cmꢁ1
,
.
Trans. 1, 1997, 3629; (b) For the conversion of a b- 67 X.-C. Huang, C. Zhou, H.-Y. Wei and X.-Y. Wang, Inorg.
acylamidoazide to a 1,2-diamine, see ref. 28. Chem., 2013, 52, 7314, and references cited therein.
58 (a) See ESI† for experimental details, spectroscopic and 68 H.-C. Wang, M. Xue, Q. Guo, J.-P. Zhao, F.-C. Liu and
analytical data including chromatograms from chiral J. Ribas, J. Solid State Chem., 2012, 187, 143.
stationary phase separations. The enantiomeric excesses 69 H. Schumann, C. Janiak and J. Pickardt, J. Organomet. Chem.,
were determined by chiral phase liquid chromatography 1988, 349, 117.
where base-line separation of the enantiomers were 70 I. Yu, A. Acosta-Ramirez and P. Mehrkhodavandi, J. Am.
observed; (b) The starting racemic aziridines were prepared Chem. Soc., 2012, 134, 12758.
by reaction of the appropriate alkene substrate with IN3 71 M. R. Kember, F. Jutz, A. Buchard, A. J. P. White and
followed by reduction with LiAlH4 and acylation with
C. K. Williams, Chem. Sci., 2012, 3, 1245.
This journal is © The Royal Society of Chemistry 2014
Chem. Sci., 2014, 5, 1102–1117 | 1117