Nonstabilised azomethine ylids from n-oxides: Unravelling the deprotonation of n-methylmorpholine n-oxide

Article abstract of DOI:10.1071/CH14217

Nonstabilised azomethine ylids (NAYs) are useful 1,3-dipoles, but their synthetic applications are restricted by the high temperatures often needed for their generation, and by an incomplete understanding of the effect of heteroatoms in cyclic systems. We have examined the behaviour of N-methylmorpholine N-oxide (NMO) as a NAY precursor in the Roussi reaction (low-temperature reaction of an N-oxide with strong base). The choice of base is critical to achieving cycloadduct formation. We report synthetic and computational (density functional theory) investigations of the products obtained with different bases and their mechanisms of formation.

Full text of DOI:10.1071/CH14217

RESEARCH FRONT
CSIRO PUBLISHING
Aust. J. Chem. 2014, 67, 1309–1317
Full Paper
http://dx.doi.org/10.1071/CH14217
Nonstabilised Azomethine Ylids from N-Oxides:
Unravelling the Deprotonation of N-Methylmorpholine
N-Oxide
A
A
Paul Malek Mirzayans, Elizabeth H. Krenske,
and Craig M. Williams
A B
,
A
School of Chemistry and Molecular Biosciences, University of Queensland,
Brisbane, Qld 4072, Australia.
B
Corresponding author. Email: c.williams3@uq.edu.au
Nonstabilised azomethine ylids (NAYs) are useful 1,3-dipoles, but their synthetic applications are restricted by the high
temperatures often needed for their generation, and by an incomplete understanding of the effect of heteroatoms in cyclic
systems. We have examined the behaviour of N-methylmorpholine N-oxide (NMO) as a NAY precursor in the Roussi
reaction (low-temperature reaction of an N-oxide with strong base). The choice of base is critical to achieving cycloadduct
formation. We report synthetic and computational (density functional theory) investigations of the products obtained with
different bases and their mechanisms of formation.
Manuscript received: 4 April 2014.
Manuscript accepted: 7 May 2014.
Published online: 4 June 2014.
Introduction
latter three are amenable to the preparation of N–R (as opposed
to N–H) substituted cycloadducts.
Azomethine ylids 1, both nonstabilised azomethine ylids
(NAYs) and stabilised azomethine ylids (SAYs), are versatile
reactive intermediates that undergo (3þ2) cycloadditions onto
double and triple bonds to afford pyrrolidines 2 and dihy-
dropyrroles 3, respectively (Scheme 1).[1] These reactions have
found widespread application in organic synthesis and have
been used in the total synthesis of complex natural products[1,2]
and in materials chemistry.[3]
Many of the leading procedures to generate NAYs require
high temperatures. A few methods can be conducted at room
temperature, but only a handful are compatible with the use of
low temperature (e.g. ꢀ78 to 08C).[1] Methods that allow the
use of low temperature involve 2-azaallyl anions (metallated
1,3-dipoles),[1b,4] desilylation of iminium salts,[1c,5] ring open-
ing of aminals,[6] and double deprotonation of tertiary amine
N-oxides (Roussi reaction[7]). Of these four classes, only the
In the Roussi reaction (Scheme 2),[8] a tertiary amine
N-oxide[9] is treated with a strong base (typically lithium
N,N-diisopropylamide, LDA) in the presence of a dipolarophile
to afford the (3þ2) cycloadduct(s). In terms of N-oxide func-
tionality, Roussi’s method has mostly been applied to N-oxides
bearing simple alkyl substituents (e.g. Scheme 2a, b). Intra-
molecular cycloadditions[10] (but not intermolecular cyclo-
additions[11]) have been accomplished with N-benzyl systems,
while there are several examples of intermolecular cyclo-
additions involving N-allyl[12] and N-phenyl[11] derivatives.
With respect to heteroatom-containing substituents, b-hydroxy
N-oxides (Scheme 2c) undergo reaction, but also lead to signifi-
cant amounts of side products, including cyclic aminals[13,14]
;
these reactions are often dominated by azomethine dimerisa-
tion,[14] which is also commonly observed in other applications
of Roussi’s method. t-Butyl ethers are tolerated,[13] and have
been used in an attempt to promote chiral induction,[15] but
the subsequent removal of the t-butyl group requires use of the
aggressive reagent trimethylsilyl iodide.[16]
During investigations directed towards the synthesis of
morpholine type-containing natural products, we sought to
perform a (3þ2) cycloaddition using the NAY generated
from N-methylmorpholine N-oxide (NMO, 4). While the corre-
sponding non-oxygenated heterocycle, N-methylpiperidine
N-oxide (NPO, 5) has been used as a NAY precursor by Roussi
(e.g. Scheme 2b),[1,17] a survey of the literature revealed that
reactions involving morpholine-derived N-oxides lacked
precedent.
ϩ
2
Ϫ
1
3
We discovered that NMO (4) is a challenging substrate for
Roussi-type chemistry. The incorporationofthe oxygenatom into
Scheme 1. Azomethine ylid (3þ2) cycloadditions.
Journal compilation Ó CSIRO 2014
www.publish.csiro.au/journals/ajc

1310
P. M. Mirzayans, E. H. Krenske, and C. M. Williams
Ϫ
ϩ
Table 1. Reaction condition screening in the synthesis of azomethine
ylids arising from NMO
ϩ
Ϫ
Entry Addition rate Base
Equiv.
Temp.
Yield. Product
1
2 or 3
1
2
3
4
5
6
7
10 min
15 min
1 h
LDA
3.5
10
3.5
3.5
3.5
3
08C
08C
08C
08C
08C
34 %
27 %
10 %
8 %
7
7
7
7
6
6
6
LDA
LDA
Ϫ
ϩ
45 min
15 min
10 min
15 min
n-BuLi
MDA
MDA
MDA
(a)
18 %
ꢀ508C - r.t. 17 %
6
08C
18 %
Ϫ
(b)
(c)
ϩ
Ϫ
ϩ
5
13
12
Ϫ
ϩ
ϩ
Scheme 4. Synthesis of N-oxide 12, a possible decomposition product
of NMO.
ϩ
slightly decreased the yield of 7 to 27 % (Table 1, entry 2).
Conceptually, the formation of 7 could commence with azo-
methine ylid formation (i.e. 9) and (3þ2) cycloaddition giving 6.
Then, a sequence involving deprotonation of 6, C–O bond
cleavage giving 10, and proton transfer from C to O may furnish
11, which gives 7 upon protonation (Scheme 3).
Scheme 2. Roussi’s method for generation and (3þ2) cycloaddition of
NAYs.
Ϫ
In an attempt to minimise the possibility of rearrangement of
6 to 7, we performed a modified protocol in which the concen-
tration of LDA was kept low by addition via syringe pump to the
suspension of NMO (4) and diphenyl acetylene (THF, 08C) over
a period of 1 h (Table 1, entry 3). This protocol, however,
decreased the yield to 10 %. Again, no dimer 8 was isolated.
