Radical chain reactions of α-azido ketones with tributyltin hydride: Reduction vs nitrogen insertion and 1,2-hydrogen shift in the intermediate N-stannylaminyl radicals

Article abstract of DOI:10.1016/S0040-4020(02)00302-2

The radical chain reactions of a variety of acyclic and cyclic α-azido ketones with tributyltin hydride have been investigated. The derived N-(tributylstannyl)aminyl radicals normally undergo H-abstraction reaction yielding corresponding amines, and thence symmetrical pyrazines by subsequent self-condensation, in competition with 1,2-H-migration from the α-carbon to nitrogen leading to α-imino ketone decomposition products with loss of the chain-carrying tributyltin radical. The noteworthy occurrence of a quite uncommon radical 1,2-hydrogen-atom shift is considered to be largely due to consequent formation of a highly stable, captodative carbon-centred radical. In contrast with our previous N-stannylaminyl radicals produced from α-azido-β-keto esters, the present aminyl congeners give poor amounts (or even none) of nitrogen-inserted amides/lactams, which are envisaged to arise from intramolecular three-membered cyclisation onto the ketone moiety followed by β-scission of the resultant alkoxyl radical. It is inferred that adequate stabilisation of the eventual ring-opened carbon radical be a major factor for the successful outcome of the regiospecific nitrogen insertion process. Evidence is also presented that chemoselective attack of tris(trimethylsilyl)silyl radical to the ketone oxygen of an α-azido ketone gives rise to deazidation as a likely consequence of β-elimination of azidyl radical by the ensuing α-silyloxyalkyl radical. X-Ray crystal structure analyses of the bromo ketone 5a, the azido ketone 5b, the caprolactam 22, and the pyrazine 26 have been performed.

Full text of DOI:10.1016/S0040-4020(02)00302-2

TETRAHEDRON
Pergamon
Tetrahedron 58 (2002) 3485±3492
Radical chain reactions of a-azido ketones with tributyltin
hydride: reduction vs nitrogen insertion and 1,2-hydrogen shift in
the intermediate N-stannylaminyl radicals
Luisa Benati,a,p Rino Leardini,a Matteo Minozzi,a Daniele Nanni,a Piero Spagnolo,a,p
Samantha Strazzari,a Giuseppe Zanardia and Gianluca Calestanib
a
Á
Dipartimento di Chimica Organica `A. Mangini', Universita di Bologna, Viale Risorgimento 4, I-40136 Bologna, Italy
b
Á
Dipartimento di Chimica Generale ed Inorganica, Analitica e Chimica Fisica, Universita di Parma, Viale delle Scienze 78,
I-43100 Parma, Italy
Received 12 February 2002; accepted 14 March 2002
AbstractÐThe radical chain reactions of a variety of acyclic and cyclic a-azido ketones with tributyltin hydride have been investigated. The
derived N-(tributylstannyl)aminyl radicals normally undergo H-abstraction reaction yielding corresponding amines, and thence symmetrical
pyrazines by subsequent self-condensation, in competition with 1,2-H-migration from the a-carbon to nitrogen leading to a-imino ketone
decomposition products with loss of the chain-carrying tributyltin radical. The noteworthy occurrence of a quite uncommon radical 1,2-
hydrogen-atom shift is considered to be largely due to consequent formation of a highly stable, captodative carbon-centred radical. In
contrast with our previous N-stannylaminyl radicals produced from a-azido-b-keto esters, the present aminyl congeners give poor amounts
(or even none) of nitrogen-inserted amides/lactams, which are envisaged to arise from intramolecular three-membered cyclisation onto the
ketone moiety followed by b-scission of the resultant alkoxyl radical. It is inferred that adequate stabilisation of the eventual ring-opened
carbon radical be a major factor for the successful outcome of the regiospeci®c nitrogen insertion process. Evidence is also presented that
chemoselective attack of tris(trimethylsilyl)silyl radical to the ketone oxygen of an a-azido ketone gives rise to deazidation as a likely
consequence of b-elimination of azidyl radical by the ensuing a-silyloxyalkyl radical. X-Ray crystal structure analyses of the bromo ketone
5a, the azido ketone 5b, the caprolactam 22, and the pyrazine 26 have been performed. q 2002 Elsevier Science Ltd. All rights reserved.
1.Introduction
tendency to react with azides yielding N-stannyl-substituted
aminyl radicals, which act as the key intermediates in
several important synthetic processes. In fact, the radical
chain reactions of tributyltin hydride (Bu3SnH) with alkyl
and acyl azides result in straightforward production of
reduced amines8,9 or rearranged imines10 through transient
N-(tributylstannyl)-aminyl radicals. Similar chain reactions
of allylstannanes with sulfonyl azides result in smooth
allylation of the derived N-(tributylstannyl)sulfonamidyl
radicals.2 Moreover, N-tributyltin-substituted aminyl
radicals, being somewhat nucleophilic than ordinary aminyl
radicals,9 are highly capable of performing intramolecular
®ve- and six-membered cyclisation onto carbonyl groups to
give otherwise inaccessible medium-sized lactams by
b-fragmentation of the ensuing alkoxyl radicals.11 Follow-
ing these data, very recently we provided a novel entry to
amides and lactams through regiospeci®c nitrogen insertion
of b-keto ester compounds by using the radical chain
reactions of the a-azido derivatives with Bu3SnH.12 In
such circumstances the derived stannylaminyl radical
undergoes 3-exo cyclisation onto the adjacent ketone
group to form an alkoxyl radical, which then undergoes
b-scission and eventual reduction of the ring-opened radical
(Scheme 1).
