Asymmetric Schmidt reaction of hydroxyalkyl azides with ketones

Article abstract of DOI:10.1021/ja0348896

An asymmetric equivalent of the Schmidt reaction permits stereocontrol in ring expansions of symmetrical cyclohexanones. The procedure involves the reaction of chiral 1,2- and 1,3-hydroxyalkyl azides with ketones under acid catalysis; the initial reaction affords an iminium ether that can be subsequently opened with base. A systematic study of this reaction is reported, in which ketone substrates, chiral hydroxyalkyl azides, and reaction conditions are varied. Selectivities as high as ca. 98:2 are possible for the synthesis of substituted caprolactams, with up to 1,7-stereoselection involved in the overall process. The fact that either possible migrating carbon is electronically identical provides an unusual opportunity to study a ring-expansion reaction controlled entirely by stereoelectronic factors. The mechanism of the reaction and the source of its stereoselectivity are also discussed.

Full text of DOI:10.1021/ja0348896

Published on Web 05/28/2003
Asymmetric Schmidt Reaction of Hydroxyalkyl Azides with
Ketones
Kiran Sahasrabudhe, Vijaya Gracias, Kelly Furness, Brenton T. Smith,
Christopher E. Katz, D. Srinivasa Reddy, and Jeffrey Aube´*
Contribution from the Department of Medicinal Chemistry, 1251 Wescoe Hall DriVe,
Malott Hall, Room 4070, UniVersity of Kansas, Lawrence, Kansas 66045-2506
Received February 26, 2003; E-mail: jaube@ku.edu
Abstract: An asymmetric equivalent of the Schmidt reaction permits stereocontrol in ring expansions of
symmetrical cyclohexanones. The procedure involves the reaction of chiral 1,2- and 1,3-hydroxyalkyl azides
with ketones under acid catalysis; the initial reaction affords an iminium ether that can be subsequently
opened with base. A systematic study of this reaction is reported, in which ketone substrates, chiral
hydroxyalkyl azides, and reaction conditions are varied. Selectivities as high as ca. 98:2 are possible for
the synthesis of substituted caprolactams, with up to 1,7-stereoselection involved in the overall process.
The fact that either possible migrating carbon is electronically identical provides an unusual opportunity to
study a ring-expansion reaction controlled entirely by stereoelectronic factors. The mechanism of the reaction
and the source of its stereoselectivity are also discussed.
Achiral ketones can be subjected to group-selective asym-
metric transformations leading to a variety of usefully func-
tionalized derivatives. Such reactions are of particular synthetic
utility because they provide ready access to medium-ring
compounds, many of which are only made with difficulty using
other methods.1 Much recent attention in this field has focused
on asymmetric deprotonation2 and ring-expansion processes. In
the latter, the formal insertion of a group X between the carbonyl
and one of the enantiotopic methylene groups in a ketone such
as 4-methylcyclohexanone can afford an asymmetric synthesis
of a seven-membered hetero- or carbocyclic product (eq 1).
Likewise, only a few methods for the asymmetric conversion
of ketones to lactams exist. Previously described methods that
furnish enantiomerically enriched nitrogen-insertion products
include the rearrangement of chiral oximes via the Beckmann
rearrangement7 and the photochemical rearrangement of chiral
oxaziridines.8 The paucity of methods available for the ste-
reospecific synthesis of enantiopure oximes9 limits the utility
of the Beckmann rearrangement for this purpose. Unlike oximes,
oxaziridines can be prepared easily in diastereo- and enantio-
(3) (a) Taschner, M. J.; Black, D. J. J. Am. Chem. Soc. 1988, 110, 6892-
6893. (b) Taschner, M. J.; Chen, Q.-Z. Bioorg. Med. Chem. Lett. 1991, 1,
535-538. (c) Taschner, M. J.; Black, D. J.; Chen, Q.-Z. Tetrahedron:
Asymm. 1993, 4, 1387-1390. (d) Lopp, M.; Paju, A.; Kanger, T.; Pehk, T.
Tetrahedron Lett. 1996, 37, 7583-7586. (e) Stewart, J. D.; Reed, K. W.;
Zhu, J.; Chen, G.; Kayser, M. M. J. Org. Chem. 1996, 61, 7652-7653. (f)
Bolm, C.; Luong, T. K. K.; Schlingloff, G. Synlett 1997, 1151-1152. (g)
Kayser, M. M.; Chen, G.; Stewart, J. D. J. Org. Chem. 1998, 63, 7103-
7106. (h) Stewart, J. D.; Reed, K. W.; Martinez, C. A.; Zhu, J.; Chen, G.;
Kayser, M. M. J. Am. Chem. Soc. 1998, 120, 3541-3548. (i) Stewart, J.
D. Curr. Org. Chem 1998, 2, 195-216. (j) Mihovilovic, M. D.; Chen, G.;
Wang, S.; Kyte, B.; Rochon, F.; Kayser, M. M.; Stewart, J. D. J. Org.
Chem. 2001, 66, 733-738.
Asymmetric Baeyer-Villiger processes using enzymatic reac-
tions,3 metal-promoted catalysis,4 or multistep equivalents
thereof5 have received the greatest share of attention in this
regard. In contrast, very few carbon-based versions of this
chemistry have been introduced to date.6
(4) (a) Bolm, C.; Schlingloff, G.; Weickhardt, K. Angew. Chem., Int. Ed. Engl.
1994, 33, 1848-1849. (b) Bolm, C.; Luong, T. K. K.; Schlingloff, G. Synlett
1997, 1151-1152. (c) Paneghetti, C.; Gavagnin, R.; Pinna, F.; Strukul, G.
Organometallics 1999, 18, 5057-5065.
(5) (a) Honda, T.; Kimura, N.; Tsubuki, M. Tetrahedron: Asymm. 1993, 4,
1475-1478. (b) Sugimura, T.; Fujiwara, Y.; Tai, A. Tetrahedron Lett. 1997,
38, 6019-6022.