Using n-butyl lithium (n-BuLi) instead of LDA, and conducting
the addition over a period of 45 min did not improve upon these
results (Table 1, entry 4). At higher rates of addition, n-BuLi
added to the triple bond of diphenyl acetylene (a similar result
was also observed for t-butyl lithium, t-BuLi). During the
slow addition of LDA described above, a transient intermediate
was detected by TLC. It is possible that this intermediate was
cycloadduct 6, however, another possibility is that one of the
C–O bonds of NMO (4) cleaves before an azomethine ylid
is generated, and it is the ring-opened ylid that undergoes
cycloaddition. To probe the latter possibility, we subjected the
acyclic N-oxide 12 (prepared as in Scheme 4) to the same
conditions. The N-oxide 12 is a plausible ring-opened decom-
position product of NMO. It was synthesised from N-methyl-
ethanolamine and ethyl bromide via intermediate 13. When
N-oxide 12 was treated with LDA at ꢀ788C, only a trace amount
of pyrrole 7 was formed. This suggests that N-oxide decompo-
sition is not the primary origin of pyrrole 7 during the reaction of
NMO (4) with LDA.
Ϫ
Ϫ
ϩ
ϩ
Њ
4
9
6
10
11
ϩ
Ϫ
8
7
Scheme 3. Reaction of NMO (4) with LDA and diphenyl acetyleneleading
to pyrrole 7.
the six-membered ring leads to a complex array of side reactions,
which are sensitive to the nature of the base. We have used a
combination of experimental and computational methods to exa-
mine the products obtained under different conditions and their
mechanisms of formation. We herein report our studies of the
reactions of NMO (4) with strong bases and reveal the factors that
must be addressed in developing a protocol for azomethine ylid
generation from this class of heteroatom-containing N-oxides.
We studied the behaviour of NMO under increasingly basic
conditions. Treatment of a suspension of NMO (4) with t-BuLi
in THF at ꢀ728C, followed by slow warming to ꢀ308C over 4 h,
afforded a complex mixture. Surprisingly, the aqueous phase
contained some unreacted NMO (4), together with minor
impurities. Similar results were observed when the reaction
was quenched at ꢀ728C rather than being allowed to warm to
ꢀ308C. Unfortunately, reliable quantification of the various
species in the aqueous phase was not possible, due to the pre-
sence of inorganic salts. Further increasing the base strength
by employing Schlosser’s base (t-BuLi/t-BuOK)[18] at ꢀ728C
Results and Discussion
Our initial studies focussed on the behaviour of NMO (4) under
the conditions typically used by Roussi (LDA, THF)
(Scheme 3). In contrast to NPO (5) (Scheme 2b), treatment of a
suspension of NMO (4) with 3.5 equiv. of LDA in the presence
of diphenyl acetylene (THF, 08C) did not lead to isolation
of the expected cycloadduct 6. Instead, pyrrole 7 was obtained
in 34 % yield (Table 1, entry 1). Although the yield of 7 was
modest, no azomethine ylid dimer 8 was isolated as a side
product. Increasing the concentration of base by three-fold

Nonstabilised Azomethine Ylids from N-Oxides
1311
Ϫ
OϪ
Nϩ
Nϩ
O
temperature (r.t.) gave a slightly lower yield (Table 1, entry 6),
while increasing the quantity of MDA from 3 to 6 equiv. gave no
change in yield (Table 1, entry 7).
CHCl3
O
KX
N
r.t.
80 %
over 2 steps
14
OH
O
4
O
15
Additives such as 1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-
pyrimidinone (DMPU) and tetramethylethylenediamine
(TMEDA)[22] were benign, whereas magnesium bromide ethe-
rate complex substantially retarded product formation. Different
combinations in rates of addition, and order of addition of
reagents, gave no improvement in the yield. No reaction was
observed when MeMgBr was substituted for MDA. Finally,
when the reaction was conducted in the absence of diphenyl
acetylene, no dimer 8 was detected, but the Polonowski-type[23]
adduct 16[24] was isolated (,70 %). A possible mechanistic
route to 16 (Scheme 6) involves mono-deprotonation of NMO
and elimination of ꢀOMg(NHiPr2)(NiPr2) to give the iminium
cation 17, which is attacked by a second molecule of NMO (4) to
give 18. In the presence of excess base, 18 decomposes with
expulsion of formaldehyde to give iminium ion 17 and morpho-
line, which combine to form 16. Interestingly, the production of
a gas was observed during the experiment, which is presumably
formaldehyde.[25b] The decomposition of NMO (4) via the
pathway 4 - 17 - 18 - 17 is an autocatalytic process that
has previously been proposed by Rosenau and co-workers, who
also observed gas production due to formation of formalde-
hyde.[25] Having observed 16 as a substantial by-product
under the aforementioned conditions we believe that the low
yield of 6 observed in the reaction of NMO (4) and diphenyl
acetylene with MDA is due to the formation of 16. (Note:
Rosenau and Kosma have proposed that the conversion of 18
to 17, plus formaldehyde and morpholine, may be either
intermolecular (as drawn), or alternatively, intramolecular via
a 6-membered transition state (TS) where deprotonation is
X ϭ H and HMDS
Scheme 5. Reaction of NMO (4) with potassium bases leading to 14 and
subsequently 15.
Ϫ
Ϫ
ϩ
ϩ
8
4
9
6
Ϫ
ϩ
ϩ
ϩ
ϩ
18
17
4
Ϫ
ϩ
Ϫ
ϩ
16
17
Scheme 6. Reaction of NMO (4) with MDA and diphenyl acetylene
leading to cycloadduct 6 and side product 16.
afforded a complex mixture of compounds that suggested
decomposition rather than azomethine ylid formation.
effected by the nitrogen lone pair.[25]
)
Because the deprotonation of NMO (4) involves metal-
coordinated species, we turned our attention to the effect of
varying the metal cation. When potassium hexamethyldisilyla-
zide (KHMDS) was added to a suspension of NMO (4) and
diphenyl acetylene (THF, 08C), the reaction cleanly afforded the
hydroxylamine 14, which was stable enough to be isolated
(Scheme 5). In chloroform, 14 converted to 15 via a reverse
Cope elimination[19] on standing overnight (80 % yield over
two steps, 4 to 15). Use of potassium hydride (KH) required
elevated temperature (608C), but likewise afforded 14 (33 %).
The non-oxygenated heterocycle NPO 5 does not react under
these conditions.
Treatment of NMO (4) and diphenyl acetylene with the
sodium base NaHMDS gave no product. The NMR spec-
trum of the crude mixture revealed unreacted NMO (4) and
diphenyl acetylene. The same outcome was obtained when
sodium N,N-diisopropylamide (NDA)[20] was used as the base.