Organic azides are versatile nitrogen intermediates whose
reactivity and synthetic potential have been investigated
extensively under thermal and photochemical conditions
both in the presence and in the absence of nucleophilic or
electrophilic species.1 On the other hand, radical reactions
of the organic azides have been by far less explored, though
recent studies have revealed that radical species can add to
the a- or g-position of the azido moiety to give an aminyl
radical after loss of molecular nitrogen by the initial 1,3-
and/or 3,3-triazenyl adduct.2 Intramolecular additions of
carbon-centred radicals, including the aryl,3 thiocarbonyl,4
alkyl,5 and vinyl6 members, to alkyl and aryl azides have
been shown to provide useful synthetic routes to N-hetero-
cycles. Moreover, very recently the intermolecular addition
of alkyl radicals to sulfonyl azides has been exploited in two
preparatively attractive methods for the radical azidation of
alkyl iodides and dithiocarbonates.7 Among heteroatom-
centred radicals, especially the stannyl ones exhibit a fair
Keywords: azides; radicals; aminyls; rearrangements; migrations.
Corresponding authors. Tel.: 139-51-2093623; fax: 139-51-2093654;
e-mail: benati@ms.fci.unibo.it
p
0040±4020/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020(02)00302-2

3486
L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
Scheme 1. Nitrogen insertion mechanism in a-azido-b-ketoesters.
The driving force of the regiospeci®c b-scission would be
provided by resonance stabilisation of the amide group
formed and, additionally, by the formation of the capto-
datively stabilised ring-opened radical. However, in certain
cases the occurrence of the nitrogen insertion process was
seriously limited by competing reduction of stannylaminyl
radical to amine and/or noticeable deazidation of the keto
ester substrate by the stannyl radical.
Figure 1. The X-ray molecular structure of 5b showing the atom-number-
ing scheme.
b-keto esters. The hitherto unknown 2-azido-6-phenyl-
cyclohexanone 5b, whose structure was fully established
by X-ray crystallographic analysis (Fig. 1), was obtained
in the nearly exclusive cis-con®guration upon reaction of
the corresponding cis-bromide 5a with sodium azide in
DMSO. The bromide 5a was in turn obtained in a stereo-
selective fashion by spontaneous isomerisation of the
bromide 4c (over ca. 12 days at room temperature), prob-
ably through initial enolisation and subsequent ionisation
yielding an allylic carbonium ion (Scheme 2). The observed
retention of con®guration in the conversion of cis-bromide
5a into cis-azide 5b might be the result of the intervention of
an SN1mechanism and/or isomerisation of the trans- into
the thermodinamically more stable cis-isomer through
enolisation. All the prepared azides were preliminarily
proven to show no evident sign of decomposition when
being heated in re¯uxing benzene.
In light of our interesting ®ndings with a-azido-b-keto
esters, we next were prompted to extend our study to
analogous radical chain reactions of simple a-azido ketones.
We, in fact, became interested in gaining further useful
information about the attractive reactivity of (tributyl-
stannyl)aminyl radicals toward ketone groups and,
especially, in ascertaining whether our previous nitrogen
insertion method might similarly apply to that type of keto
azide, despite the fact that, owing to lack of the ester sub-
stituent, no captodative stabilisation of the eventual ring-
opened intermediate would be provided. Herein we report
our results with the acyclic and cyclic azido ketones 1b±7b
which are shown in Chart 1.
2.Results and discussion
All the azides 1b±7b were reacted with Bu3SnH (1.1 equiv.)
in re¯uxing benzene in the presence of 2,20-azobis(2-
methylpropionitrile) (AIBN, 0.1equiv.), using conditions
strictly comparable with those previously employed for
the corresponding reactions of azido keto esters (Procedure
A). The reactions were normally prolonged for ca. 3±5 h
and the crude mixtures were then directly subjected to
column chromatography. Since our present azides, but to
an extent largely dependent upon their structural features,
appeared to yield a-imino ketone products with release of
the chain-carrying tributyltin radical, we were led to investi-
gate the possible effect of using a catalytic amount of
the hydride reagent under high dilution conditions
(Procedure B).
The already known azides 1b±4b,6b,7b were readily
prepared by a-bromination of the parent ketones
1a±4a,6a,7a and subsequent displacement of bromide by
azide ion, by using a procedure similar to that we had
previously employed to prepare the a-azido derivatives of
The reaction of phenacyl azide 1b with Bu3SnH, following
Chart 1.
Scheme 2. Bromide migration in ketone 4c.

L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
3487
Scheme 3. Reaction pathways for phenacyl azide 1b. Yields in round brackets refer to Procedure A, yields in square brackets refer to Procedure B.
Procedure A, gave no evidence at all for any production of
N-methylbenzamide, which was the expected product of
regiospeci®c nitrogen insertion, but instead gave the
diphenylpyrazine 10 as well as minor amounts of the
imidazole 13 as the only identi®able products (Scheme 3).
still unclear, but the reaction is likely to proceed via inter-
mediate imines.9 The imino ketone 12 might have also been
arisen from a competing process promoted by some base
present in the reaction medium (possibly the amine 8).
a-Azido ketones having a-hydrogens are indeed known to
be base-sensitive and undergo base-promoted loss of
nitrogen to form a-imino ketones (or 1,2-diketones by
hydrolysis)16 and/or tautomeric a-amino enones.17
However, we ascertained that phenacyl azide 1b, when
being heated in re¯uxing benzene in the presence of even
1equiv. of triethylamine, could just suffer conversion into
imidazole 13 at a very low rate. Indeed, no trace of 13 was
observed after 3 h (the time required for complete dis-
appearance of the starting azide in the radical reaction)
and a yield as low as 10% was obtained only after 8 h.