(1) Reviews of medium rings and their synthesis: (a) Roxburgh, C. J.
Tetrahedron 1993, 49, 10 749-10 784. (b) Rousseau, G. Tetrahedron 1995,
51, 2777-2849. (c) Tochtermann, W.; Kraft, P. Synlett 1996, 1029-1035.
(d) Molander, G. A. Acc. Chem. Res. 1998, 31, 603-609. (e) Yet, L.
Tetrahedron 1999, 55, 9349-9403. (f) Maier, M. E. Angew. Chem., Int.
Ed. 2000, 39, 2073-2077. (g) Yet, L. Chem. ReV. 2000, 100, 2963-3007.
(h) Nubbemeyer, U. Top. Curr. Chem. 2001, 216, 125-196.
(2) For representative examples, see: (a) Cox, P. J.; Simpkins, N. S.
Tetrahedron: Asymm. 1991, 2, 1-26. (b) Izawa, H.; Shirai, R.; Kawasaki,
H.; Kim, H.-d.; Koga, K. Tetrahedron Lett. 1992, 30, 7221-7224. (c)
Coggins, P.; Gaur, S.; Simpkins, N. S. Tetrahedron Lett. 1995, 36, 1545-
1548. (d) Majewski, M.; Lazny, R. J. Org. Chem. 1995, 60, 5825-5830.
(e) Toriyama, M.; Sugasawa, K.; Shindo, M.; Tokutake, N.; Koga, K.
Tetrahedron Lett. 1997, 38, 567-570. (f) Lee, J. C.; Lee, K.; Cha, J. K. J.
Org. Chem. 2000, 65, 4773-4775. (g) A number of papers on this topic
have appeared in a Symposium-in-Print on lithium amide base chemistry.
O’Brien, P., Guest Ed., Tetrahedron, 2002, 58, 4657-4727.
(6) (a) Nemoto, H.; Tanabe, T.; Fukumoto, K. Tetrahedron Lett. 1994, 35,
6499-6502. (b) Nemoto, H.; Tanabe, T.; Fukimoto, K. J. Org. Chem. 1995,
60, 6785-6790.
(7) (a) Lyle, R. E.; Lyle, G. G. J. Org. Chem. 1959, 24, 1679-1684. (b) Lyle,
G. G.; Barrera, R. M. J. Org. Chem. 1964, 29, 3311-3314. (c) Toda, F.;
Akai, H. J. Org. Chem. 1990, 55, 4973-4974.
(8) (a) Lattes, A.; Oliveros, E.; Rivie`re, M.; Belzecki, C.; Mostowicz, D.;
Abramskj, W.; Piccinni-Leopardi, C.; Germain, G.; Van Meerssche, M. J.
Am. Chem. Soc. 1982, 104, 3929-3934. (b) Aube´, J.; Burgett, P. M.; Wang,
Y. Tetrahedron Lett. 1988, 29, 151-154. (c) Aube´, J. Tetrahedron Lett.
1988, 29, 4509-4512. (d) Aube´, J.; Wang, Y.; Hammond, M.; Tanol, M.;
Takusagawa, F.; Vander Velde, D. J. Am. Chem. Soc. 1990, 112, 4879-
4891. (e) Aube´, J. Chem. Soc. ReV. 1997, 26, 269-277.
(9) Adamczyk, M.; Chen, Y.-Y.; Fishpaugh, J. R.; Gebler, J. C. Tetrahedron:
Asymm. 1993, 4, 1467-1468.
9
7914
J. AM. CHEM. SOC. 2003, 125, 7914-7922
10.1021/ja0348896 CCC: $25.00 © 2003 American Chemical Society

Asymmetric Schmidt Reaction
A R T I C L E S
Scheme 1
hydroxyalkyl azide reaction partners. The origin of the stereo-
selectivity of the reaction will also be discussed.
Results
Selection and Synthesis of Hydroxyalkyl Azides. Previous
work10 had shown that the in-situ tethering process worked best
with 1,2- and 1,3-hydroxyalkyl azides, so a representative series
of mono- and disubstituted reagents 1-12 were targeted for
study. These specific derivatives represent a selection of
variously substituted compounds bearing substituents of differing
steric demands. In some cases, diastereomers or constitutional
isomers were separately prepared to test specific mechanistic
issues. Although most of these compounds were prepared in
enantiomerically enriched form, some reagents were used as
racemates and are so indicated.
Scheme 2
merically enriched form; once made, they undergo photochemi-
cally induced, regioselective ring expansion reactions. A typical
reaction sequence is shown in Scheme 1. Using R-methylben-
zylamine as a stereocontrolling group, the overall selectivity
for the conversion of ketones to lactams reached a maximum
of ca. 88% ds.8d Enantiomerically pure lactams were obtained
by separation of the diastereomers and removal of the chiral
group on the nitrogen. Besides the relatively low stereoselec-
tivity, this methodology suffered from the need to carry out
three discrete chemical steps and moderate overall yields.
Recent work in this laboratory has circumvented these issues
by using alkyl azides in nitrogen ring-expansion reactions. In
particular, 1,2- and 1,3-hydroxyalkyl azides have proven
especially useful for the conversion of ketones to lactams via
an “in situ-tethering” strategy.10 Activation of the carbonyl group
by a protic or Lewis acid promotes the initial formation of a
hemiketal and dehydration to the oxenium ion shown (Scheme
2). Intramolecular attack by the azido group followed by
rearrangement and the loss of N2 yields an intermediate iminium
ether that can be purified or subjected to hydrolysis in a one-
pot process. This overall process is dubbed “in situ-tethering”
because alkyl azides are poor nucleophiles in intermolecular
reactions with ketones11b,d and because the tethering step can
be carried out under the same conditions used for the rear-
rangement reaction. These reactions are very closely related to
azido-Schmidt reactions with ketone equivalents11 and carboca-
tions.12
An attractive aspect of the proposed ring-expansion protocol
was the ready availability of chiral hydroxyalkyl azides from
commercial starting materials using simple chemical transfor-
mations. In general, 1,2-azido alcohols were made by ring
opening of epoxides or via analogous reactions using activated
diols (Scheme 3). In some cases, a Mitsunobu procedure carried
out on the diol directly afforded the azido alcohol without the
need for protecting groups or additional activation. This
procedure was more convenient for providing 1 in high ee
relative to the known ring opening of styrene oxide, in part
because the desired regioisomer is formed in high yield as the
major product (cf., the synthesis of 3). Bittman and co-workers
have reported a related method for the stereoselective azidation
of 1,2- and 1,3-diols using Mitsunobu chemistry.14 The ready
synthesis of highly enantiomerically enriched 1,2-diols by
asymmetric dihydroxylation chemistry15 ensures that a wide
structural variety of chiral 1,2-azido alcohols is available.