After the lack of success in generating the desired cycload-
duct 6 by using Liþ, Naþ, or Kþ bases, we investigated a
magnesium base. Pleasingly, treating a mixture of NMO (4)
and diphenyl acetylene (THF, 08C) with commercial magne-
sium bis(N,N-diisopropylamide)[21] (MDA) gave the expected
cycloadduct 6 in 18 % yield (Scheme 6) (Table 1, entry 5). With
6 at hand, we were able to obtain pyrrole 7 in 71 % yield (82 %
based on recovered starting material, brsm) by simply treating
6 with LDA at 08C. This result supports the intermediacy of 6 in
the synthesis of 7 under these conditions. Heartened by these
results, we systematically examined variations in reaction con-
ditions in an attempt to obtain a more favourable yield of 6.
Conducting the reaction at ꢀ508C and slowly warming to room
As obtention of different products upon treatment of NMO
(4) and alkyne with Liþ, Kþ, and Mg2þ bases was highly
unexpected we performed density functional theory (DFT)
calculations to explore the mechanistic origins of these pro-
ducts. The reactions of NMO (4) with diphenyl acetylene and
base in THF were computed at the SMD/B3LYP-D3/6–31G(d)
level of theory, modelling the base as MNMe2 (Mþ ¼ Liþ, Kþ,
or Mg(NMe2)þ). This simple approach to modelling bases
containing different metal cations does not make any allowance
for the effects of aggregation or solvent coordination, and
therefore will not give quantitative accuracy in the prediction
of absolute rates. Nevertheless, it is expected to give useful
qualitative insights into the way in which the identity of the
metal cation influences the barrier for a given process, and thus
the observed chemoselectivity.
The initial calculations (Fig. 1) examined the relative reac-
tivities of NMO (4) and NPO (5) towards azomethine ylid
formation under Roussi’s conditions (LDA in THF). Roussi
proposed[12] that the reaction of 5 with LDA commences with
deprotonation of the N-Me group (exo deprotonation, TS1),
which is kinetically more facile than deprotonation of one of the
methylene groups adjacent to nitrogen (endo deprotonation).
Exo deprotonation and loss of ꢀOLi(NHR2) lead to iminium
cation 19, which is deprotonated again to give azomethine ylid
20. Calculations employing LiNMe2 as the base (Fig. 1, dashed
line) indicate that the rate-limiting step of this pathway is the
cleavage of the N–O bond of the intermediate lithiated carban-
ion 21 (TS2). In agreement with Roussi’s proposal, exo depro-
tonation of NPO (5) is kinetically more favourable than endo
deprotonation. The endo TS is 1.5 kcal molꢀ1 higher in energy

1312
P. M. Mirzayans, E. H. Krenske, and C. M. Williams
Ϫ OLi(NHMe2)
ϩN
TS2
N–O
cleavage
8.7
Li(NMe2)
H
ϪO
Nϩ
ϩ
X
N
OϪ
Nϩ
X
7.5
Li(NMe )
ϩ
X
2
H
TS1
deprotonation
TS3
NMe2
Li
Ϫ
X
0
Li(NMe2)(OH)
deprotonation
X
ϭ
CH2
X ϭ O
5
4
Ϫ9.6
Ϫ9.6
Ϫ7.2
LiNMe2
Ϫ16.6
Ϫ14.2
Ϫ20.2
Ϫ19.3
[Li(NHMe2)]2O
Ϫ21.4
Ϫ25.7
Ϫ26.7
Li(NMe2)
ϪO
Nϩ
Li(NHMe2)
O
ϩ
Nϩ
N
Ϫ42.8
Ϫ42.9
Ϫ
X
X
X
ϩ
21
19
22
N
X
Ϫ
20
Fig. 1. Free energy diagrams for LiNMe2-mediated azomethine ylid formation from NMO (4) and NPO (5) in
THF, computed at the SMD/B3LYP-D3/6–31G(d) level of theory. Dashed line: NPO (5); solid line: NMO (4). DG
in kcal molꢀ1
.
than TS1 (see the Supplementary Material). Examining the
corresponding calculations for NMO (4) (Fig. 1, solid line),
the energy profile for azomethine ylid formation is very similar
to that for NPO (5). Indeed, the overall free energy barrier
(22-TS2) for NMO (4) is 2.2 kcal molꢀ1 smaller than that for
NPO (5). Thus, azomethine ylid formation from NMO (4) via
a Roussi-type mechanism appears to be more facile than the
corresponding reaction of NPO (5), and the low yields of cyc-
loadduct obtained using NMO (4) cannot be ascribed to any
inherent lack of reactivity of NMO towards ylid formation.
The calculations reveal that a crucial factor differentiating
the chemistry of NMO (4) from that of NPO (5) and other
aliphatic N-oxides is the regioselectivity of the initial deproto-
nation step. The availability of alternative deprotonation
routes for NMO leads to a complex set of reaction pathways
as depicted in Fig. 2. Whereas for NPO (5) deprotonation at NMe
cleavage (i.e. 26), deprotonation and 6p-electrocyclic ring
closure[26] of the resulting ylid 27 to intermediate 28,[27]
followed by base-catalysed tautomerisation. The acyclic
hydroxylamine 14 and five-membered N-oxide 15 are inter-
convertible by (reverse) Cope elimination, an intramolecular
process. The equilibrium positions of Cope eliminations are
known to depend on solvent, chloroform favouring cyclisa-
tion.[19f] N-Oxide 15 could be formed from the endo carbanion
23 through a sequence consisting of C–O bond cleavage to 29
followed by 5-exo-trig cyclisation (Path B3). Alternatively,
deprotonation of NMO (4) a to the endocyclic oxygen atom
(Path C) would give carbanion 30, which may undergo C–N
bond cleavage to give 14. The energies of the important inter-
mediates and transition states along these pathways coordinated
to different metal cations are shown in Fig. 2. Energies are
reported only for the transition states and intermediates preced-
ing the loss of metal oxide. Because the loss of metal oxide
provides a very large thermodynamic driving force, the subse-
quent steps are not rate-limiting. Details of the barriers of the
downstream steps are included in the Supplementary Material.
The energies of the transition states along Paths A and B1
(via 19 or 27, respectively) provide insights into the propensities
of different bases to yield cycloadduct 6. Starting from either the
exo carbanion 21 or the kinetically favoured endo carbanion 23,
the rate-determining N–O cleavage steps (TS2 and TS5) are low
in energy when the metal cation is Mg2þ (DGz E ꢀ2 kcal
molꢀ1), high in energy for Kþ ($21 kcal molꢀ1), and interme-
diate for Liþ (6–8 kcal molꢀ1). The Mg2þ cation promotes N–O
cleavage because it is a small cation with a high charge density,
which allows strong binding to the O2ꢀ that is being abstracted
from the carbanion. Conversely, the large monocation Kþ
provides the weakest stabilisation of the oxide. These results
are in agreement with the propensity towards cycloadduct
formation observed experimentally with different metal cations,
where an isolable quantity of 6 was obtained using MDA, no 6
was obtained using KH or KHMDS, and the use of LDA gave
pyrrole 7, a decomposition product of 6. Pyrrole 7 is computed to
be 21 kcal molꢀ1 lower in energy than cycloadduct 6, and hence
the equilibrium position of the base-catalysed interconversion of
(exo deprotonation) is kinetically favoured by 1.5 kcal molꢀ1
,
the preferred site of deprotonation of NMO (4) is not N-Me but
N-CH2, leading to the endocyclic carbanion 23. The endo
deprotonation TS (TS4) for NMO with LiNMe2 lies 2 kcal
molꢀ1 lower in energy than the exo TS (TS1). Similar endo
selectivities of ,2 kcal molꢀ1 are computed for the Kþ and
Mg2þ bases. The switch in the regioselectivity of deprotonation
on going from NPO (5) to NMO (4) is stereoelectronic in origin.