Consequently, a possible intervention of a base-promoted
process during the radical reaction of azide 1b was believed
to be unimportant.
In another experiment the resulting crude reaction mixture
allowed direct crystallisation of the dihydropyrazine 9; such
compound 9 was found to be fairly stable in the solid state,
but it underwent fairly rapid aromatisation to 10 in benzene
solution. The dihydropyrazine 9 (and thence the aromatised
compound 10) was presumably due to self-condensation of
the stannylated amino ketone 8, which could be formed by a
viable H-abstraction reaction of the derived stannylaminyl
radical with the tin hydride (Scheme 3). In fact, symmetrical
pyrazines, including the pyrazine 10, are known to form by
the spontaneous self-condensation of a-amino ketones
produced in situ by reduction of the corresponding azides.13
Interestingly, the imidazole 13 was instead presumably due
to self-condensation of an alternate a-imino ketone inter-
mediate 12. Although formation of the imidazole 13 by
dimerisation of the compound 12 has been previously
described,14 it is worth noting that it occurred by rearrange-
ment of an initial nitrene, obtained by pyrolysis of phenacyl
azide 1b at 180±2008C. In our case, we suggest that also the
presumed imine 12 should be attributed to a transient
stannylaminyl radical. Indeed, the intermediate 12 might
in principle result from a radical 1,2-H rearrangement
from carbon to nitrogen, followed by b-elimination of the
chain-carrying tributyltin radical from the resultant capto-
dative radical 11 (Scheme 3). Radical 1,2-H-migrations are
quite unusual in carbon-centred radicals,15 whereas such
type of migration possibly occurs, but to a (very) limited
extent, in the N-stannylaminyl radicals that are formed upon
addition of tributyltin radicals to simple alkyl azides.9 In
fact, the radical reaction of terminal alkyl azides with
Bu3SnH normally affords aldehydes in variable amounts,
depending on the nature of the substrate. The actual
mechanism for the conversion of azides to aldehydes is
With the aim of achieving some solid support to the
believed radical occurrence of the imine 12, the above reac-
tion of phenacyl azide 1b was repeated by performing very
slow addition of 0.3 equiv. of the tin hydride by syringe
pump over ca. 3 h (Procedure B). Using such a reduced
amount of Bu3SnH the azide 1b could still undergo virtually
complete reaction, this fact being consistent with some cata-
lytic role being played by the hydride. Moreover, using such
a very low concentration of the hydride there was preferen-
tial occurrence of the imidazole 13 at the expense of the
competing pyrazine 10: this is consistent with expected
enhancement of the unimolecular aminyl radical rearrange-
ment at the expense of the bimolecular reduction. Addition-
ally, the reaction also afforded a signi®cant amount of the
amine 14, which is plausibly formed by trapping of the
rearranged radical 11 by the 2-cyano-2-propyl radical
(CPR) arising from fragmentation of AIBN (Scheme 3).
These overall data furnished sound support to our view
that an initial stannylaminyl radical should be primarily
involved in the formation of the presumed intermediate 12

3488
L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
Scheme 4. Reaction products of phenacyl azide 2b. Yields in brackets refer
to Procedure A.
and thence of the isolated imidazole 13. Consequently, these
data also provided the ®rst convincing evidence in favour of
1,2-H transfer reaction in N-tributylstannyl-substituted
aminyl radicals.
Scheme 6. Reaction products of azide 5b. Yields in round brackets refer to
Procedure A, yields in square brackets refer to Procedure B.
It is worth noting that another radical route to the imine 12
might involve intermolecular H-transfer from the stannyl-
ated amine 8 to the primary stannylaminyl radical. Although
this pathway cannot be totally excluded, however we
consider it unimportant, since it seems unlikely that amine
8 could act as a better hydrogen donor than the stannane.
imino propanone analog of the imino ethanone 12 was
obtained (Scheme 4).
In an attempt to intercept the possibly formed imino propa-
none we then repeated the reaction in the presence of
o-phenylenediamine (PDA). This aromatic diamine is in
fact known to trap reactive a-imino ketones, yielding
condensed quinoxaline products.18 In the presence of PDA
we actually succeeded in isolating the disubstituted
quinoxaline 17, albeit in a poor yield (Scheme 4), thus
proving that some production of unstable 1-phenyl-2-imino-
propanone should have evidently occurred.
The stannylaminyl radical derived from phenacyl azide 1b
is thus not capable of performing regiospeci®c nitrogen
insertion to give N-methylbenzamide, probably since the
required b-fragmentation of the alkoxyl radical inter-
mediate would form a poorly stable, primary a-aminoalkyl
radical (Scheme 3). The same stannylaminyl radical is
instead interestingly prone to exhibit smooth H-shift from
carbon to adjacent nitrogen, probably owing to consequent
formation of a highly stable carbon radical such as the
captodative radical 11. That rearrangement can compete
ef®ciently with hydrogen transfer from the hydride and
becomes even preferred if the hydride concentration is
suitably lowered.
Therefore, the aminyl radical arising from the azide 2b,
similar to the one formed from 1b, can undergo both
H-abstraction reaction, to give the corresponding amine,
and 1,2-H-migration, to give a rearranged imine. However,
this radical was also enabled to perform nitrogen insertion
yielding the amide 16, although to a limited extent.
Evidently, in such case the regiospeci®c b-scission of the
cyclised alkoxyl radical becomes rather feasible since it can
form a secondary aminoalkyl radical somewhat more stable
than the primary one obtainable from 1b.
