Stoichiometric reactions of enamines derived from diphenylprolinol silyl ethers with nitro olefins and lessons for the corresponding organocatalytic conversions - A survey

Article abstract of DOI:10.1002/hlca.201300079

The stoichiometric reactions of enamines prepared from aldehydes and diphenyl-prolinol silyl ethers (intermediates of numerous organocatalytic processes) with nitro olefins have been investigated. As reported in the last century for simple achiral and chiral enamines, the products are cyclobutanes (4 with monosubstituted nitro-ethenes), dihydro-oxazine N-oxide derivatives (5 with disubstituted nitro-ethenes), and nitro enamines derived from γ-nitro aldehydes (6, often formed after longer reaction times). The same types of products were shown to be formed, when the reactions were carried out with peptides H-Pro-Pro-Xaa-OMe that lack an acidic H-atom. Functionalized components such as alkoxy enamines, nitro-acrylates, acetamido-nitro-ethylene, or hydroxylated nitro olefins also form products carrying the diphenyl-prolinol silyl ether as a substituent. All of these products must be considered intermediates in the corresponding catalytic reactions; the investigation of their chemical properties provided useful hints about the rates, the conditions, the catalyst resting states or irreversible traps, and/or the limitations of the corresponding organocatalytic processes. High-level DFT and MP2 computations of the structures of alkoxy enamines and thermodynamic data of a cyclobutane dissociation are also described. Some results obtained with the stoichiometrically prepared intermediates are not compatible with previous mechanistic proposals and assumptions. Copyright

Full text of DOI:10.1002/hlca.201300079

Helvetica Chimica Acta – Vol. 96 (2013)
799
Stoichiometric Reactions of Enamines Derived from Diphenylprolinol Silyl
Ethers with Nitro Olefins and Lessons for the Corresponding Organocatalytic
1
Conversions – a Survey )
2a
2b
by Dieter Seebach*, Xiaoyu Sun ), Marc-Olivier Ebert, W. Bernd Schweizer, Nirupam Purkayastha ),
Albert K. Beck, Jçrg Duschmal e´ , and Helma Wennemers
Laboratorium fꢀr Organische Chemie, Departement Chemie und Angewandte Biowissenschaften,
ETH-Zꢀrich, Hçnggerberg HCI, Wolfgang-Pauli-Strasse 10, CH-8093 Zꢀrich
(
phone: þ 41-44-632-2990; fax: þ 41-44-632-1144; e-mail: seebach@org.chem.ethz.ch)
2c
2d
and Takasuke Mukaiyama ), Meryem Benohoud ), and Yujiro Hayashi*
Tokyo University of Science, Department of Industrial Chemistry, Faculty of Engineering, Kagurazaka,
Shinjuku-ku, Tokyo 162-8601, Japan
and Department of Chemistry, Tohoku University, 6-3 Aramaki-Aza, Aoba-ku, Sendai, Miyagi 980-8578,
Japan (phone: þ 81-22-795-3554; fax: þ 81-22-795-6566; e-mail: yhayashi@m.tohoku.ac.jp)
and Markus Reiher*
Laboratorium fꢀr Physikalische Chemie, Departement Chemie und Angewandte Biowissenschaften,
ETH-Zꢀrich, Hçnggerberg HCI F 235, Wolfgang-Pauli-Strasse 10, CH-8093 Zꢀrich
(
phone: þ 41-44-633-4308; fax: þ 41-44-633-1595; e-mail: markus.reiher@phys.chem.ethz.ch)
Dedicated to Jack D. Dunitz – friend and ꢁprofessor for the professorsꢂ of LOC – on the occasion of his
9
0th birthday
The stoichiometric reactions of enamines prepared from aldehydes and diphenyl-prolinol silyl ethers
intermediates of numerous organocatalytic processes) with nitro olefins have been investigated. As
(
reported in the last century for simple achiral and chiral enamines, the products are cyclobutanes (4 with
monosubstituted nitro-ethenes), dihydro-oxazine N-oxide derivatives (5 with disubstituted nitro-
ethenes), and nitro enamines derived from g-nitro aldehydes (6, often formed after longer reaction
times). The same types of products were shown to be formed, when the reactions were carried out with
peptides H-Pro-Pro-Xaa-OMe that lack an acidic H-atom. Functionalized components such as alkoxy
enamines, nitro-acrylates, acetamido-nitro-ethylene, or hydroxylated nitro olefins also form products
carrying the diphenyl-prolinol silyl ether as a substituent. All of these products must be considered
intermediates in the corresponding catalytic reactions; the investigation of their chemical properties
provided useful hints about the rates, the conditions, the catalyst resting states or irreversible traps, and/
or the limitations of the corresponding organocatalytic processes. High-level DFTand MP2 computations
of the structures of alkoxy enamines and thermodynamic data of a cyclobutane dissociation are also
described. Some results obtained with the stoichiometrically prepared intermediates are not compatible
with previous mechanistic proposals and assumptions.
1
)
)
Part of the results described herein had been reported in a preliminary communication [1].
) Postdoctoral fellow at ETH Zꢀrich 2011/12, partially financed by Swiss National Science
2
a
Foundation (SNF-No. 200020-126693). All the isolations of intermediates and preparations of single
b
crystals, described herein, have been carried out by X. S. ) Postdoctoral fellow at ETH Zꢀrich 2012/
c
1
3. ) Some of the experiments were carried out by T. M. during a three-month stay at ETH Zꢀrich
d
(
August – October 2012), financed by Tokyo University of Science. ) Postdoctoral fellow at Tokyo
University of Science (2010 – 2012), financed by JSPS (PE10034; P10817).
ꢃ 2013 Verlag Helvetica Chimica Acta AG, Zꢀrich

