Reactions of alkyl azides and ketones as mediated by Lewis acids: Schmidt and Mannich reactions using azide precursors

Article abstract of DOI:10.1021/ja000490v

The Lewis acid-promoted reactions of alkyl azides with ketones can afford several products. Chief among these result from a Schmidt-like insertion of the azide into the carbon - carbon bond adjacent to the carbonyl group. Alternatively, an acid-promoted r

Full text of DOI:10.1021/ja000490v

7226
J. Am. Chem. Soc. 2000, 122, 7226-7232
Reactions of Alkyl Azides and Ketones as Mediated by Lewis Acids:
Schmidt and Mannich Reactions Using Azide Precursors
Pankaj Desai, Klaas Schildknegt, Konstantinos A. Agrios, Craig Mossman,
Gregory L. Milligan, and Jeffrey Aube´*
Contribution from the Department of Medicinal Chemistry, UniVersity of Kansas,
Lawrence, Kansas 66045-2506
ReceiVed February 8, 2000
Abstract: The Lewis acid-promoted reactions of alkyl azides with ketones can afford several products. Chief
among these result from a Schmidt-like insertion of the azide into the carbon-carbon bond adjacent to the
carbonyl group. Alternatively, an acid-promoted rearrangement of the azide to an iminium species can occur,
mostly with benzylic azides; the iminium species can then be trapped by the enol of the carbonyl compound
in a variation of the Mannich reaction. The scope of each of these reactions, the dependence of the observed
products upon azide and ketone structure, and the nature of acid promotion are discussed. In broad strokes,
cyclohexanones and other cyclic ketones react in the presence of TiCl4 to afford insertion products, whereas
the Mannich route predominates when benzyl azide and triflic acid are used. The features that lead to each
reaction type and possible mechanistic implications are discussed.
The insertion of an alkyl azide into a carbonyl compound to
Scheme 1
afford an amide is analogous to the Schmidt reaction of
hydrazoic acid and ketones (Scheme 1).1 Mostly based on
experiments carried out by Briggs2 and Smith3 in the 1940s,
conventional wisdom had long dictated that a Schmidt reaction
using alkyl azides was not possible. However, work in this
decade clearly demonstrated intramolecular reactions of alkyl
azides with ketones,4 oxonium ions5,6 and carbocations7 to be
efficient and synthetically useful processes.
The Mannich reaction and variations thereof have also loomed
large in the synthesis of nitrogenous compounds.8 The traditional
* Corresponding author. Telephone: (785) 864-4496. Fax: (785) 864-
5326. E-mail: jaube@ukans.edu.
(1) For relevant reviews, see: (a) Wolff, H. Org. React. 1946, 3, 307-
336. (b) Smith, P. A. S. In Molecular Rearrangements; de Mayo, P., Ed.;
John Wiley & Sons: New York, 1963; Vol. 1, pp 457-591. (c) Abramovich,
R. A.; Kyba, E. P. In The Chemistry of the Azido Group; Patai, S., Ed.;
John Wiley & Sons: London, 1971; pp 221-329. (d) Kyba, E. P. In Azides
and Nitrenes: ReactiVity and Utility; Scriven, E. F. V., Ed.; Academic:
Orlando, 1984; pp 2-34. (e) Krow, G. R. Tetrahedron 1981, 37, 1283-
1307.
(2) Briggs, L. H.; De Ath, G. C.; Ellis, S. R. J. Chem. Soc. 1942, 61-
63.
(3) Smith, P. A. S. J. Am. Chem. Soc. 1948, 70, 320-323.
(4) (a) Aube´, J.; Milligan, G. L. J. Am. Chem. Soc. 1991, 113, 8965-
8966. (b) Le Dre´au, M.-A.; Desmae¨le, D.; Dumas, F.; d’Angelo, J. J. Org.
Chem. 1993, 58, 2933-2935. (c) Norris, P.; Horton, D.; Levine, B. R.