In NMO, the equatorial N-CH2 proton is antiperiplanar to the
C–O bond, allowing the endo deprotonation TS (and the result-
ing carbanion) to be stabilised by negative hyperconjugation
(see TS4Li, Fig. 2).
Each of the observed products 6, 7, 14, and 15 can
be accessed via multiple mechanistic pathways. In principle,
cycloadduct 6 could be formed by the sequence of N-oxide
deprotonation, N–O cleavage, and iminium ion deprotonation
following either an exo or endo mode (Path A or Path B1,
respectively, in Fig. 2) followed by 1,3-dipolar cycloaddition.
Pyrrole 7 may (as proposed above) be derived from cycloadduct
6 by the action of base (Path A or Path B1), or it may
alternatively arise via Path B2, which consists of nucleophilic
addition by the endo carbanion 23 to diphenyl acetylene (i.e. 24),
followed by intramolecular proton transfer (i.e. 25), N–O

Nonstabilised Azomethine Ylids from N-Oxides
1313
1.28
TS2
N–O cleavage
1.50
TS1Li
TS1
O
M(NHMe2)
ϩ
exo deprotonation
ϩ N
N
TS3
deprotonation
7.5
(24.6)
[Ϫ1.6]
Ϫ
O
O
19
Ϫ21.4
(Ϫ5.7)
[Ϫ26.1]
21
M(NMe2)
ϪO
Nϩ
H
Ph
Ph
Liϩ Ϫ9.6
ϩ
ϩ
ϪHϩ
O
(Kϩ Ϫ3.0)
PhCCPh
TS5
N
N
N
[Mg2ϩ Ϫ11.2]
N–O
O
O
20Ϫ
cleavage
6
27
O
6.5
(21.1)
[Ϫ1.6]
Liϩ Ϫ42.8
[Mg2ϩ Ϫ64.5]
base
Ph
ϪO
Nϩ
TS4
O
(Me2HN)M
Ph
Ph
Nϩ
Ph
endo deprotonation
OH
OϪ
OH
Hϩ
O
O
Ϫ
Ϫ
N
N
N
Ph
Ph
ϪO
23
4
M(NMe2)
7
Nϩ
10
11
Ϫ22.5
(Ϫ8.4)
[Ϫ23.5]
Ϫ11.6
H
O
(Ϫ5.1)
(Me2HN)
M(NHMe2)
M
O
Nϩ
[Ϫ13.4]
TS10
Hϩ
transfer
O
Nϩ
11
10
Ϫ
TS6
TS8
deprotonation
Me2NH
O
O
O
PhCCPh
Ϫ21.2
H
TS11
N–O
cleavage
M
O
Nϩ
α to O
Ϫ
addition
Ϫ7.8
Ϫ
Ph
Ph
M
Ph
Me2N
Ph
24
25
(1.8)
O
OϪ
Nϩ
[Ϫ3.5]
1.28
1.51
TS4Li
Ϫ18.2
Ϫ49.7
Ϫ41.8
H
30
TS9
C–N
cleavage,
Ϫ7.3
(5.1)
[Ϫ10.6]
then Hϩ
base
Ϫ6.1
6π
electro-
ϩ
ϩ
(2.5)
O
N
Ϫ
N
N
[n.d]
base
cyc.
O
O
O
H
H
H
N
Ph
Ph
Ph
Ph
Ph
14
Ph
OH
26
27
28
TS7 Ϫ2.9
(3.7)
C–O
cleavage
(reverse) Cope
elimination
[5.3]
1.30
1.48
OϪ
Nϩ
(NHMe2)
5-exo-trig,
M
Ϫ O
Hϩ
O
Nϩ
O
15
29
TS8K
2.30
TS6Li
2.33
1.93
1.92
1.84
TS7K
TS9K
TS2Mg
TS5Mg
Fig. 2. Pathways for deprotonation of NMO (4) and conversion of the three regioisomeric carbanions (21, 23, and 30) to the observed products (6, 7, 14, and
15). Free energies (kcal molꢀ1) of selected species complexed to different metal cations are shown. Liþ numbers shown in normal type; Kþ numbers shown in
parentheses; Mg2þ numbers shown in square brackets.
6 and 7 lies strongly in favour of 7. That 6 is decomposed by
excess LDA, but not by excess MDA, may reflect differences in
steric effects that may hamper the deprotonation of 6 to give 10
when the base is MDA.
Alternatively, pyrrole 7 can be accessed through Path B2. It is
noteworthy that the initial step of this pathway (addition of endo
carbanion 23 to diphenyl acetylene giving 24) is most facile for
Liþ (TS6). Indeed, for Liþ, nucleophilic addition to the alkyne
via TS6 is favoured by 14.3 kcal molꢀ1 over N–O cleavage via
TS5. With Mg2þ, nucleophilic addition to the alkyne via TS6 is
favoured also, but only by 2 kcal molꢀ1. Given the simple model
species used, and the fact that the rate of passage through TS6
(but not TS5) is dependent on alkyne concentration, it is unlikely
that such a small difference in barriers would be a reliable

1314
P. M. Mirzayans, E. H. Krenske, and C. M. Williams
predictor of differences in reaction rates. The main conclusions
that can be drawn from the calculated data are that Liþ promotes
reaction through Path B2, and Mg2þ provides strong stabilisa-
tion to the N–O cleavage transition states involved in ylid
formation via Paths A and B1.
a different outcome – a highly unusual result. Based on our
studies, MDA has emerged as the base of choice for the for-
mation of 6, avoiding an otherwise elimination–aromatisation
sequence affording pyrrole 7 under LDA conditions. This
finding is likely to hold true in other morpholine containing
heterocycles generated under these conditions.
Concerning products 14 and 15, which have not incorporated
any alkyne and are obtained only with potassium bases, the
calculations indicate that when Mþ ¼ Kþ, the preferred reaction
of the thermodynamic endo carbanion 23 is Path B2. However,
Path B3 leading to 14 and 15 via TS7 lies only 2 kcal molꢀ1
higher in energy, again within the error expected from these
simple models. Transition state TS7 (C–O cleavage) is also
predicted to be relatively low in energy when the metal ion is Liþ
(ꢀ2.9 kcal molꢀ1), but no formation of 14 or 15 is observed
using Liþ bases since Path B2 leading to pyrrole 7 is preferred.