8
00
Helvetica Chimica Acta – Vol. 96 (2013)
1. Introduction. – Shortly after the renaissance [2] of organocatalysis [3] had started
in the year 2000, the enantioselective Michael addition of aldehydes and ketones to
nitro olefins, catalyzed by chiral sec-amines, was investigated [4 – 6]. This type of
reaction, leading to g-nitro carbonyl compounds with up to three new stereogenic
centers, became a general test track for those organocatalysts that form enamines as
nucleophilic reactive intermediates (Scheme 1). Most of the sec-amino derivatives used
3
are of the pyrrolidine type A ). The underlying reactions of chiral and achiral enamines
(cf. B) with nitro olefins in stoichiometric applications have been studied in the years
between 1964 and 2000 [8], and have led a rather shadowy existence. A careful study of
this old literature provides the mechanistic manifold presented in Scheme 2. In these
stoichiometric reactions, a mono- or disubstituted enamine and a nitro olefin are mixed
under anhydrous conditions in an aprotic solvent such as MeCN, Et O, or hexanes,
2
often with cooling, to give, after non-aqueous workup, either a cyclobutane C
(
(
preferentially with monosubstituted nitro olefins) or an oxazine derivative D
preferentially with disubstituted nitro olefins), or a nitro enamine, E/E’. All of these
–
isolable – primary products, C – E, are converted to the actual Michael adducts F upon
hydrolysis under acidic conditions (cf. EtOH/H O/HCl). Possible routes for these
2
hydrolytic conversions, mostly already proposed in the old literature [8], are included
in Scheme 2. The elusive iminium-nitronate zwitterion G plays a central role in these
proposals. For the reactions of the (E)- and (Z)-1-morpholinopropene (enamines from
propanal and morpholine) with nitrostyrene, and of (E)- and (Z)-nitrostyrene with
morpholino-cyclohexene (enamine from cyclohexanone and morpholine), the Huisgen
test [9] for (2 þ 2) cycloadditions via zwitterionic intermediates was positive: there is
scrambling between the (E)- and (Z)-isomers, and the reaction is non-stereospecific
4
(
Scheme 3; see [10] and 1985 publication (with Laube) [8]) ). Trapping of the
Scheme 1. sec-Amine (A)-Catalyzed Michael Additions of Aldehydes and Ketones to Nitro Olefins and
Enamine Intermediate B
3)
4)
The number of reports on this type of reactions is so large (many dozens!) that we can only refer to
some selected reviews [7].
In the case of the oxazine derivatives D (Scheme 2), a formally allowed [11], non-synchronous,
concerted (4 þ 2) cycloaddition, a so-called inverse-electron-demand DielsꢀAlder reaction, without
a zwitterion intermediate, could be involved. In a recent DFT calculation, no zwitterion of type G
was located as a low-lying energy minimum [12]. For synthetic applications and mechanistic
discussions of oxazine derivatives resulting from (4 þ 2) cycloadditions to nitro olefins of enes other
than enamines (e.g., enol ethers), see [13].