Tetrahedron Lett. 1995, 36, 7811-7814. (d) Milligan, G. L.; Mossman, C.
J.; Aube´, J. J. Am. Chem. Soc. 1995, 117, 10449-10459. (e) Wendt, J. A.;
Aube´, J. Tetrahedron Lett. 1996, 37, 1531-1534.
(5) (a) Boyer, J. H.; Hamer, J. J. Am. Chem. Soc. 1955, 77, 951-954.
(b) Mossman, C. J.; Aube´, J. Tetrahedron 1995, 52, 3403-3408. (c)
Badiang, J. G.; Aube´, J. J. Org. Chem. 1996, 61, 2484-2487.
(6) (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, 16241-16252.
(7) (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, 10183-10194.
(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, 12039-12048.
(8) Tramontoni, M.; Angiolini, L. Tetrahedron 1990, 46, 1791-1837.
formulation of this process as applied to aniline is shown in
Scheme 1. It involves the condensation of an amine with an
aldehyde to afford a reactive iminium salt (formaldehyde, shown
in the Scheme, is a common partner). Condensation with an
enolizable ketone then affords the product, which is often called
a Mannich base.
10.1021/ja000490v CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/15/2000

Reactions of Alkyl Azides with Ketones
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7227
Table 1. Effect of Lewis Acids on the Reaction of Benzyl Azide
with Cyclohexanone
Our continuing studies of the Schmidt reaction of alkyl azides
led us to consider whether simple azides (i.e., those lacking
additional functionality) could react with appropriately activated
ketones after all. In particular, the use of Lewis acid activation
instead of the protic conditions almost exclusively used in the
era of Briggs’s and Smith’s work was thought worthy of
examination. The initial results of this study, which showed that
TiCl4-promoted Schmidt reactions did occur with a limited range
of cyclic ketones, have been published in preliminary form.9
When the same reaction conditions were attempted with acyclic
ketones, however, no insertion chemistry was observed. Instead,
a rearrangement/addition product reminiscent of the Mannich
reaction was the only tractable substance obtained (the “azido-
Mannich” process shown at the bottom of Scheme 1). The sub-
stitution of triflic acid for TiCl4 resulted in a significant
improvement in the efficiency, cleanliness, and scope of the
reaction.10
product ratiosa
entry
acid
TFA
TfOH
TiCl4
TiCl4
SnCl4
equiv
1
2
recovered azide
100
1
2
3
4
5
6
7
8
9
excess
1.1
79b
39
15
11
?
1.1
2.5
1.1
2.5
1.1
2.5
1.1
2.5
45
85
16
89
90
89
92
100
100
SnCl4
BF3‚OEt2
BF3‚OEt2
AlCl3
11
8
In addition to providing the details of new variations of two
venerable synthetic methods, this full account will examine the
factors that determine whether the Schmidt or Mannich manifold
will predominate. The emergence of several interesting side
reactions and the relationship of these methods to other
techniques will also be discussed.
10
AlCl3
a
Ratios are normalized to 100 and calculated on the basis of proton
NMR integrations of crude reaction mixtures after workup (except
b
where noted). Isolated yield.
activation led to variable results, but far and away the most
effective promoter of the insertion reaction was TiCl4 (generally
used in methylene chloride solution). Although decent conver-
sions to lactam could be obtained using approximately equimolar
ratios of reactants and promoter, best results were obtained when
the azide was used in excess (cf. entries 3 and 4). Under no
circumstances was less than a full equivalent of Lewis acid
sufficient, presumably because the product amide is a better
Lewis base than the reactant ketone and therefore holds onto
the promoter until workup.