In the alternative mechanism leading to 14 and/or 15
(Path C), the initial TS (TS8, deprotonation a to oxygen) has
DGz ¼ 2.5 kcal molꢀ1 for Mþ ¼ Kþ, which is $5 kcal molꢀ1
higher than the TSs for exo or endo deprotonation of NMO
(TS1, TS4). If the deprotonation of NMO is kinetically con-
trolled, then the calculations predict that the route to 14 and/or
15 is Path B3. However, the overall rate-determining transition
state in Path C (TS9, C–N cleavage in carbanion 30) with
Mþ ¼ Kþ lies at 5.1 kcal molꢀ1, only 1.4 kcal molꢀ1 higher than
the rate-determining TS in Path B3 (TS7, C–O cleavage in
carbanion 23). Taking into account the approximations inherent
in the calculations (see above), the small difference between
these barriers suggests that under conditions of thermodynamic
control, both Path B3 and Path C are feasible for Kþ bases.
The possibility that the reactions of NMO (4) with bases may
follow radical mechanisms was explored by conducting a
reaction in the presence of (2,2,6,6-tetramethylpiperidin-1-yl)
oxidanyl (TEMPO), in an attempt to trap any radical intermedi-
ates. No TEMPO adducts were detected when the reaction of
NMO (4) with diphenyl acetylene was run in the presence of
TEMPO, and the reaction outcome remained unaltered from that
obtained in the absence of TEMPO.[28]
Lastly, we considered the possibility that the differences in
observed products obtained from NMO (4) and NPO (5) may
result from differences in the reactive conformation of the
N-oxide. It is well known that cyclic N-oxides can exist as an
equilibrium mixture of conformers, and in the case of NPO (5)
the O-axial conformer has been determined to be the major
conformer (95 %).[29c] The 1H NMR spectrum of NMO (4)
showed only a single set of signals (at 258C), confirming the
results reported by Pohl et al. that the axial is the major
conformer.[29c] Calculations indicate that the O-axial conformer
of NMO (4) is favoured by 2.8 kcal molꢀ1, which is 0.7 kcal
molꢀ1 greater than the O-axial preference of NPO (5) (2.1 kcal
molꢀ1) (see also Pohl et al.[29c]). Additionally, the O-axial
conformers of both N-oxides are computed to undergo depro-
tonation more readily than the O-equatorial conformers. These
results suggest that Roussi reactions involving NMO (4) and
NPO (5) both involve the O-axial conformers.
General Experimental
1H and 13C NMR spectra were recorded on Bruker AV300
(300.13, 75.47 MHz) and DRX500 (500.13, 125.76 MHz)
instruments in CDCl3 unless otherwise stated. Coupling con-
stants are given in Hz and chemical shifts are expressed as
d values in ppm. High resolution electrospray ionisation
(HRESIMS) accurate mass measurements were recorded in
positive mode on a Bruker MicrOTOF-Q (quadrupole–time of
flight) instrument with a Bruker ESI source using sodium for-
mate as a reference calibrant. Column chromatography was
undertaken on silica gel (flash silica gel, 230–400 mesh) with
distilled solvents. All reagents and solvents were purified before
use according to literature methods.[30] NMO (4) was dried at
458C, 0.1 mmHg, for a period of 2 h before use. THF was freshly
distilled from a sodium/benzophenone still. Fine chemicals
were purchased from Sigma-Aldrich.
Computational Methods
DFT calculations were performed using the Gaussian 09 soft-
ware.[31] Geometries wereoptimisedatthe B3LYP/6–31G(d)level
of theory in the gas phase.[32] Stationary points were characterised
by vibrational frequency calculations at this level, and transition
states were further verified by means of IRC calculations.[33]
Single-point calculations at the B3LYP/6–31G(d) level using the
SMD implicit solvent model were performed to obtain solvation
energies in THF.[34] Dispersion energy corrections were computed
for the gas-phase geometries using Grimme’s D3 model[35] with
Becke–Johnson damping.[36] Free energies in THF solution (258C)
were obtained by adding the SMD solvation energy and D3 dis-
persion energy to the B3LYP/6–31G(d) gas-phase free energy and
were corrected to a standard state of 1 mol Lꢀ1
.
Experimental
1-N-(2-Hydroxyethyl)-2-methyl-3,
4-diphenyl-1H-pyrrole (7)
A mixture of NMO (4) (235 mg, 2.01 mmol) and diphenyl
acetylene (357 mg, 2.01 mmol) in anhydrous THF (10 mL)
under an argon atmosphere was treated with freshly prepared
LDA (12.5 mL, 0.56 M in THF) at 08C. The reaction was stirred
at this temperature for a period of 4 h and a further 17 h at r.t.
after which time the reaction was quenched by the addition of a
minimal amount of NH4Cl(aq). The solvent was removed under
vacuum using a liquid nitrogen trap to afford a crude mixture.
Purification by silica column chromatography (light petroleum :
EtOAc, 2 : 1) afforded 7 (16.3 mg, 34 %) as a pale yellow oil.
dH (500 MHz, CDCl3) 7.33–7.28 (m, 2H), 7.25–7.09 (m, 8H),
6.84 (s, 1H), 4.07 (t, J 5.5, 2H), 3.94 (t, J 5.5, 2 H), 2.26 (s, 3 H).
dC (125 MHz, CDCl3) 136.3, 135.7, 130.5, 128.0 (2C), 127.9,
126.9, 125.6, 125.2, 123.4, 120.8, 118.5, 62.5, 49.1, 10.5. HRMS
Anal. Calc. for C19H20NO: 278.1539; found: 278.1541.
Conclusion
The Roussi reaction is one of only a handful of procedures that
allow NAYs to be generated at low temperature. The study
reported herein describes a detailed investigation into the
complex array of reactions undergone by NMO (4) under
Roussi’s conditions. We have found that the base counterion
(i.e. metal cation) has a critical role in directing the course of
reaction and that Liþ, Naþ, Kþ, and Mg2þ cations each produce
2-(N-Ethyl-N-methylamino)ethan-1-ol (13)
To neat N-methyl-ethanolamine (0.86 mL, 10.65 mmol) at 08C
was added ethylbromide (0.80 mL, 10.65 mmol) dropwise.
After 20 min at 08C the ice bath was removed. Upon warming an

Nonstabilised Azomethine Ylids from N-Oxides
1315
exotherm was observed. When no further heat was generated
(30 min) the reaction was cooled to 08C and a further portion of
ethylbromide (0.20 mL, 0.25 equiv.) was added and the ice bath
was removed. After stirring at r.t. for a period of a 100 min gentle
vacuum was applied to remove volatiles and the crude was
washed with acetone. After applying gentle vacuum to remove
the remaining acetone, excess solid pellets of NaOH was added
and the mixture was stirred for a period of 30 min. The crude
residue was placed under high vacuum (,0.1 Torr) and the
product was captured using a liquid nitrogen trap. When most of
the product was captured, the crude was slowly heated to 608C
for a brief period to afford the tertiary amine 13 (527 mg, 48 %)
as a colourless liquid. dH (300 MHz, CDCl3) 3.58 (dd, J 5.5, 2H),
2.55–2.44 (m, 4H), 2.24 (s, 3H), 1.06 (t, J 7.1, 3H). This material
is commercially available.