Helvetica Chimica Acta – Vol. 96 (2013)
801

8
02
Helvetica Chimica Acta – Vol. 96 (2013)
Scheme 3. Test for Zwitterionic Intermediates [9] in the (2 þ 2) Cycloaddition of an (E)- and (Z)-
Enamine, and an (E)- and (Z)-Nitro Olefin. The same results (yield, diastereoselectivity) were obtained
with the (E)- and (Z)-starting materials (see 1985 paper with Laube in [8], and [10]).
zwitterion in the course of aqueous hydrolysis has been considered crucial for the
cyclobutane C to be converted to the nitro carbonyl compound F, while the dihydro-
oxazine N-oxide D, a cyclic nitronate ester, could also undergo acidic hydrolysis to an
5
6
iminium-nitronic acid or nitro iminium ion ) without the intermediacy of a zwitterion ).
Except on route G ! E ! F, the absolute (with chiral enamines) and relative
1
3
configurations of the two stereogenic centers (bearing R and R ) in a- and b-position
5)
This was suggested by us (Footnote 5 in [1]) as an alternative to other ꢁexitsꢂ from the cyclic
intermediate D to open-chain precursors of the final nitro carbonyl compound F. Computational
results by Pihko and co-workers [12] indicate that protonation of an oxazine derivative of type D in
the 3-position of the heterocycle is ꢁclearly favored kineticallyꢂ (vide infra Footnote 38), generating a
ꢀ
nitro-alkane moiety; we are unable to judge whether the alternative protonation on O , with
subsequent formation of a nitronic acid group, has been probed in this computational analysis.
Results of deuterolysis experiments are not compatible with C(3)-protonation (see Scheme 10,e).
Rather than simple hydrolysis to the nitro carbonyl derivative of type F, Nef reactions have been
observed to occur during acidic hydrolyses of the primary products (cf. Valentin and co-workers in
6)
[
8], as well as Yoshikoshi and co-workers in [13]).