Results and Discussion
Survey of Reaction Conditions. The success of the intramo-
lecular Schmidt reaction had clearly established alkyl azides as
useful nucleophilic partners with ketones. In addition, this work
unambiguously showed for the first time that the azidohydrin
intermediate initially formed in the Schmidt reaction of either
hydrazoic acid or an alkyl azide could undergo nitrogen
elimination and directly afford lactam (Scheme 2). In contrast,
When benzyl azide, cyclohexanone, and triflic acid were
combined, no Schmidt reaction product was obtained. Instead,
a structure strongly suggestive of a Mannich process was
obtained in this experiment (entry 2). However, ∼30% yields
of insertion product had been elsewhere noted when triflic acid
was used with nonbenzylic substrates.12 The (phenylamino)-
ketone was also formed in significant amounts when a single
equivalent of TiCl4 was added to the two reactants (entry 3).
This reaction is thought to proceed as indicated in Scheme 3.
Scheme 2
Scheme 3
the Schmidt reaction of hydrazoic acid with ketones probably
undergoes initial dehydration followed by rearrangement of an
iminodiazonium species, reminiscent of the Beckmann rear-
rangement.11 The failure of simple alkyl azides to react with
ketones under classical protic conditions, therefore, seemed most
likely to result from the unfavorable formation of the initial
azidohydrin adduct (i.e., Keq in Scheme 2 is too small). In the
intramolecular case, this equilibrium might be more favorable
due to entropy. It was reasoned that more aggressive Lewis acid
activation of the carbonyl should also facilitate azidohydrin
formation in intermolecular cases, and thus a range of acids
was surveyed with benzyl azide and cyclohexanone as repre-
sentative reactants (Table 1).
In accord with our previously published account,9 trifluoro-
acetic acid did not effectively promote the reaction. Lewis acid
Initial activation of the azide functional group by protonation
(or by Lewis acid coordination, not shown) leads to a migration
of the phenyl substituent with loss of nitrogen gas, which serves
as a driving force. This leads to the Mannich intermediate shown
(9) Aube´, J.; Milligan, G. L.; Mossman, C. J. J. Org. Chem. 1992, 57,
1635-1637.
(10) Schildknegt, K.; Agrios, K. A.; Aube´, J. Tetrahedron Lett. 1998,
39, 7687-7690.
(11) Richard, J. P.; Amyes, T. L.; Lee, Y.-G.; Jagannadham, V. J. Am.
Chem. Soc. 1994, 116, 10833-10834.
(12) Gracias, V.; Frank, K. E.; Milligan, G. L.; Aube´, J. Tetrahedron
1997, 53, 16241-16252.

7228 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Desai et al.
Table 2. Effect of Lewis Acids on the Reaction of Benzyl Azide
Table 3. TiCl4-Promoted Reactions of Substituted Cyclohexanones
with 3-Pentanone
product ratiosa
entry
acid
equiv
3
4
recovered azide
1
2
3
4
5
6
7
TfOH
TiCl4
TiCl4
SnCl4
SnCl4
BF3‚OEt2
BF3‚OEt2
1.1
1.1
2.5
1.1
2.5
1.1
2.5
84b
62
62
38
38
100
100
70
30
10
90
a
Ratios are normalized to 100 and calculated on the basis of proton
NMR integrations of crude reaction mixtures after workup (except
b
where noted). Isolated yield.
which is then captured by the enol of the ketone component of
the reaction.13
A few experiments explored the effects of changing the order
of reagent addition. Little difference was observed when TiCl4
was added to either a mixture of ketone and azide or when added
to benzyl azide followed immediately by addition of ketone (4-
5:1 ratio of Schmidt to Mannich product being observed in each
case). However, when benzyl azide was allowed to react with
TiCl4 overnight (to room temperature), a powdery precipitate
formed. Removal of solvent by cannula was followed by adding
fresh methylene chloride; when the resulting suspension was
cooled to 0 °C and cyclohexanone added, only Mannich product
was observed by NMR. Standard workup resulted in the
isolation of 18% of 2-(phenylamino)methylcyclohexanone.
These results are consistent with the powderswhich was often
observed in the reactions of benzyl azide but absent in those of
hexyl azide (see below)sbeing the proposed iminium ion
intermediate.