In preliminary work, we reported that chiral hydroxy alkyl
azides can lead to highly selective nitrogen asymmetric ring
expansion reactions as shown in eq 1 above.10a,13 In this paper,
we report studies on the scope and utility of this process focusing
on the effects of structural variation in both the ketone and
(12) (a) Pearson, W. H.; Schkeryantz, J. M. Tetrahedron Lett. 1992, 33, 5291-
5294. (b) Pearson, W. H.; Walavalkar, R.; Schkeryantz, J. M.; Fang, W.-
k.; Blickensdorf, J. D. J. Am. Chem. Soc. 1993, 115, 10 183-10 194. (c)
Pearson, W. H.; Fang, W.-k.; Kampf, J. W. J. Org. Chem. 1994, 59, 2682-
2684. (d) Pearson, W. H.; Fang, W.-k. J. Org. Chem. 1995, 60, 4960-
4961. (e) Pearson, W. H. J. Het. Chem. 1996, 33, 1489-1496. (f) Pearson,
W. H.; Gallagher, B. M. Tetrahedron 1996, 52, 12 039-12 048. (g) Pearson,
W. H.; Fang, W.-k. Isr. J. Chem. 1997, 37, 39-46. (h) Pearson, W. H.;
Fang, W.-k. J. Org. Chem. 2000, 65, 7158-7174.
(10) (a) Gracias, V.; Milligan, G. L.; Aube´, J. J. Am. Chem. Soc. 1995, 117,
8047-8048. (b) Gracias, V.; Frank, K. E.; Milligan, G. L.; Aube´, J.
Tetrahedron 1997, 53, 16 241-16 252. (c) Forsee, J. E.; Smith, B. T.; Frank,
K. E.; Aube´, J. Synlett 1998, 1258-1260. (d) Smith, B. T.; Gracias, V.;
Aube´, J. J. Org. Chem. 2000, 65, 3771-3774.
(11) (a) Aube´, J.; Milligan, G. L. J. Am. Chem. Soc. 1991, 113, 8965-8966.
(b) Aube´, J.; Milligan, G. L.; Mossman, C. J. J. Org. Chem. 1992, 57,
1635-1637. (c) Mossman, C. J.; Aube´, J. Tetrahedron 1995, 52, 3403-
3408. (d) Milligan, G. L.; Mossman, C. J.; Aube´, J. J. Am. Chem. Soc.
1995, 117, 10449-10459. (e) Desai, P.; Schildknegt, K.; Agrios, K. A.;
Mossman, C.; Milligan, G. L.; Aube´, J. J. Am. Chem. Soc. 2000, 122, 7226-
7232. (f) Wrobleski, A.; Aube´, J. J. Org. Chem. 2001, 66, 886-889.
(13) Furness, K.; Aube´, J. Org. Lett. 1999, 1, 495-497.
(14) He, L.; Wanunu, M.; Byun, H.-S.; Bittman, R. J. Org. Chem. 1999, 64,
6049-6055.
(15) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. ReV. 1994,
94, 2483-2547.
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7915

A R T I C L E S
Sahasrabudhe et al.
Scheme 3
Scheme 5
Scheme 6
Scheme 4
Table 1. Reactions of 1 with 4-Methylcyclohexanone
ratio of
14a: 14bb
entry
solvent
THF
T (°C)
time (h)
yield (%)a
1
2
3
4
5
6
7
8
9
-78f rt
0
36
21
18
18
18
100
21
18
0
3c
74
-
Et2O
73:27
73:27
69:31
83:17
70:30
69:31
78:22
78:22
The 1,3-azido alcohols were synthesized by a variety of
techniques (Scheme 4). In some cases, simple azide displace-
ment of an alkyl halide provided the desired reagent in a single
step. For example, treatment of commercially available (R)-3-
chloro-1-phenyl-1-propanol and (S)-3-bromo-2-methyl-1-pro-
panol with NaN3 directly led to enantiomerically pure 5 and 8,
respectively. (S)-3-Chloro-1-phenyl-1-propanol could also be
converted to the phenyl-containing (R)-6 by displacement of
the chloride with acetate, hydrolysis, and Mitsunobu azida-
tion (inversion) of the benzylic center. Conversion of the C2-
symmetrical chiral diol (R,R)-2,4-dihydroxypentane to the
corresponding azide was accomplished by Mitsunobu inversion
of one of the homotopic hydroxyl groups, affording 11.
Mitsunobu reactions were also used to prepare compounds 7
and 10 in racemic form (routes not shown; see the Supporting
Information).
cyclohexane
CH3CN
CH2Cl2
CH2Cl2
CH2Cl2
CCl4
3
-30
-30
-30
0
-20
-20
41
15c
62
86
97
100
n-pentane
23
a Isolated yields except where noted. b Ratios determined by HPLC of
crude reaction mixtures; see text and Supporting Information for stereo-
structure determinations. c Yield for this example estimated from HPLC
trace.
azide 1 and 4-methylcyclohexanone (Scheme 6). Our initial
studies with achiral hydroxyalkyl azide addition reactions
established BF3‚Et2O as a superior Lewis acid for the nitrogen-
insertion reactions.10 Thus, 1.5 equiv of 1 and 3.0 equiv of BF3‚
Et2O were added to ketone under the conditions shown in Table
1. Presumably, an excess of Lewis acid is needed to accom-
modate the disproportionation of the reagent into tetrafluorobo-
rate; see below. After the reactions were carried out for the time
noted, the intermediate iminium ether 13 was hydrolyzed by
the addition of aqueous NaHCO3 or KOH, affording 14a,b as
Compounds 9 and 12 were synthesized using the multistep
pathways shown in Scheme 5. These reagents were needed for
certain stereochemical assignments and for mechanistic studies.