The reaction of the methylated phenacyl azide 2b, under the
standard conditions of Procedure A, similarly furnished the
corresponding pyrazine 15 in a comparable yield with that
of the above diphenylpyrazine 10, but, interestingly, it
concomitantly furnished a signi®cant amount of the
nitrogen-inserted benzamide 16. The reaction gave also a
small amount of the ketone 2a, whose formation can be
ascribed to a known deazidation process involving pre-
liminary attack of the stannyl radical to the terminal azide
nitrogen.12 No diagnostic evidence for any occurrence of an
Our successive ®ndings obtained with the cyclic azide 3b
revealed that the respective aminyl radical behaves in a
fashion similar to that displayed by the radical congener
produced from the acyclic counterpart 2b. When the azide
3b was in fact reacted with the tin hydride according to
Procedure A, the tricyclic pyrazine 18 was obtained in rela-
tively high yield along with a small amount of the ring
expanded caprolactam 19 (Scheme 5). Moreover, when
the reaction was repeated using Procedure B and in the
presence of PDA, the formation of 18 was found to be
largely depressed in favour of the tetrahydrophenazine 20,
that under these conditions was formed to a noticeable
extent as a result of effective trapping of labile a-imino-
cyclohexanone19 by the aromatic diamine (Scheme 5).
The stannylaminyl radicals that were subsequently
produced from the isomeric phenyl-substituted 2-azido-
cyclohexanones 4b and 5b showed a (much) higher
tendency to give rise to their nitrogen-inserted lactam
products. The usual reaction of the 2-azido-6-phenylcyclo-
Scheme 5. Reaction products of azide 3b. Yields in round brackets refer to
Procedure A, yields in square brackets refer to Procedure B.

L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
3489
Scheme 7. Reaction products of azide 4b. Yields in brackets refer to
Procedure A.
hexanone 5b indeed led to isolation of the pyrazine 21,
which still occurred as the major product, together with
the caprolactam 22, which was obtained in a comparable,
signi®cant yield. This reaction also afforded a small amount
of the hydroxyenone 23, which was the hydrolytic product
of the corresponding phenyl-substituted a-imino-cyclo-
hexanone (Scheme 6).
Scheme 9. Reaction products of azide 7b. Yields in round brackets refer to
Procedure A, yields in square brackets refer to Procedure B.
It is worth noting that, in line with similar evidence provided
by the above azides 1b and 3b, the alternate use of
Procedure B resulted in marked suppression of pyrazine
21, and to some extent of lactam 22, in favour of enone
23 (and, plausibly, of its imino ketone precursor)
(Scheme 6).
conversion of 2-bromo-2-phenylcyclohexanone 4c into the
6-bromide 5a (Scheme 2). However, in the case of the azide
4b, the presence of the stannane would play some activating
role in view of the fact that 4b remained totally unchanged
in re¯uxing benzene in the absence of the stannane.
The isomeric 2-azido-2-phenylcyclohexanone 4b led to
isolation of the respective caprolactam 25 in a yield syn-
thetically useful and identical to that of the reduced amine
24, that was seemingly unable to undergo self-condensation
to pyrazine (Scheme 7).
Despite their analogies with the above congeners, the
stannylaminyl radicals obtained from the benzocyclic
azidoalkanones 6b and 7b were curiously highly reluctant
to exhibit any nitrogen insertion process. Under the con-
ditions of Procedure A, the indanone azide 6b gave the
pyrazine 26 as the sole identi®able product (Scheme 8),
whereas the chromanone azide 7b, in addition to its
pyrazine 28, also furnished small amounts of the enamine
29 and the dione 30. These latter compounds were the
expected products of tautomerisation and hydrolysis of an
initially formed iminochromanone (Scheme 9).
Whereas the marked propensity of the aminyl radical
obtained from 4b to yield the corresponding caprolactam
25 is likely ascribable to fair stability of the tertiary benzyl
radical formed by usual fragmentation of the alkoxyl radical
precursor, at this stage we have no convincing explanation
for the enhanced tendency seemingly displayed by the
aminyl radical of the cyclohexanone azide 5b to give rise
to its caprolactam 22.20
Under the alternate conditions of Procedure B, the azide 6b
only allowed isolation of a modest yield of the amino-
indanone 27, ascribable to coupling between CPR and the
captodative radical deriving from the stannylaminyl radical
by 1,2-H-shift (Scheme 8). The azide 7b, instead, only gave
the enamine 29 to a noticeable extent (Scheme 9). We
suggest that the aminyl radicals derived from these benzo-
cyclic azides, owing to possible conformational restraints,
are essentially discouraged to afford their cyclised alkoxyl
adducts and thence only undergo H-abstraction and/or
additional 1,2-H-shift. Once again, for both reactions,
high-dilution conditions make the hydrogen atom migration
the major process of the reaction mechanism.
It is worth noting that the azide 4b, besides affording the
amine 24 and the lactam 25 (as well as some deazidated
ketone 4a), additionally gave small amounts of the isomeric
lactam 22 and the enone 23 (Scheme 7), thus suggesting that
under the reaction conditions the azide itself could also
undergo isomerisation to 5b. In a control experiment,
TLC actually showed that the initial production of the
compounds 24 and 25 was accompanied by concomitant
formation of the azide isomer 5b. It is possible that the
observed isomerisation of 4b to 5b might proceed through
a mechanism similar to that above postulated for the related
In a search for a possible extension of our a-azido ketone
reactions, in the present work we brie¯y studied the radical
chain reactions of tris(trimethylsilyl)silane with the
indanone azide 6b and also with 2-ethoxycarbonyl-1-tetra-
lone azide.12 In light of the previous ®ndings of Kim and
co-workers that azides are relatively inert toward tris(tri-
methylsilyl)silyl radicals,5 the silyl radical was expected
to add selectively on the ketone oxygen rather than on the
azido moiety. We therefore aimed at investigating whether
the resultant a-(silyloxy)alkyl radical could be able to
perform intramolecular 3-exo cyclisation onto the adjacent
Scheme 8. Reaction products of azide 6b. Yields in round brackets refer to
Procedure A, yields in square brackets refer to Procedure B.