An analogous study carried out with 3-pentanone as the
carbonyl component revealed contrasting results in the sense
that no insertion product whatsoever was observed in these cases
(Table 2). Instead, both triflic acid and TiCl4 resulted in the
formation of Mannich-type product, although the former was
the superior reagent for preparative use. As in the case of
cyclohexanone, other Lewis acids were relatively inert. We have
yet to observe an acyclic ketone that undergoes a Schmidt-type
insertion reaction with an alkyl azide (besides 2-pentanone, we
have tried acetone, 2-butanone, 2-hexanone, and several â-keto
esters and â-diketones).
With boundary conditions thus laid out, we set about the task
of determining the scope of these two divergent processes. The
goals were (1) to establish the level of synthetic utility that could
be expected from each reaction, (2) to further delineate when
each reaction pathway was likely to predominate, and (3) to
learn what we could about the mechanisms underlying this
chemistry.
TiCl4-Promoted Reactions of Substituted Cyclohexanones.
Having established that only cyclic ketones apparently undergo
Schmidt reaction and that TiCl4/CH2Cl2 offers the best reaction
conditions for effecting the process, a variety of cyclohexanones
were reacted with benzyl azide and n-hexyl azide, the latter
serving as a prototypical aliphatic alkyl azide (Table 3).
a
Isolated as an inseparable mixture of regiosiomers.
The results of these experiments led to several useful
generalizations. Unhindered cyclohexanones and their deriva-
tives generally worked well in these reactions. However,
dramatically lowered yields and sluggish reactions resulted when
even modest R-substitution was added to the carbonyl substrate.
(Of less practical importance, due to the poor chemical yields,
is the observation that the reactions are also poorly regioselec-
tive.)
The effect of more distant substitution can be seen in a series
of progressively methylated cyclohexanones (entries 9-12).
Although it was possible to optimize the yields of Schmidt
reactions of 3,5-dimethylcyclohexanone to the 50-60% range,
we never saw any analogous product arise from 3,3,5-trimeth-
ylcyclohexanone. It is evident that 1,3-axial methylation suffices
to lower the yields of insertion product dramatically. These
results are probably due to steric crowding in the transition
structures as it is unlikely that such conservative substitution
can sufficiently perturb the electronic nature of the system to
have this effect. It is instructive that, in some cases, products
arising from Mannich processes resulted even when benzyl azide
and TiCl4 were used (entries 9 and 10).
Reactions of Other Cyclic Ketones with Alkyl Azides and
TiCl4. A selection of other representative ketones yielded further
information (Table 4). Again, the generalization that sterically
unhindered ketones underwent the most efficient insertion
reactions seemed to hold. In these experiments, we did not
assiduously search out Mannich products for all cases. Rigidified
or “tied-back” examples such as adamantanone or the steroid
derivative shown gave the best results. However, simple
cyclopentanones gave very poor yields (although the five-
membered ring ketone embedded in norbornanone did prove a
(13) Pearson, W. H.; Fang, W.-k. Isr. J. Chem. 1997, 37, 39-46.

Reactions of Alkyl Azides with Ketones
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7229
Table 4. TiCl4-Promoted Reactions of Ketones with Alkyl Azides
Table 5. Reaction of 3-Pentanone with Alkyl Azides Using Triflic
Acid
product
entry
R
R1 (yield, %)
notesa
1
2
3
4
phenyl
phenyl
phenyl
H
H
H
H
3 (84)
3 (14) ether
3 (59) hexane
p-methoxy phenyl
34 (22) compound 35b
also isolated (19%)
5
6
7
8
9
m-methoxy phenyl
m-methoxy phenyl
m-methoxy phenyl
m-methoxy phenyl
phenyl
H
H
H
H
36 (44)
36 (49) 2 equiv of azide used
36 (49) 2 equiv of TfOH used
36 (49) 2 equiv of ketone used
Me 37 (29)
10 p-carbomethoxyphenyl H
11 p-bromophenyl
12 n-hexyl
38 (69)
39 (77)
H
H
no product isolated
a Standard conditions (except where noted): equivalent amounts of
all reagents, with reactions carried out in CH2Cl2. b See Scheme 4.