Reactions of 1,2-Hydroxyalkyl Azides. Optimal solvent and
temperature conditions were determined using the hydroxyalkyl
9
7916 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Asymmetric Schmidt Reaction
A R T I C L E S
Table 2. Reactions of 1,2-Hydroxyethyl Azides with 4-Substituted
Ketones
entry
azide
ketone R
major product
yield (%)a
ratio a:bb
1
2
3
4
5
1
1
2
Me
t-Bu
Me
t-Bu
t-Bu
14a
15a
16a
97
83
51
64
73
78:22
85:15
87:13
88:12
56:44
2
17a
3c
18a or 18bd
Figure 1. Ball-and-stick depiction of the X-ray crystallographic analysis
of iminium ether 13, showing the correct relative stereochemistry of the
intermediate. Note that the enantiomeric series of the crystal structure was
arbitrarily assigned and so this drawing depicts the enantiomer of 13 as
synthesized from (R)-1.
a Isolated yield. b Ratio determined by HPLC trace of the crude reaction
mixture; see text and Supporting Information for stereostructure determina-
tions. c This experiment was done using a racemic hydroxyalkyl azide. d Not
assigned for this example.
a readily separated mixture of diastereomers (column chroma-
tography). The products were assigned by removing the nitrogen
substituent to afford the known 5-substituted caprolactam,8b,d
and measuring its specific rotation. For the series of N-phenyl-
(hydroxyethyl) derivatives examined, this step generally oc-
curred in g80% yield and resulted in the convenient removal
of the chiral induction moiety from the product.
CCl4 or 1:3 n-pentane/CCl4), furnished lactams in the yields
and selectivities shown in Table 2. The reactions of 4-methyl-
cyclohexanone and 4-tert-butylcyclohexanone with azides 1 and
2 proceeded with high levels of stereoselectivity, whereas the
reaction of azide 3 with 4-tert-butylcyclohexanone proceeded
with poor selectivity.
Reactions of 1,3-Hydroxyalkyl Azides. The reactions of 1,3-
hydroxyalkyl azides with substituted cyclohexanones proved
more facile than their 1,2-disubstituted cousins, taking place at
lower temperatures. A brief optimization exercise allowed us
to identify standard conditions of -82 °C to 25 °C, BF3‚Et2O,
CH2Cl2, which were applied to various combinations of 4-meth-
yl- or 4-tert-butylcyclohexanone and hydroxyalkyl azides 5-10
(Table 3). As before, reactions were allowed to warm to room
temperature to ensure completion.
The stereoisomeric products were generally separable and the
structures of major isomers determined by conversion to the
known8a,b,d N-dealkylated lactams. For lactams 22 and 23, this
was accomplished by dissolving metal reduction (cf., Scheme
6). The removal of nonbenzylic lactam side chains was carried
out by oxidation to the aldehyde or ketone followed by
elimination of the lactam through treatment with base (Scheme
7). In the case of lactam 21a, shown, it was additionally
necessary to establish that hydrolysis did not occur by attack at
the benzylic position (which has been observed with other
nucleophiles16 and would here lead to the enantiomeric form
of 21b; Scheme 7). This was established through X-ray
crystallographic analysis of 21a, which nailed down the relative
stereochemistry of this lactam and, with the absolute chemistry
as ascertained through degradation of 21, fully established the
stereochemical course of this reaction. Such precautions were
not necessary for lactams 26 and 28 because the tether was
terminated with a primary alcohol group. Since the stereose-
lectivity of each case derived from 4-methylcyclohexanone
closely mirrored those obtained in the tert-butyl series, the
directionality of the former reactions were assigned by analogy,
as were lactams 24, 29, and 30.
Although crude solutions of the iminium ether were generally
hydrolyzed by direct addition of base, in some cases the
intermediate iminium ethers were isolated. Concentration of the
reaction mixture of 1 and 4-methylcylohexanone followed by
chromatography with CH2Cl2/MeOH afforded a solid that could
be recrystallized from 3:1 hexane/EtOAc to afford X-ray quality
crystals of 13 as the tetrafluoroborate salt (Figure 1). The
intermediacy of iminium ethers in this chemistry had previously
been proposed based on spectroscopic evidence, elemental
analysis, and on their reactions with other nucleophiles.10,16 This
structure, which was the first we were able to obtain in this
series, clearly indicates relative stereochemistry and the presence
-
of the BF4 counterion. The latter was presumably formed by
disproportionation of the Lewis acid, which is used in excess;
however, neither the details of this disproportionation nor the
presence of other boron species were explicitly examined.
A survey of reaction conditions showed that Et2O and THF
were inappropriate media for this reaction (Table 1). Methylene
chloride and hydrocarbon solvents were considerably better, with
less polar solvents such as n-pentane and CCl4 affording the
highest yields in the shortest time. The optimum temperature
range for the reactions was determined to be between -30 °C
and -20 °C, although it was necessary to allow the reaction to
warm to room temperature and to carry out the iminium ether
hydrolyses at room temperature to obtain the stated yields. Most
importantly, the selectivities obtained using a hydroxyethyl azide
were comparable to the best obtained using the multistep
oxaziridine route (ratios ranging from 74:26 to 84:16 were
obtained for 4-methylcyclohexanone using the latter8d), sug-
gesting that the Schmidt-based technique would be competitive
with earlier technology.