Helvetica Chimica Acta – Vol. 96 (2013)
803
to C¼O in Fare set in the coupling step between the trigonal centers of the enamine and
7
the nitro-olefin component ).
All the products shown (in orange and blue) and all the processes outlined in
Schemes 2 and 3 for the stoichiometric reactions between enamines and nitro olefins
must be considered for the sec-amine-catalyzed addition of aldehydes and ketones to
nitro olefins presented in Scheme 1. At the beginning of the competitive run in
enantioselective organocatalysis, there was no time for mechanistic investigations, so
8
that no or only a simplistic catalytic cycle ) was proposed, focusing on the relative
7
8
topicity of the primary C,C-bond formation [4 – 7] ) ).
Intrigued by the observation that the addition of aldehydes to nitro olefins,
catalyzed by the diphenyl-prolinol silyl ethers 1a and 1b under standard conditions
(
5 mol-% 1, 5 mol-% 4-nitrophenol, 24 h, room temperature) [14], appeared to be
subject to steric hindrance, resulting from the size and the number of substituents on
the two reactants (an effect which had not been noticed in the stoichiometric
9
reactions) ) two of our groups (Y. H. and D. S. ) decided to study the addition of 1-
derived enamines to various nitro olefins under stoichiometric, anhydrous condi-
10 11
tions ) ). In this work, we present hitherto unpublished (experimental) details and
case studies, which show that, for the catalytic reaction to take place, and for identifying
the best conditions, it is useful to know the chemical properties of the intermediates
involved.
2
. Stoichiometric Reactions of Enamines, Derived from the Diphenylprolinol Silyl
Ethers 1 and Aldehydes, with Mono- and Disubstituted Nitro Olefins. – We prepared
1
2
solutions, in dry (D )benzene or (D )toluene, of enamines 2 (formed immediately ))
6
8
1
3
from aldehydes and the pyrrolidines 1, removing H O with molecular sieves )
2
(
Scheme 4). The solution of the enamines was then mixed with the nitro olefin,
dissolved in the same solvent, and NMR spectra were recorded immediately and after
certain periods of time. In fact, and as expected, the three types of products identified
and reported in the old literature (Scheme 2) were detected by NMR analysis: the
cyclobutanes 4 (Fig. 1; cf. C) the dihydro-oxazine N-oxides 5 (Fig. 2; cf. D), and the
nitro enamines 6 (Fig. 3; cf. E). The ratio depended upon the particular pair of
reactants 2 and 3, upon the temperature, and upon the reaction time, with the
7)
Following the general ꢁTopological Rule for C,C-Bond Forming Processesꢂ involving coupling of two
trigonal centers (see 1981 publication (with Golinski) in [8]).
See, e.g., Scheme 1 in [14] and refs. cit. therein.
In fact, when mixing enamines with nitro olefins an exothermic reaction takes place, so that cooling
or starting the reaction at dry-ice temp. is sometimes recommended [8].
See [14] and the preliminary communication [1]. In the original work of Hayashi et al., no acid
additive was used, and the solvent was hexane [15].
For independent works along these lines, see [12][16][17]. For early mechanistic studies on the
catalytic reaction, see Sect. 3 and refs. cit. therein.
In all cases, the enamine had formed completely after a few minutes (ꢁbefore we arrived at the
NMR machine for recording the spectrumꢂ).
In the case of propanal, an excess aldehyde had to be employed to achieve complete conversion in
the presence of 4-ꢄ molecular sieves (MS), but not with 3-ꢄ MS. We interpret this observation by
assuming that propanal can enter the cavities of 4-ꢄ but not of 3-ꢄ MS.
8
)
9
)
10)
11)
12)
13)