Scheme 4
fair substrate). On the other hand, the reaction of hexyl azide
with 3-phenylcyclobutanone gave a somewhat better result. The
observation that cyclopentanones are not good Schmidt reaction
partners for alkyl azides is one of the most striking and
unexpected outcomes of this study. Besides cyclopentanone
itself (entry 2), we have examined 1- and 2-tetralones and several
additional substituted cyclopentanones without success. For
larger (seven- and eight-membered) ketones, only Mannich
pathway was observed.
Scope of the Azido-Mannich Process. Additional studies
pertaining to the rearrangement of alkyl azides prior to reaction
with ketone are presented in Table 5. A series of aromatic azides
gave various levels of the Mannich bases along with, in one
instance, an interesting side product 35 (entry 4). The particu-
larly low yields associated with p-methoxybenzyl azide and the
byproduct obtained through this process could both be explained
through the intermediacy of a stabilized benzylic cation resulting
from loss of hydrazoic acid from the protonated substrate
(Scheme 4). Alternatively, the low yields of the experiment in
entry 4 might be due to competitive destruction of the reagent
by R-hydride migration, which in this case would afford a
benzylic cation stabilized by the p-methoxy group. In contrast,
no Mannich product was observed when 3-pentanone was
reacted with n-hexyl azide (entry 12).
Pragmatically, the most important data gleaned from this
survey were that (1) approximately equimolar quantities of all
reactants suffice for good results, (2) triflic acid in methylene
chloride is a superior medium relative to ether or hexane, (3)
when TfOH is used as the acid, the order of addition is relatively
unimportant, and (4) a good migrating group like phenyl is
apparently necessary for a successful reaction. Relevant and
complementary results along these lines have been obtained by
other workers. Kuwajima has reported an analogous amino-
methylation procedure beginning with azido(trimethylsilyl)-
methane, in which the silyl group undergoes migration to
generate an electrophilic Mannich species that reacts with
ketones.14 In addition, Pearson has reported that the iminium
species from benzyl azide as shown in Scheme 3 can react as
a partner in a Diels-Alder-style cycloadditions to afford 1,2,3,4-
tetrahydroquinolines and 1,2-dihydroquinolines.13 The effect of
aromatic substitution in the Pearson chemistry mirrors that noted
here.
(14) Tanino, K.; Takahashi, M.; Murayama, K.; Kuwajima, I. J. Org.
Chem. 1992, 57, 7009-7010.

7230 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Desai et al.
Table 6. Triflic Acid Promoted Reactions of Benzyl Azide with
Carbonyl Substrates
As might be expected, the ratios of the regioisomeric
R-phenylaminomethylated ketones roughly reflected the ex-
pected thermodynamic ratios of the intermediate enols. We
briefly examined some silyl enol ethers as starting points for
the reaction (Scheme 5). Although these compounds turned out
Scheme 5
to be acceptable substrates, the product ratios did not reflect
the ratios of the starting enol ethers. This suggests that these
reactions proceed through initial hydrolysis of the enol ethers
and that the resulting ketones react with the aminomethylation
reagent present in situ.
An entirely different type of product was obtained from the
reaction of 2-carboethoxycyclopentanone and benzyl azide in
the presence of TiCl4 (Table 6, entry 12 and Scheme 6). In this
Scheme 6
a Isolated yields of reactions promoted by the indicated acid.