The reactions of chiral 1,3-hydroxyalkyl azides were generally
more selective than those of their 1,2-disposed counterparts. For
example, a 93:7 ratio of isomers was obtained from the reaction
of hydroxyalkyl azide 5 with 4-methylcyclohexanone (Table
3, entry 1). Although results differed according to the structure
The reactions of 4-methylcyclohexanone and 4-tert-butylcy-
clohexanone with the hydroxyalkyl azides 1, 2 and 3, carried
out under these optimized conditions (-20 °C, n-pentane or
(16) Gracias, V.; Milligan, G. L.; Aube´, J. J. Org. Chem. 1996, 61, 10-11.
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7917

A R T I C L E S
Sahasrabudhe et al.
Table 3. Reactions of Monosubstituted 1,3-Hydroxyalkyl Azides
Scheme 7
with 4-Substituted Cyclohexanones
azide
entry ketone R azide
R
1
R
2
R
3
products ratioa:ba yield (%)b
1
2
3
4
5
6
7
8
9
Me
Ph
5
5
5
H
H
H
H
Ph
Ph
Ph
H
H
H
H
H
H
H
19
20
21
93:7
96:4
95:5
98
99
100
96
94
90
93
98
93
98
88
85
H
H
H
H
H
Me
Me
Ph
t-Bu
Me
t-Bu
t-Bu
Me
6c Ph
6c Ph
22c
23c
24d
25
89:11
90:10
90:10d,e
78:22
74:26
60:40
60:40
88:12d
88:12d
7d -C6H(OMe)3
8
8
H
H
H
H
H
H
t-Bu
Me
26
9c
27c
28c
29d
30d
10 t-Bu
11 Me
12 t-Bu
9c
Ph
i-Pr
i-Pr
10d
10d
H
H
a Ratio determined by HPLC of crude reaction mixtures, except where
noted. See text and Supporting Information for stereostructure determina-
tions. b Total yields of isolated purified lactams. c This reaction was done
in the enantiomeric series to that shown in this table. It is depicted in this
manner to allow easy comparison between examples. See the text and
Supporting Information for the correct enantiomers. d This experiment was
done using a racemic hydroxyalkyl azide. e Ratio estimated by 1H NMR
examination of the crude reaction mixture.
Scheme 8
of the hydroxy azide, very similar ratios were obtained in the
reactions of a given hydroxyalkyl azide and either 4-methyl or
4-tert-butylcyclohexanone. The best ratios were obtained from
reactions containing alkyl substitution at either the 1 or 3
position of the hydroxyalkyl side chain, with somewhat poorer
results being obtained for 2-alkyl or (especially) 2-aryl examples.
Overall yields for the entire sequence including iminium
formation and hydrolysis were excellent throughout. It is
noteworthy that outstanding levels of up to 1,7-stereoselection
were observed in this series.
A particularly instructive pair of examples is shown in
Scheme 8. The reactions of the disubstituted 3-carbon chain
hydroxyalkyl azide 11 with 4-substituted ketones proceeded with
high diastereoselectivity (g49:1), whereas those using the
epimeric azide 12 led to poor asymmetric induction for the same
ketones (ca. 1.5:1). These results, among others, provide
evidence for a chairlike transition structure for the heterocyclic
ring generated in the azide addition step of the reaction (see
below).
Additional opportunities for this reaction were briefly exam-
ined as shown in Scheme 9. As expected, the use of a meso-
disubstituted cyclohexanone provided product in excellent yield
and stereoselectivity, consistent with a hypothesis that selectivity
partially depends on a reasonably conformationally fixed ketone
(see below). Very good selectivity was also observed in the
reaction of an unsymmetrical ketone, (R)-3-methylcyclohex-
anone, but in this case the reaction leads to constitutional isomers
rather than stereoisomers (Scheme 9b). The fact that a stereo-
center is already present in the starting material means that
enantiomerically pure ketone must be used to avoid a 1:1
mixture of products (in this case, the enantiomeric (S)-3-
methylcyclohexanone would react with 5 to give mostly the
methyl epimer of compound 36b).17
The reactions of additional ketones have also been found to
proceed with good (unoptimized) yields but poor stereoselec-
tivity (Scheme 9c). Thus, one cyclobutanone and a meso bicyclic
(17) Aube´, J.; Hammond, M. Tetrahedron Lett. 1990, 31, 2963-2966.
9
7918 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Asymmetric Schmidt Reaction
A R T I C L E S
Scheme 9
Scheme 10
preparing it through debromination of pure 39. Although these
results could be complicated by a variety of factors (such as
instability of the alternative axial addition product or stereo-
chemical direction by the bromine group), they are at least
consistent with equatorial addition of azide in the 1,2-azido
alcohol series.
Discussion
The advantages of the in situ tethering method for the
stereoselective ring expansion of ketones to lactams are sever-
alfold. (1) It is a one-pot, one-workup procedure that provides
diastereomers that can generally be separated on a preparative
scale by nonheroic means. The removal of the chiral group can
be accomplished via several methods depending on structure.
By comparison, the oxaziridine method8 (Scheme 1) requires
three chemical steps to get to the diastereomeric lactam products.
(2) In the best cases, the selectivity is considerably greater than
previous methods; this is especially true for propane-derived
hydroxyalkyl azides (which also react more readily and at a
lower temperature). (3) Although limited, for the time being,
to substituted cyclohexanones, the reaction reliably provides
extremely easy access to a variety of chiral caprolactams with
up to 98% 1,7-selectivity being obtained.
cyclopentanone were examined but gave unpromising results.
An analogous reaction of a bicyclic cyclopentanone with 5 has
been reported by Vidari et al. with similar results.18 We propose
that such substrates have a poor level of diastereofacial
selectivity with respect to the plane of the ketone and that this
leads to the observed results.
Elimination Products from 2-Bromo-4-tert-Butylcyclohex-
anone. In previous work, we had learned that 2-phenyl-2-
azidoethanol reacted with 2-bromocyclohexanone to afford
modest yields of spirocyclic oxazoline products in place of the
usual ring-expanded lactams.10d As both reactions presumably
go through the same N-diazonium intermediate, it occurred to
us that this process could provide evidence for equatorial or
axial addition of azide leading to formation of the spirocyclic
stereogenic center (Scheme 10).