3490
L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
aminyl reduction. Moreover, trapping of the rearranged
captodative radical by CPR, which was observed with two
different azides (1b, 6b), provides an additional, ultimate
proof of the migration.
Finally, the present work has also shown that when using
tris(trimethylsilyl)silane in place of the tin hydride the
radical reaction of an a-azido ketone would only result in
deazidation, owing to preferred elimination of azidyl radical
by the resultant a-(silyloxy)alkyl radical.
Scheme 10. Reaction of azide 6b with silyl radicals.
4.Experimental
azido group to form a synthetically attractive aminyl radical
by subsequent loss of nitrogen. However, treatment of 6b
with a slight excess of (TMS)3SiH and a catalytic amount of
AIBN in re¯uxing benzene gave only the deazidated inda-
none 6a in fairly good yield. Strictly comparable results
were obtained from the same reaction with 1-tetralone
azide. In both cases the derived oxyalkyl radical was there-
fore seemingly able to perform only b-elimination of azidyl
radical to give an intermediate enol ether and hence the
deazidated substrate by eventual hydrolysis (Scheme 10).
4.1. General procedures
All melting points (Ko¯er melting point apparatus) are
uncorrected. 1H and 13C NMR spectra were normally carried
out in CDCl3 solutions, using tetramethylsilane as the
internal standard. Mass spectra were determined by the
electron impact method (70 eV). IR spectra were recorded
in CHCl3 solutions. Column chromatography was
Ê
performed on ICN silica gel (63±200 A) by gradual elution
with light petroleum (bp 40±708C)/diethyl ether and/or
ethyl acetate/diethyl ether mixtures and ®nal elution with
ethyl acetate/diethyl ether/methanol mixtures.
3.Conclusions
Our overall present and previous ®ndings have shown that
the radical chain reactions of Bu3SnH with a-azido ketones
can usefully afford amides and lactams provided that the
alkoxyl radical arising from 3-exo cyclisation of a transient
(tributylstannyl)aminyl radical onto the ketone moiety
could undergo b-scission to form an adequately stabilised
carbon radical. Very ef®cient or fair stabilisation can be
furnished by ester12 or aryl (4b) moieties, respectively,
whereas alkyl groups can provide stabilisation only to
some extent (2b, 3b, 5b). The outcome of the nitrogen
insertion process can however be limited by the possible
occurrence of conformational restraints in the 3-exo
cyclisation step, as suggested for compounds 6b and 7b.
4.2. Materials
The starting ketones 1a±7a, caprolactam 19, as well as
n-tributyltin hydride were commercially available, and
were used as received. 2,20-Azobis(2-methylpropionitrile)
(AIBN, Fluka) and dibenzoyl peroxide (DBP) were
recrystallised from CHCl3/CH3OH.
2-Azido-1-phenyl-1-propanone (2b) was prepared accord-
ing to the literature.22 The already known azido ketones 1b,
3b, 4b, and 7b were prepared in 70±85% yield through
bromination of the parent ketones 1a, 3a, 4a, 7a with
N-bromosuccinimide (NBS, 1equiv.) and a catalytic
amount of DBP in re¯uxing tetrachloromethane (15 min±
10 h) and subsequent treatment of the crude bromide with
sodium azide (2 equiv.) in dimethyl sulfoxide (DMSO).12
The azidoindanone 6b was similarly prepared from the
corresponding bromide, in turn obtained by treatment of
indanone 6a with bromine (1equiv.) in diethyl ether at
room temperature.
When the driving force of the nitrogen insertion is not strong
enough, the competitive reduction of the original stannyl-
aminyl radical leads to an a-amino ketone and then to the
corresponding symmetrical pyrazine, provided that no addi-
tional substituent is present in the precursor a-position (like
in 4b).
The most noteworthy ®ndings are that the stannylaminyl
radicals are normally found to yield a-imino ketone
decomposition products as a result of unusually easy 1,2-
H-migration, whose driving force is essentially ascribable to
the consequent formation of a captodative, highly stabilised
carbon-centred radical.21 The occurrence of this rearrange-
ment, already suggested for similar intermediates but never
proved, was de®nitely supported by the results of the
reactions carried out under catalytic conditions. Indeed, a
1,2-H-shift followed by loss of a tributyltin radical is con-
sistent with the complete disappearance of the starting azide
when using signi®cantly less then 1equiv. of Bu 3SnH
(Procedure B). The observed enhancement of the yields of
a-imino ketone decomposition products at the expense of
the pyrazines is also consistent with the occurrence of a
unimolecular mechanism competing with the bimolecular
The azides 1b, 3b, 6b, and 7b had spectral data consistent
with those previously reported.23a Analytical data of azide
4b were as follows: mp 27±288C (lit,24 an oil), nmax 2099
(N3), 1717 (CO) cm21, 300 MHz 1H NMR d 1.62±2.02 (5H,
m), 2.32±2.54 (2H, m), 2.78±2.85 (1H, m), 7.27±7.50 (5H,
m), 50 MHz 13C NMR d 21.92, 27.45, 36.34, 39.91, 74.46
(q), 126.72, 129.05, 129.42, 135.20 (q), 206.93 (q).
Hitherto unknown cis-2-azido-6-phenylcyclohexanone (5b)
was similarly prepared by reacting sodium azide in DMSO
with the cis-bromide 5a, which was available by spon-
taneous isomerisation of the bromide 4c both in the solid
state and in solution (over ca. 12 days at room temperature).