8
04
Helvetica Chimica Acta – Vol. 96 (2013)
Scheme 4. Preparation of Enamine 2 Solutions from Aldehydes and Pyrrolidines 1, and Reactions with
Nitro Olefins 3 with Formation of Mixtures of Cyclobutanes 4, Dihydrooxazine N-Oxides 5, and Nitro
Enamines 6. Cyclobutanes are preferred with monosubstituted nitro olefins, and oxazine derivatives are
preferred with disubstituted nitro olefins, and nitro enamines are often the major products after longer
periods of time.
cyclobutane 4 prevailing with monosubstituted nitro olefins, the heterocycle 5 being the
only product detected with disubstituted nitro olefins, and the nitro enamine 6 often
becoming the major product after extended periods of time. Equilibria between the
four- and the six-membered rings, and between these cyclic compounds and the starting
materials, enamine and nitro olefin, were observed. As will be discussed in the
following Sections in detail, some of the four- and six-membered-ring compounds 4 and
5
are stable in solution at up to 508, or even to 1008, and in five cases crystalline samples,
14
suitable for single-crystal X-ray structure determination, could be prepared (Fig. 4) ).
Most of the NMR data of compounds 4, 5, and 6, presented in the Exper. Part, have been
obtained by analysis of the complex spectra of mixtures of – sometimes equilibrating –
compounds, using 2D-COSY, NOSY, HSQC, and EXSY techniques ) ).
.1. The Cyclobutanes 4. As we had expected, cyclobutanes, 4a – 4q, are always
formed from the enamines 2 and the monosubstituted nitro olefins 3 (R ¼ H), not only
in those cases, in which the corresponding catalytic reaction under our standard
conditions (toluene, room temperature, 5 mol-% 1a, 5 mol-% 4-nitrophenol, up to
15 16
2
3
48 h; Scheme 1) [14] takes place (cf. 4a – 4j in Fig. 1), but also with the sterically more
hindered reactants (cf. 4k – 4q in Fig. 1). The reaction becomes slower with increasing
14)
As can be seen from Fig. 4 of the crystal structures, the virtual electron pair on the pyrrolidino N-
atom is in all cases antiperiplanar to the most polar CꢀCꢀN (4q) and CꢀOꢀN (5b – 5d and 5g) bond
within the ring, as if the molecules would be ready for opening up to zwitterions in a
stereoelectronically assisted process.
See Figs. 1 – 3 and Scheme 6 in our preliminary communication [1].
The isolation and characterization of the cyclobutanes 4e and 4q have been described in our
previous paper [14]; improved procedures are presented herein.
1
1
5
6
)
)

Helvetica Chimica Acta – Vol. 96 (2013)
805

8
06
Helvetica Chimica Acta – Vol. 96 (2013)
Fig. 2. Dihydro-oxazine N-oxides 5 formed under the conditions specified in Scheme 4. Identification by
in situ NMR analysis, by isolation (chromatography, crystallization), and/or by X-ray crystal-structure
determination (see Exper. Part).
Fig. 3. Nitro Enamines 6 formed, especially after prolonged reaction times, from enamines 2 and nitro
olefins 3 under anhydrous conditions (Scheme 4). Identification by in situ NMR analysis. Under carefully
controlled conditions, solutions of 90% pure nitro enamine 6a can be prepared. The letters in the
compound numbering 6a – 6j correlate with those in Fig. 1.

Helvetica Chimica Acta – Vol. 96 (2013)
807
Fig. 4. X-Ray crystal structures of the cyclobutane 4q, and of the dihydro-1,2-oxazine N-oxides 5b – 5d and
g. The structures of 4q, 5b, and 5d have been reported in our preliminary communication [1]. The
5
structure of 5g was determined by the Gellman group in 2009, but submitted to the Cambridge
Crystallographic Data Centre (CCDC) only recently. In all structures, there is stereoelectronic n ! s*
N
C,C
or n ! s* stabilization [18a] (see also Footnote 14). Note that the exocyclic bond of the pyrrolidine
N
C,O
ring has the generally preferred [18b] exo-synclinal conformation only in 5b, 5c, and 5g, while 4q and 5d
have the antiperiplanar conformation of this bond (cf. results of calculations in Fig. 8 and structure of 21
in Scheme 11).
size of the substituents on the enamine and on the nitro olefin precursors, and only with
t
too much bulk (cf. two Bu groups) no cycloaddition product could be detected (see
dotted-line box in Fig. 1). Thus, the lack of nitro-aldehyde formation under the