Thus, 2-bromo-4-tert-butylcyclohexanone was prepared as a
mixture of isomers and subjected to reaction with racemic 1 to
give a modest yield of four isomeric spirocyclic oxazolines, of
which the major isomeric compound 39 could be isolated by
crystallization. X-ray crystallography showed 39 to have the
structure shown in Scheme 10, in which the azide has attacked
the conformationally biased substrate from an equatorial direc-
tion. Alternatively, a thermodynamic equilibrium of isomeric
N-diazoniumoxazolidines could selectively undergo reaction to
predominantly afford elimination product containing equatorial
nitrogen. It was difficult to accurately determine the ratio of
isomers present at this stage, but removal of the bromide group
under radical conditions smoothly led to a ca. 10:1 mixture of
40a,b. The structure of 40a was confirmed by separately
Mechanistic Hypotheses: Three-Carbon Tethers. The
overall stereochemical course of the reaction has been deter-
mined for most of the examples examined, but of course only
two products are possible from the reaction of a chiral
hydroxyalkyl azide with a symmetrical ketone. We believe that
the stereochemistry depends on three independent variables,
illustrated for the specific example of 4-tert-butylcyclohexanone
with hydroxyalkyl azide 5: (1) addition of the azide onto the
oxenium ion from an equatorial direction or an axial direction
relative to resident ketone substitution (Scheme 11a), (2) the
reactive orientation of the newly formed heterocyclic ring
+
(Scheme 11b), and (3) the relationship between the leaving N2
group and the migrating carbon (Scheme 11c). These variables
will be discussed in reverse order in the context of the substituted
azidopropanol reagents, with the two-carbon case to be discussed
in the following section. One interesting point is that each
potentially migratory carbon is identically substituted, so that
electronic influences on migratory aptitude are absent, allowing
a purely stereoelectronic analysis of the rearrangement reac-
tion.19 Another interesting point is that the reactive oxenium
(18) Vidari, G.; Tripolini, M.; Novella, P.; Allegraucci, P.; Garlaschelli, L.
(19) The proposed transition state in our original communication10a was
incorrectly depicted and should be replaced by the following arguments.
Tetrahedron: Asymm. 1997, 8, 2893-2903.
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7919

A R T I C L E S
Sahasrabudhe et al.
Scheme 11
It seems very reasonable to propose synchronous migration
of carbon antiperiplanar to the leaving N2 moiety based on
+
the above precedent. In addition, the observation of high
selectivity in a number of examples examined suggests that prior
ionization of the aminodiazonium group followed by carbon
migration to a resulting nitrenium ion is less likely as there is
no clear way of rationalizing selective migration in such a
process. In the present case, antiperiplanar C f N migration
requires an axial orientation of the departing diazonium ion
+
(Scheme 11c). An equatorial N2 group would only have
synplanar migrating groups available to it, both of which would
have nearly the same dihedral angle between either migrating
group. Again, this situation would make it difficult to rationalize
the high stereoselectivity observed in this reaction. Thus,
although we do not have direct evidence of antiperiplanar and
synchronous migration in this step, such a postulate is strongly
in accord with previous work, theoretical considerations, and
our observations of highly stereoselective migration in the
absence of electronically differentiable migrating groups.
The analysis of the other two variables, shown in Scheme
11 (parts a and b), is complicated by the lack of evidence
regarding the reversibility of the azide addition to the oxenium
ion. It is likely, though, that similar steric interactions predomi-
nate in either eventuality and this principle will form the basis
for the following discussion.
A structure/selectivity analysis of the variously substituted
hydroxyalkyl azides is strongly supportive of the intermediacy
of a chairlike N-diazoniumtetrahydrooxazine ring. As shown
in Scheme 11b, the intermediacy of the two different chair forms
would lead to different isomers, provided that no other
mechanistic step were changed at the same time. Thus, better
selectivity should correlate with the stability of a given chair
form over its alternative. For monosubstituted hydroxyalkyl
azides, one presumes that the single substituent will occupy the
equatorial position (Scheme 11b). A particularly relevant
comparison is that between the reactions of syn- vs anti- 1,3-
dimethylazidopropanol, in which outstanding selectivity was
observed in the former but essentially disappeared in the latter
(Scheme 8). In the anti case, at least one of the methyl groups
must occupy an axial position and each chairlike form would
be expected to have nearly equal energy.
The series of 2-substituted azidopropanols is also instructive.
As expected, the size of the substituent has a profound effect
on the selectivity of the reaction (Table 3, cf. entries 7/9/11 or
8/10/12). The low selectivity observed with azide 9 relative to
either 8 or 10 was somewhat surprising in that the phenyl group
is generally thought to be nearly isosteric with an isopropyl
group. This is a matter for further exploration, but for the
moment it is worth mentioning that phenyl groups are sometimes
more stable in an axial orientation relative to their aliphatic
counterparts.26 Also, the lower selectivity for each of the
2-substituted azidopropanols relative to the 1- or 3-monosub-
stituted analogues makes sense because an axial group in this
position is in a 1,3-relationship to an oxygen and an aminodia-
zonium group, which should provide relatively small 1,3-diaxial
effects compared to the other two monosubstitution types (see
Scheme 11b for an example).
ion could, in principle, exist as two geometrical isomers with
the alkyl group coplanar to one or another of the adjacent
methylene groups. It is likely that these two forms are
interconvertible through rapid bond rotation and this point is
not considered in the analysis below.