10 min (effervescence was observed over the course of few min
and the reaction became homogeneous and yellow in colour).
The reaction was allowed to slowly reach r.t. and was subse-
quently quenched (saturated aqueous solution of NH4Cl) 3 h
after addition of MDA. The aqueous phase was extracted using
EtOAc, and the combined organic phase was washed with brine
and dried over MgSO4. The crude mixture was purified by silica
column chromatography (EtOAc) to afford 6 (109 mg, 18 %) as
a colourless oil. dH (500 MHz, CDCl3) 7.31–7.14 (m, 10H), 4.29
(dd, J 12.8, 3.7, 1H), 4.11 (dq, J 9.6, 3.6, 1H), 3.88 (dd, J 12.7, 3.4,
1H), 3.85 (dd, J 11.0, 3.6, 1H), 3.78–3.72 (m,2H), 3.52(dd, J 11.2,
9.6, 1H), 3.08 (ddd, J 12.7, 8.7, 5.7, 1H), 2.97 (dt, J 12.7, 2.7, 1H).
dC (125 MHz, CDCl3) 136.9, 136.1, 135.0, 134.9, 128.6, 128.3,
128.2, 127.8, 127.4, 127.3, 68.3, 68.1, 64.3, 60.4, 48.6. HRMS
Anal. Calc. for C19H20NO: 278.1539; found: 278.1532.
N-Ethyl-2-hydroxy-N-methylethan-1-amine N-oxide (12)
1-N-(2-Hydroxyethyl)-2-methyl-3,4-diphenyl-1H-pyrrole
(7) from LDA Treatment of 6
To neat compound 13 (322 mg, 3.12 mmol) was added H2O2
(0.41 mL, 30 % aqueous solution) and the resulting mixture was
stirred overnight (20 h). Toluene was then added with vigorous
stirring, and after ,5 min the solvent was removed under high
vacuum to afford the title compound 12 (314 mg, 84 %) as a
viscous clear oil. dH (500 MHz, [D4]methanol) 4.07–3.96
(m, 2H), 3.52–3.39 (m, 4H), 3.18 (s, 3H), 1.36 (t, J 7.2, 3H).
dC (125 MHz, [D4]methanol) 70.1, 65.7, 57.2, 55.3, 8.8. HRMS
Anal. Calc. for C5H14NO2: 120.1019; found: 120.1019.
To a solution of 6 (90 mg, 0.324 mmol) in anhydrous THF
(9 mL) at 08C under an argon atmosphere, was added freshly
prepared LDA (3.0 mL, 1.14 mmol, 0.37 M THF solution) in
three portions at 30 min intervals. The reaction was quenched
15 min after the last addition using a saturated aqueous solution
of NH4Cl. The aqueous phase was extracted using EtOAc, and
the combined organic phase was dried over MgSO4. Purification
by silica column chromatography (EtOAc) afforded 7 (64 mg,
71 %, 82 % brsm) and recovered starting material 6 (12 mg).
N-Methyl-N-(2-(vinyloxy)ethyl)hydroxylamine (14)
and 2,3-dimethyloxazolidine 3-N-oxide (15)
Dimorpholinomethane (16)
A suspension of NMO (4) (137 mg, 1.17 mmol) and diphenyl
acetylene (208 mg, 1.17 mmol) in anhydrous THF (10 mL) was
treatedwithKHMDS(9.36 mL, 0.5 M intoluene)at 08Cunderan
argon atmosphere. The reaction was stirred at this temperature
for a period of 1 h and a further 24 h at r.t. after which time the
reaction was quenched using a saturated aqueous solution of
NH4Cl. The aqueous phase was extracted using EtOAc, dried
over MgSO4, and concentrated under vacuum to afford
hydroxylamine 14 as a crude oil. dH (500 MHz, [D6]benzene)
6.47 (dd, J 14.3, 6.6, 1H, 14), 4.25 (dd, J 14.3, 1.8, 1H, 14),
4.18–4.12(m, 1H, 15), 3.97(dd, J 6.6, 1.8, 1H, 14), 3.83(brs, 2H,
14), 3.71 (q, J 5.5, 1H, 15), 3.43 (ddd, J 10.5, 7.2, 3.3, 1H, 15),
3.23 (dt, J 8.1, 3.3, 1H, 15), 2.76 (br t, J 5.5, 2H, 14), 2.53 (br s,
3H, 14), 2.45 (s, 3H, 15), 2.46–2.39 (m, 1H, 15), 1.32 (d, J 5.5,
3H, 15). dC (125 MHz, [D6]benzene) 152.6 (14), 100.1 (15), 86.4
(14), 67.0 (15), 66.5 (14), 64.2 (15), 61.6 (14), 52.6 (14), 49.7
(15), 13.0(15). Compound 14 converts to 15 in chloroform-d and
[D6]benzene; thus, only a mixture could be observed by NMR.
The crude oil was dissolved in CHCl3 (used as obtained from
supplier) and stirred for a period of 20 h at r.t. after which time
the solvent was removed under vacuum and the crude material
washed with light petroleum to afford oxazolidine N-oxide 15
(110mg, 80%) as a pale solid. dH (500MHz, CDCl3) 4.64 (q, J
5.5, 1H), 4.46 (dt, J 8.6, 7.3, 1H), 4.06 (dt, J 8.6, 4.0, 1H), 3.79
(ddd, J10.9, 7.2, 3.9, 1H), 3.71–3.63(m, 1H), 3.20(s, 3H), 1.54(d,
J 5.5, 3H). dC (125 MHz, CDCl3) 100.4, 67.6, 63.9, 53.3, 12.6.
HRMS Anal. Calc. for C5H12NO2: 118.0863; found: 118.0858.
To a suspension of NMO (4) (412 mg, 3.51 mmol) in anhydrous
THF (20 mL) under an argon atmosphere, was added MDA
(7.54 mL, 5.27 mmol, 0.7 M THF solution) dropwise over a
period of 15 min. The reaction was quenched 1 h after the
addition using a saturated aqueous solution of NH4Cl. The
aqueous phase was extracted using EtOAc and the combined
organic phase was dried over MgSO4 to afford 16 in a crude
yield of ,70 %. The NMR spectra were in accordance with
those reported previously.[24]
Supplementary Material
Copies of 1H and 13C NMR spectra, further discussion of
computational studies of mechanism, and a compilation of
computational data can be found in the Supplementary Material.