The range of possible ketone partners was explored through
the series of experiments compiled in Table 6. In some
examples, both TfOH and TiCl4 were assayed. Good yields
could be obtained through most of the substrate types examined
as long as reasonable amounts of enol could be generated in
the reaction mixture. As was our experience with the intramo-
lecular Schmidt reaction of alkyl azides, R,â-unsaturated ketones
continue to pose a problem (entry 15).4 On the other hand,
â-keto esters generally gave good results, although we did not
obtain tractable products from acetoacetone or dimedone.
example, an 88% yield of the tricyclic compound 50 was
obtained and its structure confirmed through X-ray crystal-
lography. Although it is possible that the reaction occurs via a
Mannich reaction followed by an intramolecular Friedel-Crafts
process, this seems unlikely in view of the relatively poor yield
obtained when bona fide Mannich product 49 was submitted to
the reaction conditions that led to the tricyclic material. Another
possibility is that the Lewis acid-activated iminium ion under-
goes a hetero-Diels-Alder reaction with the enolized form of

Reactions of Alkyl Azides with Ketones
J. Am. Chem. Soc., Vol. 122, No. 30, 2000 7231
the â-keto ester. Related Diels-Alder processes were indepen-
dently observed by Pearson from rearrangement processes.13 At
any rate, this reaction remains a curiosity for the time being as
it has not been observed in any other case, including the
homologous 2-carbomethoxycyclohexanone.
Mechanistic Issues and Scope. It is apparent from these
studies that the Schmidt reaction of alkyl azides with ketones
is highly sensitive to steric effects. Furthermore, when an azide
has an alternative pathway available to it, other processes begin
to take over. It is also clear that powerful Lewis acid activation
is needed for efficient addition of azide to the carbonyl group.
There are a variety of mechanistic possibilities available to the
reaction. Most similar to the classical Schmidt reaction mech-
anism1 would be the simple Lewis acid activation of the
carbonyl followed by the addition of alkyl azide (Scheme 7,
position based on observed chemistry. Such examples would
include the azidohydrin intermediates proposed for both inter-
molecular and intramolecular Schmidt reactions, and protonation
prior to imine formation in the Mannich process (Scheme 3).
Olah and co-workers detected exclusively proximal protonation
of methyl and ethyl azides in superacid medium; in addition,
aminodiazonium ions are favored over the tautomeric imino-
diazenium ions.16 For these reasons and for the sake of simpli-
city, we prefer the former mechanism for rationalizing our
results.
Significant steric crowding results from the addition of both
Lewis acid and alkyl azide to the carbonyl component. Thus,
unhindered cyclohexanones or multicyclic ring systems in which
the ketone-bearing ring is immobilized without nearby substit-
uents work best. However, the ketone’s enthusiasm for ring-
expansion diminishes with even moderate steric bulk or, it would
seem, conformational flexibility. This would account for the
lack of observed Schmidt reactivity with alkyl azides and acyclic
ketones. As shown in Scheme 8, a terminal methyl group in
Scheme 7
Scheme 8
the conformationally flexible butanone derivative shown can
swing into the vicinity of the carbonyl, in a placement
reminiscent of the 2-methyl group of 2-methylcyclohexanone,
which is also a poor Schmidt substrate. In either case, the steric
environment of the ketone is significantly crowded and appar-
ently unable to accommodate both the Lewis acid and the poorly
nucleophilic azide as is required for addition to the carbonyl.
It seems unlikely that steric hindrance alone can account for
the dependence of Schmidt reaction yields on ring size. In
general terms, additions to cyclic ketones (such as cyanohydrin
formation) are favorable for three-, four-, and six-membered
rings but less facile for all other common ring systems.17 These
phenomena are usually attributed to the emergence of greater
1,2-torsional interactions encountered during additions to five-
and seven-to-ten-membered ring systems or, alternatively, due
to an increase in I-strain during the conversion of the carbonyl
carbon from the sp2 to the sp3 hybridization state.17 The present
work indicates that the addition of an alkyl azide to a carbonyl,
even when activated by a powerful Lewis acid, is a marginally
acceptable process that only occurs when everything is right.