The bromide 5a had: mp 102±1038C, 300 MHz 1H NMR d
1.91±2.40 (5H, m), 2.74 (1H, m), 3.70 (1H, br dd,

L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
3491
J1ˆ5.6 Hz, J2ˆ12.4 Hz), 4.78 (1H, ddd, J1ˆ6.0 Hz,
J2ˆ12.4 Hz, J3ˆ1.0 Hz), 7.10±7.40 (5H, m), 50 MHz 13C
NMR d 27.09, 36.07, 40.91, 57.25, 57.73, 128.04, 129.08,
129.28, 138.22 (q), 200.78 (q). Anal. calcd for C12H13BrO:
C, 56.94; H, 5.18; Br, 31.57. Found: C, 56.86; H, 5.17; Br,
31.51. The structure of 5a was con®rmed by X-ray
diffraction.
128.56, 130.48, 141.19 (q), 165.55 (q), MS m/e (rel. inten.)
234 (M1, 48), 131 (100), 103 (92). Anal. calcd for
C16H14N2: C, 82.02; H, 6.02; N, 11.96. Found: C, 81.95;
H, 6.01; N, 12.00.
Using Procedure A or B, the reaction mixture was even-
tually concentrated under reduced pressure and the resultant
residue was normally subjected to direct column chromato-
graphy. In the reaction of azide 7b, following procedure A,
chromatographic product separation proved to be quite
dif®cult due to slow oxidation of the resulting dihydro-
pyrazine (M1, 290), attempted oxidisation of this compound
by bubbled air resulted in essential production of an
unidenti®ed compound (M1, 306).
The azide 5b had: mp 72±748C, nmax 2105 (N3), 1731
(CO) cm21, 200 MHz 1H NMR d 1.70±2.13 (4H, m),
2.21±3.49 (2H, m), 3.55±3.68 (1H, m), 4.00±4.12 (1H,
m), 7.10±7.40 (5H, m), 50 MHz 13C NMR d 24.32, 35.17,
36.25, 57.62, 66.63, 128.87, 130.11, 130.38, 138.77 (q),
206.94 (q). The structure of 5b was con®rmed by X-ray
diffraction.
Yields of the isolated products 2a, 4a, 10, 13±30 are given
in Schemes 3±9. Analytical data of the new compounds 14,
21, 26±28 were as follows.
All the prepared azides 1b±7b were found to be quite stable
in re¯uxing benzene. In re¯uxing benzene and in the
presence of 1equiv. of triethylamine the phenacyl azide
1b was found to remain virtually unchanged over ca. 3 h,
but to suffer some decomposition to imidazole 13 (ca. 10%)
after 8 h.
4.3.4.
3-Amino-2,2-dimethyl-4-oxo-4-phenylbutane-
nitrile (14). Oil, nmax 3380, 3310 (NH2), 2239 (CN), 1680
(CO) cm21, 200 MHz 1H NMR d 1.38 (3H, s), 1.45 (3H, s),
1.90 (2H, br s), 4.39 (1H, s), 7.47±7.54 (2H, m), 7.60±7.64
(1H, m), 7.94±8.02 (2H, m), 50 MHz 13C NMR d 23.79,
25.15, 37.91 (q), 59.70, 123.84 (q), 129.26, 129.64, 134.72,
136.81 (q), 199.39 (q). Anal. calcd for C12H14N2O: C, 71.26;
H, 6.98; N, 13.85. Found: C, 71.18; H, 6.98; N, 13.82.
The known reaction products 10,25 15,26 16,27 17,28 18,29
20,30 22,31 23,32 24,33 and 2534 were authenticated by
spectral comparison with the literature data. 2-Benzoyl-5-
phenylimidazole (13),14,35 caprolactam 19 and 3-amino-4H-
chromen-4-one (29)23b were identi®ed by spectral com-
parison with authentic samples. The dione 30 was identi®ed
following a procedure reported in the literature.36
4.3.5. 1,6-Diphenyl-1,2,3,4,6,7,8,9-octahydrophenazine
(21). (Seemingly one stereoisomer) mp 170±1728C,
1
200 MHz H NMR d 1.70±2.35 (8H, m), 2.73±3.09 (4H,
m), 4.33 (2H, t, Jˆ6.0 Hz), 6.95±7.08 (4H, m), 7.19±7.44
(6H, m), 50 MHz 13C NMR d 19.69, 32.31, 33.00, 47.57,
126.71, 128.88, 129.45, 146.19 (q), 151.49 (q), 151.61 (q).
MS m/e (rel. inten.) 340 (M1, 100). Anal. calcd for
C24H24N2: C, 84.67; H, 7.11; N, 8.23. Found: C, 84.81; H,
7.09; N, 8.30.
4.3. Reactions of azido ketones with tributyltin hydride
4.3.1. Procedure A. A benzene (80 mL) solution containing
the appropriate keto azide (4 mmol), the tin hydride
(4.4 mmol), and AIBN (0.4 mmol) was re¯uxed under a
nitrogen atmosphere for 3±5 h, until TLC or IR monitored
the virtual disappearance of the starting azide. In the case of
azide 2b, the reaction was carried out also in the presence of
o-phenylenediamine (2 mmol) (Scheme 4).
4.3.6. 6,12-Dihydrodiindeno[1,2-b:1,2-e]pyrazine (26).
Mp 268±2698C, 200 MHz H NMR d 4.04 (4H, s), 7.41±
1
7.56 (4H, m), 7.60±7.67 (2H, m), 8.09±8.15 (2H, m),
50 MHz 13C NMR d 39.98, 121.00, 125.43, 127.67,
129.07, 138.84 (q), 142.12 (q), 151.60 (q), 157.18 (q). MS
m/e (rel. inten.) 256 (M1, 100). Anal. calcd for C18H12N2: C,
84.35; H, 4.72; N, 10.93. Found: C, 84.50; H, 4.70; N, 10.96.