Acknowledgements
We thank the Australian Research Council and the University of Queensland
for financial support. C.M.W. and E.H.K. are ARC Future Fellows supported
by grants FT110100851 and FT120100632, respectively. Computational
resources were provided by the National Facility of the National Compu-
tational Infrastructure through the Merit Allocation Scheme, and by the
University of Queensland Research Computing Centre.
References
[1] (a) I. Coldham, R. Hufton, Chem. Rev. 2005, 105, 2765. doi:10.1021/
CR040004C
(b) W. H. Pearson, P. Stoy, Synlett 2003, 903. doi:10.1055/
S-2003-39285
(c) E. Vedejs, F. G. West, Chem. Rev. 1986, 86, 941. doi:10.1021/
CR00075A014
7,8-Diphenyl-3,4,6,8a-tetrahydro-1H-pyrrolo[2,1-c]
[1,4]oxazine (6)
To a suspension of NMO (4) (252 mg, 2.15 mmol) and diphenyl
acetylene (383 mg, 2.15 mmol) in anhydrous THF (14 mL) at
08C under an argon atmosphere, was added MDA (9.22 mL,
6.45 mmol, 0.7 M THF solution) dropwise over a period of
(d) K. V. Gothelf, K. A. Jorgensen, Chem. Rev. 1998, 98, 863.
doi:10.1021/CR970324E
(e) C. Najera, J. M. Sansano, Curr. Org. Chem. 2003, 7, 1105.
doi:10.2174/1385272033486594

1316
P. M. Mirzayans, E. H. Krenske, and C. M. Williams
(f) L. M. Harwood, R. J. Vickers, in Synthetic Applications of
1,3-Dipolar Cycloaddition Chemistry Toward Heterocycles and
Natural Products (Eds. A. Padwa, W. H. Pearson) 2003, Vol. 59, Ch. 3,
pp. 169–252.
(b) C. H. DePuy, R. W. King, Chem. Rev. 1960, 60, 431. doi:10.1021/
CR60207A001
(c) H. O. House, D. T. Manning, D. G. Melillo, L. F. Lee, O. R. Haynes,
B. E. Wilkes, J. Org. Chem. 1976, 41, 855. doi:10.1021/JO00867A023
(d) D. St. C. Black, J. E. Doyle, Aust. J. Chem. 1978, 31, 2317.
doi:10.1071/CH9782317
(e) E. Ciganek, J. Org. Chem. 1990, 55, 3007. doi:10.1021/
JO00297A011
(f) E. Ciganek, J. M. Read, Jr, J. C. Calabrese, J. Org. Chem. 1995, 60,
5795. doi:10.1021/JO00123A013
[2] (a) See for example: I. Coldham, S. Jana, L. Watson, N. G. Martin,
Org. Biomol. Chem. 2009, 7, 1674. doi:10.1039/B822743H
(b) N. Shimada, Y. Abe, S. Yokoshima, T. Fukuyama, Angew. Chem.
Int. Ed. 2012, 51, 11824. doi:10.1002/ANIE.201206863
(c) A. J. M. Burrell, I. Coldham, L. Watson, N. Oram, C. D. Pilgram,
N. G. Martin, J. Org. Chem. 2009, 74, 2290. doi:10.1021/JO8019913
(d) A. J. M. Burrell, I. Coldham, N. Oram, Org. Lett. 2009, 11, 1515.
doi:10.1021/OL9001653
(g) W. Oppolzer, A. C. Spivey, C. G. Bochet, J. Am. Chem. Soc. 1994,
116, 3139. doi:10.1021/JA00086A060
(e) A. J. M. Burrell, I. Coldham, Curr Org Synth 2010, 7, 312.
doi:10.2174/157017910791414472
(h) N. J. Cooper, D. W. Knight, Tetrahedron 2004, 60, 243.
doi:10.1016/J.TET.2003.10.043
(f) C. Fang, C. S. Shanahan, D. H. Paull, S. F. Martin, Angew. Chem.
Int. Ed. 2012, 51, 10596. doi:10.1002/ANIE.201205274
[3] (a) See for example: C. Me´nard-Moyon, N. Izard, E. Doris,
C. Mioskowski, J. Am. Chem. Soc. 2006, 128, 6552. doi:10.1021/
JA060802F
(b) V. Georgakilas, K. Kordatos, M. Prato, D. M. Guldi, M. Holzinger,
A. Hirsch, J. Am. Chem. Soc. 2002, 124, 760. doi:10.1021/JA016954M
[4] See for example: S. Pandiancherri, D. W. Lupton, Tetrahedron
Lett. 2011, 52, 671 and references therein. doi:10.1016/J.TETLET.
2010.11.142
(i) E. H. Krenske, E. C. Davison, I. T. Forbes, J. A. Warner,
A. L. Smith, A. B. Holmes, K. N. Houk, J. Am. Chem. Soc. 2012, 134,
2434. doi:10.1021/JA211568K
[20] (a) For this work, sodium N,N-diisopropylamide (NDA) was generated
in situ from n-butyl lithium, N,N-diisopropylamine and sodium
t-butoxide in THF using a similar procedure to that described by
L. Lochmann, J. Trekoval, J. Organomet. Chem. 1979, 179, 123.
doi:10.1016/S0022-328X(00)95213-6
(b) For isolation of NDA see: T. Munguia, Z. A. Bakir, F. Cervantes-
Lee, A. Metta-Magana, K. H. Pannell, Organometallics 2009, 28,
5777. doi:10.1021/OM9006032
[5] See for example: R. J. Carra, M. T. Epperson, D. Y. Gin, Tetrahedron
2008, 64, 3629. doi:10.1016/J.TET.2008.02.008
[21] P. E. Eaton, C.-H. Lee, Y. Xiong, J. Am. Chem. Soc. 1989, 111, 8016.
doi:10.1021/JA00202A054
[22] (a) D. B. Collum, Acc. Chem. Res. 1992, 25, 448. doi:10.1021/
AR00022A003
[6] (a) See for example: P. Deprez, J. Royer, H.-P. Husson, Tetrahedron:
Asymmetry 1991, 2, 1189. doi:10.1016/S0957-4166(00)80019-1
(b) P. Deprez, J. Rouden, A. Chiaroni, C. Riche, J. Royer,
H.-P. Husson, Tetrahedron Lett. 1991, 32, 7531. doi:10.1016/
0040-4039(91)80526-C
(b) R. E. Mulvey, S. D. Robertson, Angew. Chem. Int. Ed. 2013, 52,
11470. doi:10.1002/ANIE.201301837
[7] (a) R. Beugelmans, G. Negron, G. Roussi, J. Chem. Soc., Chem.
[23] D. Grierson, Org. React. 1990, 39, 85.
Commun. 1983, 31. doi:10.1039/C39830000031
(b) G. Roussi, Trends Org. Chem. 1993, 4, 101.