These considerations can explain the failure of cyclopent-,
cyclohept- and cyclooctanones to afford insertion product under
the standard conditions described.
top). As noted above, similarities to the Schmidt reaction of
hydrazoic acid with ketones must stop with the formation of
the initial azidohydrin adduct as dehydration to afford an
iminodiazonium ion is not an option for this intermediate. Once
formed, it can only undergo bond reorganization to yield the
product lactam (as its TiCl4 complex) or revert back to starting
materials. (Note that the values of n for TiCln are left open
throughout this discussion.)
An alternative would be for the azide to form an adduct with
the Lewis acid (Scheme 7, bottom). The complexation of methyl
azide with SbCl5 has been reported; interestingly, the adduct
leads to an imine upon treatment with HCl.15 The titanium could
coordinate azide at either the proximal (alkyl group-bearing)
or distal nitrogen. In the present context, complexation of the
TiCl4 group with the ketone carbonyl might set up a six-
membered ring transition structure allowing for transfer of the
alkyl azide to the carbonyl group and establishing the azidohy-
drin intermediate. However, the addition of electrophiles to an
alkyl azide is generally considered to occur at the proximal
(16) Mertens, A.; Lammertsma, K.; Arvanaghi, M.; Olah, G. A. J. Am.
Chem. Soc. 1983, 105, 5657-5660.
(17) For a general discussion of these concepts and leading references,
see: Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of Organic
Compounds; John Wiley & Sons: New York, 1994; pp 762 and 769-771.
(15) Goubeau, J.; Allenstein, E.; Schmidt, A. Chem. Ber. 1964, 97, 884-
890.

7232 J. Am. Chem. Soc., Vol. 122, No. 30, 2000
Desai et al.
formed after 15 min, and the suspension was stirred for a total of 16
h at which time it was diluted with 20 mL of ethyl acetate and
partitioned between 200 mL of ethyl acetate and 30 mL of saturated
sodium bicarbonate. The organic layer was extracted with 30 mL of
brine and dried over anhydrous sodium sulfate. Evaporation of the
solvent followed by flash chromatography (ethyl acetate-hexanes (1:
4)) gave 197 mg of a clear oil (100%). 1H NMR (300 MHz, CDCl3) δ
0.90 (br t, J ) 4.78 Hz, 3H), 1.31 (br s, 6H), 1.41-1.55 (m, 2H), 1.72
(br s, 2H), 1.72-1.85 (m, 4H), 1.85-2.00 (m, 5H), 2.06 (br s, 2H),
2.80-2.87 (m, 1H), 3.37 (br t, J ) 6.4 Hz, 2H); 13C NMR (75.6 MHz,
CDCl3) δ 13.9, 22.4, 26.2, 26.5, 27.9, 31.0, 31.6, 34.4, 35.7, 42.4, 48.8,
52.7, 178.2; IR (neat) 2914, 1630 cm-1; MS (EI) m/e 249 (M+), 220,
206, 178. HRMS m/e calcd for C16H27NO: 249.2093, found 249.2092.
General Procedure for Mannich Reactions Using Triflic Acid.
2-Phenylaminomethyl Cyclohexanone (1). To a solution of benzyl
azide (500 mg, 3.76 mmol) in 8 mL of CH2Cl2 at 0 °C was added
cyclohexanone (0.39 mL, 3.76 mmol). The mixture was stirred for 10
min and TfOH (0.37 mL, 4.14 mmol) was added dropwise; gas
evolution was observed. The reaction was allowed to warm to room
temperature and stirred for 20 h at which time 0.2 g of solid NaHCO3,
10 mL of CH2Cl2, and 20 mL of H2O were added. The aqueous layer
was extracted with CH2Cl2 (4 × 10 mL), and the combined organic
layers were washed with brine and dried over anhydrous MgSO4.
Concentration followed by chromatography (15% ether/hexanes) gave
480 mg of an off-white solid (63%): Rf ) 0.20 (20% ether/hexanes);
Conversely, acyclic and larger-ring ketones appear to be
perfectly good substrates for the Mannich process so long as
benzyl azide is used (Table 4, entries 9 and 10, and Tables 5
and 6). The main requirement for reaction in this manifold seems
to be simple enolizability, although additional functionality could
prove problematic in more complex examples (Table 6, entry
15).