The structure of 26 was con®rmed by X-ray diffraction.
4.3.2. Procedure B. A benzene (12 mL) solution of Bu3SnH
(1.2 mmol) and AIBN (0.4 mmol) was added by a syringe
pump over 3 h to a re¯uxed benzene (65 mL) solution
containing the appropriate keto azide (4 mmol) under a
nitrogen atmosphere. The resulting mixture was re¯uxed
for additional 3±5 h until the virtual disappearance of the
starting azide. In the case of azide 3b, the reaction was
performed in the presence of o-phenylenediamine
(2 mmol) (Scheme 5).
4.3.7. 2-(2-Amino-1-oxo-2,3-dihydro-1H-inden-2-yl)-2-
methyl-propanenitrile (27). Oil, nmax 3381, 3308 (NH2),
1
2221(CN), 1713 (CO) cm 21, 300 MHz H NMR d 1.30
(3H, s), 1.52 (3H, s), 2.80 (2H, br s), 3.05 (1H, d,
Jˆ17.8 Hz), 3.54 (1H, d, Jˆ17.8 Hz), 7.37±7.49 (2H, m),
7.66 (1H, m), 7.77 (1H, d, Jˆ7.6 Hz), 75.5 MHz 13C NMR
d 21.76, 23.01, 40.73 (q), 41.01, 65.21 (q), 124.19 (q),
125.44, 126.98, 128.93, 135.71 (q), 136.72, 151.3 (q),
205.89 (q). MS m/e (rel. inten.) 146 (M1 2C(CH3)2CN,
100). Anal. calcd for C13H14N2O: C, 72.87; H, 6.59; N,
13.07. Found: C, 72.75; H, 6.62; N, 13.15.
4.3.3. Reactions. All the reactions performed according to
Procedure A were monitored by H NMR and found to
1
exhibit the intermediate formation of unstable dihydro-
pyrazines, being normally converted to the corresponding
aromatised compounds by air oxidation. In the reaction of
the azide 1b the resultant 3,6-diphenyl-2,5-dihydropyrazine
(9) could be actually isolated as a solid compound upon
addition of pentane to a diethyl ether solution of the crude
reaction mixture. The compound 9, stable in the solid state
but fairly rapidly oxidisable in solution, had: mp 169±
4.3.8. 6,13-Dihydrodichromeno[3,4-b:3,4-e]pyrazine (28).
Mp 264±2668C, 300 MHz 1H NMR d 5.41(4H, s),
6.98±7.22 (4H, m), 7.30±7.41(2H, m), 8.1±8.20 (2H,
m). MS m/e (rel. inten.) 288 (M1, 100). Anal. calcd for
1
1708C, 200 MHz H NMR d 4.76 (4H, s), 7.31±7.50 (6H,
m), 7.72±7.89 (4H, m), 50 MHz 13C NMR d 50.06, 126.11,

3492
L. Benati et al. / Tetrahedron 58 (2002) 3485±3492
10. Kim, S.; Do, J. Y. J. Chem. Soc., Chem. Commun. 1995, 1607.
11. (a) Kim, S.; Joe, G. H.; Do, J. Y. J. Am. Chem. Soc. 1993, 115,
3328. (b) Kim, S.; Yoon, K. S.; Kim, S. S.; Seo, H. S.
Tetrahedron 1995, 51, 8437.
C18H12N2O2: C, 74.99; H, 4.20; N, 9.72. Found: C, 74.89; H,
4.20; N, 9.70.
4.4. Reactions of azidoindanone 6b and 2-azido-2-
ethoxycarbonyl-1-tetralone with tris(trimethylsilyl)-
silane
12. Benati, L.; Nanni, D.; Sangiorgi, C.; Spagnolo, P. J. Org.
Chem. 1999, 64, 7836.
13. (a) Nakajima, M.; Loeschorn, C. A.; Cimbrelo, W. E.;
Anselme, J.-P. Org. Prep. Proced. Int. 1980, 12, 265.
(b) Suzuki, H.; Kawaguchi, T.; Tanaoka, K. Bull. Chem.
Soc. Jpn 1986, 59, 665. (c) LaCour, T. G.; Guo, C.; Boyd,
M. R.; Fuchs, P. L. Org. Lett. 2000, 2, 33. (d) Jeong, J. U.;
Guo, C.; Fuchs, P. L. J. Am. Chem. Soc. 1999, 121, 2071and
references therein. (e) Zbiral, E.; Stroh, J. Liebigs Ann. Chem.
1969, 727, 1969.
A benzene (40 mL) solution of the azide 6b (2 mmol) was
treated with the silane (2.2 mmol) and AIBN (0.2 mmol)
and then re¯uxed for 3 h. Removal of the solvent and
column chromatography of the residue gave the deazidated
indanone 6a in 65% yield. Using the same procedure, the
azidotetralone12 gave the deazidated compound12 in 74%
isolated yield.
14. Boyer, J. H.; Straw, D. J. Am. Chem. Soc. 1952, 74, 4506.
15. Carey, F. A.; Sundberg, R. J. Part A: Structure and
Mechanisms. 4th ed; Advanced Organic Chemistry,
Academic±Plenum: New York, 2000 Chapter 12.
16. (a) Edwards, O. E.; Purushothaman, K. K. Can. J. Chem.
1964, 42, 712. (b) Boyer, J. H.; Canter, F. C. Chem. Rev.
1954, 54, 1 .