[8] (a) For additional examples, see: S. Takano, Y. Sugihara,
K. Ogasawara, Heterocycles 1992, 34, 1519. doi:10.3987/
COM-92-6107
[24] A. Porzelle, C. M. Williams, Synthesis 2006, 3025.
[25] (a) T. Rosenau, A. Potthast, P. Kosma, C.-L. Chen, J. S. Gratzl, J. Org.
Chem. 1999, 64, 2166. doi:10.1021/JO982350Y
(b) T. Rosenau, A. Potthast, P. Kosma, Tetrahedron 2002, 58, 9809.
doi:10.1016/S0040-4020(02)01291-7
(b) J. E. Davoren, D. L. Gray, A. R. Harris, D. M. Nason, W. Xu, Synlett
2010, 2490. doi:10.1055/S-0030-1258026
(c) B. De, J. F. DeBernardis, R. Prasad, Synth. Commun. 1988, 18, 481.
doi:10.1080/00397918808060740
(d) G. Negro´n, A. Fuentes, M. Romero, G. Madrid, R. Cruz, Rev. Soc.
Qu´ım. Me´x. 1999, 43, 123.
(c) For further studies on the degradation of NMO, see: T. Rosenau,
A. Potthast, P. Kosma, Tetrahedron 2004, 60, 301. doi:10.1016/
J.TET.2003.11.015
(d) F. Liebner, P. Schmid, C. Adelwo¨hrer, T. Rosenau, Tetrahedron
2007, 63, 11817. doi:10.1016/J.TET.2007.09.055
(e) Furthermore, see: T. Rosenau, A. Potthast, H. Sixta, P. Kosma,
Prog. Polym. Sci. 2001, 26, 1763. doi:10.1016/S0079-6700(01)
00023-5
[9] (a) D. Bernier, U. K. Efelscheid, S. Woodward, Org. Prep. Proced. Int.
2009, 41, 173. doi:10.1080/00304940902955756
(b) A. Albini, Synthesis 1993, 263. doi:10.1055/S-1993-25843
[10] G. Roussi, Heterocycles 1990, 31, 1445. doi:10.3987/COM-90-5321
[11] R. Beugelmans, L. Bemadjilal-Iguestira, J. Chastanet, G. Negron,
G. Roussi, Can. J. Chem. 1985, 63, 725. doi:10.1139/V85-120
[12] J. Chastanet, G. Roussi, J. Org. Chem. 1988, 53, 3808. doi:10.1021/
JO00251A026
[26] (a) E. Taylor, I. J. Turchi, Chem. Rev. 1979, 79, 181. doi:10.1021/
CR60318A003
(b) R. Huisgen, Angew. Chem. Int. Ed. Engl. 1980, 19, 947 [Angew.
Chem. 1980, 92, 979]. doi:10.1002/ANIE.198009473
(c) T. M. V. D. Pinho e Melo, Eur. J. Org. Chem. 2006, 2873.
doi:10.1002/EJOC.200500892
[13] G. Negron, G. Roussi, J. Zhang, Heterocycles 1992, 34, 293.
doi:10.3987/COM-91-5900
[14] G. Roussi, J. Zhang, Tetrahedron 1991, 47, 5161. doi:10.1016/
S0040-4020(01)87128-3
[15] (a) J. Chastanet, H. Fathallah, G. Negron, G. Roussi, Heterocycles
[27] An alternative pathway, calculated to be low in energy, involves ring
closure by intramolecular alkylation of iminium cation 26.
[28] Using the radical trapping agent g-tocopherol, Kosma and co-
workers identified the radical intermediates derived from reduction
of NMO (4) with FeII, MnII, or CoII salts. Radical pathways were
shown to lead to decomposition of NMO (4) to N-methyl morpholine,
morpholine, and formaldehyde. See: T. Rosenau, A. Potthast, H. Sixta,
P. Kosma, Tetrahedron 2002, 58, 3073. doi:10.1016/S0040-4020(02)
00163-1
1992, 34, 1565. doi:10.3987/COM-92-6052
(b) J. Chastanet, G. Roussi, G. Negro´n, Carbohydr. Res. 1995, 268,
301. doi:10.1016/0008-6215(94)00331-9
[16] G. Roussi, J. Zhang, Tetrahedron Lett. 1991, 32, 1443. doi:10.1016/
0040-4039(91)80353-8
[17] (a) J. Chastanet, G. Roussi, J. Org. Chem. 1985, 50, 2910. doi:10.1021/
JO00216A020
[29] (a) A. Katritzky, J. Chem. Soc. B 1971, 1330.
(b) Y. Shvo, E. D. Kaufman, J. Org. Chem. 1981, 46, 2148 and
references therein. doi:10.1021/JO00323A035
(b) J. Chastanet, G. Roussi, Heterocycles 1985, 23, 653. doi:10.3987/
R-1985-03-0653
[18] M. Schlosser, Pure Appl. Chem. 1988, 60, 1627. doi:10.1351/
PAC198860111627
(c) R. Pohl, M. Dracˇ´ınsky´, L. Slaveˇt´ınska´, M. Budeˇsˇ´ınsky´, Magn.
Reson. Chem. 2011, 49, 320. doi:10.1002/MRC.2750
[30] D. D. Perrin, W. L. F. Armarego, Purification of Laboratory Chemi-
cals, 3rd edn 1988 (Pergamon Press: Oxford).
[19] (a) A. C. Cope, N. A. Lebel, J. Am. Chem. Soc. 1960, 82, 4656.
doi:10.1021/JA01502A052
[31] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci,

Nonstabilised Azomethine Ylids from N-Oxides
1317
G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian,
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada,
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima,
Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr,
J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers,
K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand,
K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi,
M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross,
V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann,
O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski,
R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth,
P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas,
J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian 09,
Revision A.02 2009 (Gaussian, Inc.: Wallingford, CT).
(b) A. D. Becke, J. Chem. Phys. 1993, 98, 1372. doi:10.1063/
1.464304
(c) A. D. Becke, J. Chem. Phys. 1993, 98, 5648. doi:10.1063/
1.464913
(d) P. J. Stephens, F. J. Devlin, C. F. Chabalowski, M. J. Frisch, J. Phys.
Chem. 1994, 98, 11623. doi:10.1021/J100096A001
[33] (a) C. Gonzalez, H. B. Schlegel, J. Chem. Phys. 1989, 90, 2154.
doi:10.1063/1.456010
(b) C. Gonzalez, H. B. Schlegel, J. Phys. Chem. 1990, 94, 5523.
doi:10.1021/J100377A021
[34] A. V. Marenich, C. J. Cramer, D. G. Truhlar, J. Phys. Chem. B 2009,
113, 6378. doi:10.1021/JP810292N
[35] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, J. Chem. Phys. 2010, 132,
154104. doi:10.1063/1.3382344
[32] (a) C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785.
doi:10.1103/PHYSREVB.37.785
[36] S. Grimme, S. Ehrlich, L. Goerigk, J. Comput. Chem. 2011, 32, 1456.
doi:10.1002/JCC.21759