Theoretical studies have characterized the aminodiazonium
group involved in this process as pyramidal and having a very
small barrier (ca. 1 kcal/mol) to stereochemical inversion.20
Numerous 1,2-migration reactions are believed to involve
antiperiplanar rearrangements of leaving group relative to the
migrating bond,21 including the Beckmann22 and classical
Schmidt23 rearrangements. Additionally, Pearson has invoked
antiperiplanar migration in studies in the closely analogous
reactions of alkyl azides with unstabilized carbocations12b,h and
Hoffman has also proposed this relationship in rearrangements
of N-(arylsulfonoxy)amines.24 Antiperiplanar migration has also
recently been supported experimentally for the Criegee and
Baeyer-Villiger reactions by Kishi, by Chandrasekar, and by
Crudden.25
(20) (a) Glaser, R.; Choy, G. S.-C. J. Phys. Chem. 1991, 95, 7682-7693. (b)
Glaser, R.; Choy, G. S.-C. J. Org. Chem. 1992, 57, 4976-4986. (c) Cacase,
F.; Attina`, M.; Speranza, M.; de Petris, G.; Grandinetti, F. J. Org. Chem.
1993, 58, 3639-3642. (d) Glaser, R.; Choy, G. S.-C. J. Am. Chem. Soc.
1993, 115, 2340-2347.
(21) Deslongchamps, P. Stereoelectronic Effects in Organic Chemistry; Perga-
mon: Oxford, 1983.
(22) Gawley, R. E. Org. React. 1988, 35, 1-420.
(23) (a) Wolff, H. Org. React. 1946, 3, 307-336. (b) Banthorpe, D. V. In The
Chemistry of the Azido Group; Patai, S., Ed.; John Wiley & Sons: London,
1971; pp 397-440. (c) Smith, P. A. S. In Molecular Rearrangements; de
Mayo, P., Ed.; John Wiley & Sons: New York, 1963; Vol. 1, pp 457-
591.
(24) Hoffman, R. V.; Kumar, A. J. Org. Chem. 1985, 50, 1859-1863.
(25) (a) Chandrasekhar, S.; Roy, C. D. Tetrahedron Lett. 1987, 28, 6371-6372.
(b) Chandrasekhar, S.; Roy, C. D. J. Chem. Soc., Perkin Trans. 2 1994,
2141-2143. (c) Goodman, R. M.; Kishi, Y. J. Am. Chem. Soc. 1998, 120,
9392-9393. (d) Crudden, C. M.; Chen, A. C.; Calhoun, L. A. Angew.
Chem., Int. Ed. Engl. 2000, 39, 2851-2855.
Finally, two ways of gaining entry into the two conformers
are possible. In one, each chair form could kinetically form in
(26) Eliel, E. L.; Manoharan, M. J. Org. Chem. 1981, 46, 1959-1962.
9
7920 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003

Asymmetric Schmidt Reaction
A R T I C L E S
Scheme 12
the course of azide attack onto the oxenium ion. In this case,
selectivity would result from the minimization of nonbonded
interactions in the course of this step. Alternative, the two
conformers could equilibrate following ring closure; if this
equilibrium is rapid, selectivity would arise by minimizing steric
interactions in the migration step itself. (This kind of equilibrium
could involve one of two distinct mechanisms: simple confor-
mational equilibrium or, alternatively, retroaddition of the azide
species to regenerate the oxenium ion followed by readdition
of azide.) We have no experimental evidence to favor either
view at this point, but note once again that either mechanism
would be subject to a similar analysis of conformer stability.
Related arguments can be used to address the third issue,
i.e., the relative stereochemistry of azide addition relative to
the preexisting 4-substituent on the original cyclohexanone ring
(Scheme 11a). There is no direct precedence to indicate whether
a kinetic azide attack should occur from an equatorial or an
axial orientation onto a relatively fixed cyclohexanone-based
oxenium ion. In either case, spirocyclization would occur with
one group in the usually disfavored axial position; a simplistic
analysis would indicate that the smaller oxygen group might
prefer to emerge in this position relative to the substituted
nitrogen atom. Again, the azide addition might be reversible,
in which case the relative stability of the two spirocyclic
products would determine the overall stereoselectivity of the
process (thermodynamic control). (Here, equilibrium between
the two forms could only occur through a retroaddition/
readdition mechanism and not simple conformational equilib-
rium).
Scheme 13
Initially, we settled on a favored equatorial addition (either
kinetic or thermodynamic) by (1) assuming antiperiplanar
migration, (2) assuming formation of the more stable chairlike
tetrahydrooxazine ring, and (3) correlating the absolute con-
figurations with the starting materials and products. Gathering
confidence in assumptions (1) and (2) for the reasons noted
above, we have additional circumstantial evidence for equatorial
addition in spirocyclizations of two-carbon-chain hydroxyalkyl
azides step through the experiments shown in Scheme 10. Again,
we take comfort from the relatively high stereoselectivity of
the overall process: if a change in equatorial/axial addition were
to occur, it would have to be dramatic to account for the
relatively high selectivities observed throughout.
Taken together, these considerations lead to the overall
pathway summarized in Scheme 12.19 Addition of the azide onto
the oxenium ion sets up an equilibrium between two chairlike
heterocycles (as discussed above but not shown in the scheme,
this equilibrium may be set via various mechanisms). The above-
noted arguments suggest that the predominant isomer of these
processes is that shown to the left in Scheme 12, resulting from
equatorial addition and formation of a tetrahydrooxazine ring
in which the phenyl substituent occupies the more stable
equatorial position. Antiperiplanar migration of carbon to
nitrogen proceeds predominantly from this intermediate to yield
the observed major product. The minor product is depicted as
resulting from the less-favored heterocyclic chair. This is
proposed based on the observations that there is little, if any,
dependence of selectivity on the substitution of the starting
ketone (see Table 3, and cf. entries 1/2/3, 4/5, 7/8, etc.) but
changes in substitution on the hydroxyalkyl azide have a
significantly larger effect.
Mechanistic Hypotheses: Two-Carbon Tethers. Although
somewhat less selective than their three-carbon-chain counter-
parts, the azidoethanol derivatives still provide some caprolac-
tams in ratios rivaling the previous-best oxaziridine technology.8
In this case, some circumstantial evidence supports equatorial
addition of azide (Scheme 10), whereas the relationship of the
N-diazonium group and the existing tetrahydrooxazine’s ste-
reocenter is less certain due to their presence on a five-
membered ring. Accordingly, the mechanism shown in Scheme
13 is proposed, wherein the key source of stereoselectivity is
the minimization of steric interactions between the phenyl group
present on the hydroxyalkyl azide and the migrating carbon.