4.5. X-Ray crystallographic analyses
Crystallographic data (excluding structure factors) for the
structures in this paper have been deposited with the
Cambridge Crystallographic Data Centre as supplementary
publication numbers CCDC 175948 (5a), CCDC 175949
(26), CCDC 175950 (22), and CCDC 175951 (5b). Copies
of the data can be obtained, free of charge, on application to
CCDC, 12 Union Road, Cambridge CB2 1EZ, UK [fax:
144-(0)1223-336033 or e-mail: deposit@ccdc.cam.ac.uk].
17. (a) Nakazumi, H.; Endo, T.; Nakaue, T.; Kitao, T. J. Hetero-
cycl. Chem. 1985, 22, 89. (b) Watanabe, S.; Nakazumi, H.;
Kitao, T. J. Chem. Soc., Perkin Trans. 1 1988, 1829. (c) Van
Sant, K.; South, M. S. Tetrahedron Lett. 1987, 28, 6019.
18. Boyer, J. H.; Straw, D. J. Am. Chem. Soc. 1953, 75, 1642.
19. Edwards, O. E.; Purushothaman, K. K. Can. J. Chem. 1964,
42, 712.
Acknowledgements
20. To fully prove the actual formation of the 3-phenyl-substituted
caprolactam 22, its structure was unambiguously established
by X-ray diffraction.
The authors gratefully acknowledge ®nancial support from
MURST (2000±2001grant for `Radical Processes in
Chemistry and Biology: Syntheses, Mechanisms and Appli-
cations') and the University of Bologna (1999±2001 Funds
for Selected Research Topics).
21. Stella, L.; Harvey, J. N. Synthetic Utility of the Captodative
Effect. Radicals in Organic Synthesis; Renaud, P., Sibi, M.,
Eds.; Wiley±VCH: Weinheim, 2001; Vol. 1 Chapter 3.3.
22. Besse, P.; Veschambre, H.; Dickman, M.; Chenevert, R.
J. Org. Chem. 1994, 59, 8288.
References
23. (a) Patonay, T.; Hoffman, R. V. J. Org. Chem. 1994, 59, 2902.
(b) Patonay, T.; Hoffman, R. V. J. Org. Chem. 1995, 60, 2368.
24. Crotti, P.; Chini, M.; Uccello-Barretta, G.; Macchia, F. J. Org.
Chem. 1989, 54, 4525.
1. For leading information concerning the chemistry of azides
and nitrenes see: (a) The Chemistry of The Azido Group, Patai,
S., Ed.; Wiley: New York, 1971. (b) Azides and Nitrenes-
Reactivity and Utility, Scriven, E. F. V., Ed.; Academic:
New York, 1984. (c) Scriven, E. F. V.; Turnbull, K. Chem.
Rev. 1988, 88, 297.
25. Armand, J.; Chekir, K.; Pinson, J. Can. J. Chem. 1974, 52,
3971.
26. Suzuki, H.; Kawaguchi, T.; Takaoka, K. Bull. Chem. Soc. Jpn.
1986, 59, 665.
2. Roberts, B. P.; Dang, H.-S. J. Chem. Soc., Perkin Trans. 1
1996, 1493.
27. Murphy, J. A.; Roome, S. J. J. Chem. Soc., Perkin Trans. 1
1995, 1349.
3. Benati, L.; Montevecchi, P. C.; Spagnolo, P. Tetrahedron Lett.
1978, 815.
28. Barluenga, J.; Aznar, F.; Ramon, L.; Cabal, M. P. Synthesis
1985, 313.
4. Benati, L.; Montevecchi, P. C. J. Org. Chem. 1981, 46, 4570.
5. Kim, S.; Joe, G. H.; Do, J. Y. J. Am. Chem. Soc. 1994, 116,
5521.
29. Kobayashi, T.; Yamamoto, S.; Kato, H. Bull. Chem. Soc. Jpn
1997, 70, 1193.
30. Petukhov, P. A.; Tkachev, A. V. Tetrahedron 1997, 53, 9761.
6. Montevecchi, P. C.; Navacchia, M. L.; Spagnolo, P. Eur. J.
Org. Chem. 1998, 1219.
Â
31. Aube, J.; Hammond, M.; Gherardini, E.; Takusagawa, F.
J. Org. Chem. 1991, 56, 499.
7. Ollivier, C.; Renaud, P. J. Am. Chem. Soc. 2001, 123, 4717.
8. (a) Redlich, H.; Roy, W. Liebigs Ann. Chem. 1981, 1215.
(b) Samano, M. C.; Robins, M. J. Tetrahedron Lett. 1991,
32, 6293. (c) Poopeiko, N. E.; Pricota, T. I.; Mikhailopulo,
I. A. Synlett 1991, 342. (d) Hays, S. D.; Fu, G. C. J. Org.
Chem. 1998, 63, 2796.
32. Horiuchi, C. A.; Kiyomiya, H.; Takahashi, M.; Suzuki, Y.
Synthesis 1989, 785.
33. Parcell, R. F.; Sanchez, J. P. J. Org. Chem. 1981, 46, 5229.
34. Yanagisawa, H.; Ishihara, S.; Ando, A.; Kanazaki, T.;
Miyamoto, S.; Koike, H.; Iyime, Y.; Oizumi, K.; Matsushita,
Y.; Hate, T. J. Med. Chem. 1988, 31, 422.
9. Kim, S.; Yeon, K. M.; Yoon, K. S. Tetrahedron Lett. 1997, 38,
3919.
35. Kong, Y. C.; Kim, K. J. Heterocycl. Chem. 1999, 36, 911.
36. Vinot, N.; Maitte, P. J. Heterocycl. Chem. 1980, 17, 855.