Summary. The Lewis acid-mediated reactions of simple
ketones with alkyl azides have been explored. It has been shown
that the Schmidt reaction of alkyl azides, a reaction proposed
over 50 years ago, can occur but is only synthetically useful
for a fairly restricted set of carbonyl-containing substrates.
Fortunately, it is possible to overcome many of these limitations
in the reactions of ketones with hydroxy azides,6 which offer
considerably greater synthetic flexibility (due to the presence
of an ω-hydroxy or other18 functional group in the product) as
well as efficiency and scope. In the course of this work, the
rearrangement of benzyl azides and subsequent Mannich reac-
tion was discovered. The availability of this pathway imposes
a limitation on the use of benzyl azide in the intermolecular
Schmidt (and, to a lesser degree, on the intramolecular version
as well19) but is significantly more forgiving with respect to
ketone structure. Overall, these reactions further solidify the
view of azides as viable nucleophilic species under favorable
conditions.
1
mp 86-87 °C; IR (CH2Cl2) 3360, 1690 cm-1; H NMR (400 MHz,
CDCl3) δ 7.18 (dd, J ) 8.5, 7.4 Hz, 2H), 6.71 (t, J ) 7.3 Hz, 1H),
6.61 (d, J ) 7.7 Hz, 2H), 4.17 (br s, 1H), 3.45 (dd, J ) 13.6, 7.7 Hz,
1H), 3.13 (dd, J ) 13.6, 4.6 Hz, 1H), 2.66 (m, 1H), 2.44 (m, 2H), 2.33
(m, 1H), 2.22-2.07 (m, 2H), 1.94 (m, 1H), 1.78-1.63 (m, 2H), 1.51
(dq, J ) 12.7, 3.8 Hz, 1H); 13C NMR (400 MHz, CDCl3) δ 213.5,
148.5, 129.7, 117.7, 113.3, 50.3, 44.3, 42.6, 32.5, 28.2, 25.3; CIMS
m/z (relative intensity) 204 (MH+, 30), 106 (95), 77 (15); HRMS calcd
for C15H19NO3: 261.1365, found: 261.1342.
Experimental Section
Spectral data for all new compounds are collected in the Supporting
Information.
General Experimental Procedure for Schmidt Reactions Using
TiCl4. 4-Hexyl-4-azahomoadamantane (30). To a solution of 2-ada-
mantanone (119 mg, 0.79 mmol) and 1-azidohexane (200 mg, 1.58
mmol) in 2.5 mL of methylene chloride, in an ice bath, was added
dropwise TiCl4 (380 mg, 2.0 mmol). The reaction was allowed to warm
to room temperature with immediate gas evolution noted. A precipitate
Acknowledgment. The authors thank Aaron Wrobleski for
experimental assistance and Lawrence Seib for X-ray crystal-
lographic studies. This work was funded by the National
Institutes of Health (GM-49093).
(18) Gracias, V.; Milligan, G. L.; Aube´, J. J. Org. Chem. 1996, 61, 10-
11.
Supporting Information Available: Characterization data
for all new compounds, details of the X-ray crystallographic
determination of compound 50, and copies of the NMR spectra
of compounds 13-16 (PDF). This material is available free of
charge via the Internet at http://pubs.acs.org.
(19) In some cases, alternative reaction pathways have been observed
to intervene when intramolecular Schmidt reactions are attempted with
benzylic azides (Morton, M. and Wrobleski, A., unpublished results from
this laboratory).
(20) Oliveros, E.; Rivie`re, M.; Lattes, A. NouV. J. Chim. 1979, 3, 739-
753.
(21) Murahashi, S.-I.; Naota, T.; Saito, E. J. Am. Chem. Soc. 1986, 108,
8, 7846-7847.
JA000490V