This hypothesis, which results in a reactive conformation in
which the diazonium ion is on the same face as the phenyl
group, arises because of the need to correlate the structures of
starting materials and products while still assuming antiperipla-
nar migration. The cis N2+/Ph relationship is reasonable because
9
J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003 7921

A R T I C L E S
Sahasrabudhe et al.
over 5 min. The solution was stirred for 30 min, and partitioned between
CH2Cl2 (20 mL) and H2O (10 mL). The layers were separated and the
aqueous layer was extracted with CH2Cl2 (2 × 10 mL). The combined
organic layers were washed with NH4Cl (5 mL), dried (MgSO4), filtered
and concentrated. HPLC analysis of the crude reaction mixture showed
a 93:7 ratio of diastereomeric lactams. Flash chromatography (1:1f9:1
EtOAc/hexane) gave 19a and 19b as transparent oils in a combined
yield of 296 mg (98%). TLC (1:1 hexane/EtOAc): Rf (19a) ) 0.15, Rf
(19b) ) 0.20; HPLC: tR major (19a) ) 19.9 min, tR minor (19b) )
18.2 min (Chirobotic T; 90% hexane/EtOH; flow rate 1 mL/min; UV
254 nm). Major diastereomer (19a): [R]D ) -4.2 (c 1.02, CHCl3). 1H
NMR (400 MHz, CDCl3) δ 0.99 (d, J ) 6.6 Hz, 3H), 1.12-1.31 (m,
3H), 1.62-1.75 (m, 1H), 1.76-1.98 (m, 4H), 2.42-2.61 (m, 2H), 3.12
(dt, J ) 14.2, 4.4 Hz, 1H), 3.24 (dd, J ) 6.5, 15.2 Hz, 1H), 3.51 (dd,
J ) 10.9, 15.2 Hz, 1H), 4.09-4.19 (m, 1H), 4.62 (m, 1H), 7.23-7.38
(m, 5H); 13C NMR (100.6 MHz, CDCl3) δ 22.6, 31.4, 35.7, 36.1, 37.2,
37.7, 45.7, 49.3, 69.8, 125.5, 127.0, 128.3, 144.1, 177.1. MS (EI) m/e
262 (M++1), 244 (100); HRMS m/e calcd for C16H24NO2: (M++1)
262.1807, found: (M++1) 262.1798. Minor diastereomer (19b): 1H
NMR (400 MHz, CDCl3) δ 0.97 (d, J ) 6.5 Hz, 3H), 1.18-1.30 (m,
2H), 1.65-1.95 (m, 5H), 2.50-2.56 (m, 2H), 3.01 (m, 1H), 3.32 (dd,
J ) 15.1, 5.9 Hz, 1H), 3.47 (J ) 15.1, 11.0 Hz, 1H), 4.10-4.25 (m,
1H), 4.53 (br d, J ) 10.4 Hz, 1H), 4.75-4.85 (m, 1H), 7.21-7.39 (m,
5H); 13C NMR (100.6 MHz, CDCl3) δ 22.6, 31.3, 35.7, 35.8, 36.3,
37.5, 44.9, 48.4, 69.6, 125.7, 127.0, 128.2, 144.1, 177.2. MS (EI) m/e
262 (M+ + 1), 244; HRMS m/e calcd for C16H24NO2: (M++1)
262.1807, found: (M++1) 262.1797.
of the small size of the azido group compared to the steric
repulsion between the phenyl group and the migrating carbon,
which increases as the reaction proceeds. Finally, we note that
this hypothesis is consistent with the steep drop in selectivity
incurred when moving the phenyl group from its position next
to the azido group (C-2) to C-1 (Table 2, entry 5).
Summary
An asymmetric equivalent of the Schmidt reaction is pre-
sented, in which in situ tethering permits stereocontrol in ring
expansions of symmetrical cyclohexanones. Selectivities as high
as ca. 98:2 are possible for the synthesis of substituted
caprolactams, with up to 1,7-stereoselection involved in the
overall process. The synthetic utility of this reaction derives
from the lack of comparably useful methods to accomplish this
sort of asymmetric nitrogen ring expansion. Both 1,2- and 1,3-
azido alcohols are useful and bring complementary advantages
and liabilities to the experimentalist. The 1,3-azido alcohols are
generally more reactive and provide superior selectivities.
However, the lactams arising from 1-azidophenylethanol are
easier to N-dealkylate.
The fact that either possible migrating carbon is electronically
identical provided an unusual opportunity to study a ring-
expansion controlled entirely by stereoelectronic factors. Al-
though there is much to learn about the details of this process,
a stereochemical analysis has allowed us to make some initial
suggestions as to how the selectivity obtained in the overall
process might be achieved.
Acknowledgment. Financial support was provided by the
National Institutes of Health (GM-49093). The authors also
thank Lawrence Seib and Doug Powell for crystallographic data,
and Robie Lucas for carrying out some early experiments. We
thank the referees of the paper for helpful comments.
Experimental Section
Representative Procedure: Synthesis of 19a,b. A solution of (R)-
3-azido-1-phenylpropanol 5 (205 mg, 1.15 mmol) and 4-methylcyclo-
hexanone (188 mg, 1.68 mmol) in 3 mL of CH2Cl2 was cooled to -82
°C (ether/dry ice bath), and BF3‚OEt2 (0.56 mL, 4.48 mmol) was added
dropwise. The reaction was gradually warmed to room temperature over
a period of 48 h. The resulting crude iminium ether was diluted with
Et2O (5 mL) and hydrolyzed with 50% KOH (1 mL) added dropwise
Supporting Information Available: Characterization data for
new compounds and additional experimental details. This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA0348896
9
7922 J. AM. CHEM. SOC. VOL. 125, NO. 26, 2003