8
08
Helvetica Chimica Acta – Vol. 96 (2013)
standard catalytic conditions with more bulky substituents is, in many cases, not due to
lack of reactivity between the intermediate enamine and the nitro olefin.
2
.1.1. Equilibrium between a Cyclobutane of Type 4 and Enamine þ Nitro Olefin. A
17
solution in C D of the cyclobutane 4e, prepared as shown in Scheme 4 ), containing
6
6
18
no NMR-detectable amount of oxazine 5 or nitro enamine 6 derivative ), but a few
percent of nitrostyrene and of the enamine 2 (from 3-methylbutanal and 1a), was
stepwise heated to 668 and cooled back to r.t.; the ratio 4e/enamine was determined by
integration of suitable NMR signals after each change of temperature (for details, see
Exper. Part). The results are presented in Fig. 5 and show that there is a clean, fully
reversible equilibrium between the (2 þ 2) cycloadduct 4e and its precursors; due to the
ring strain of the cyclobutane and the necessarily large entropy factor, the cyclization is
only weakly exergonic (DG ¼ ꢀ4.3 kcal/mol). We have not systematically looked for
such equilibria with other cyclobutanes 4, but it must be expected that they occur quite
Fig. 5. Equilibrium between the amino-nitro-cyclobutane 4e, and its precursors enamine and nitro olefin.
There is an excess nitro olefin in the reaction mixture (for details, see Table 5 in Exper. Part). According
to the NMR analysis, the solution of the equilibrating species contains no oxazine derivative 5, and no
nitro enamine 6 is detected during this experiment. The iminium-nitronate zwitterion is assumed to be
involved in this equilibration process (cf. Scheme 3).
17)
18)
A similar experiment, conducted with a less pure sample of 4e, has been described in our previous
paper [14].
... at temperatures between 25 and 668.

Helvetica Chimica Acta – Vol. 96 (2013)
809
generally when there is not too much bulk of substituents on the four-membered ring
cf. 4q).
The conclusion for the corresponding catalytic reaction (in toluene, Table 4 in [14])
(
would be that a zwitterion of type G is an intermediate en route from the cyclobutane to
the open-chain nitro aldehyde of type F19).
2
.1.2. Stability of Cyclobutane 4q. Our most stable amino-nitro-cyclobutane 4q
t
(
Fig. 4 and Footnote 16) was prepared from the Bu-substituted nitro-ethene and the
i
Pr-substituted enamine as outlined in Scheme 4; formation of 4q took 3 d in
D )benzene (85% yield after crystallization; m.p. 138 – 1448 (dec.)) and 18 h in the
(
6
20
presence of 1 equiv. 4-nitrophenol ). To test the thermal stability of this cyclobutane,
we heated its solution in (D )toluene in the presence of 1 equiv. of nitrostyrene (i.e.,
8
(
E)-b-nitrostyrene; sealed NMR tube). The result is outlined in Scheme 5: up to 808,
nothing happened; at 908, we identified traces of the enamine precursor, and after
heating to 1208 the cyclobutane 4e (product of scrambling) was detected. When the
solution of 4q was kept at room temperature in the presence of 4-nitrophenol and H O,
2
there was no product of hydrolysis after 24 h; addition of acidic alumina (activity grade
1
) or of ClCH CO H to the CHCl solution, or chromatography of 4q on silica gel did
2 2 3
not lead to the corresponding nitro enamine or nitro aldehyde.
Scheme 5. Thermal and Hydrolytic Stability of the Cyclobutane 4q in (D )Toluene. a) The solution
8
containing 1 equiv. of nitrostyrene was heated for 30 min at the indicated temperature in a sealed NMR
tube, and, after cooling to ambient temperature, the spectra were recorded. The cyclobutane 4e (with
ꢁ
exchangedꢂ nitro olefin-derived part of the molecule) is detected only after heating above 1008. b) The
mixture containing 10 equiv. of H O and 4-nitrophenol was kept at r.t. (cf. the catalytic conditions
2
[
1][14]): no NMR-detectable trace of the corresponding nitro aldehyde was formed under these
conditions. For other tested hydrolysis/decomposition conditions, see accompanying text.
1
2
9
)
)
See the article by R. Huisgen entitled ꢁCan Tetramethylene Intermediates Be Intercepted?ꢂ [9b].
Acid additives are essential for effective catalysis of the Michael addition of carbonyl compounds to
nitro olefins (Scheme 1); see discussions and references in [7][14] and in [19].
0

8
10
Helvetica Chimica Acta – Vol. 96 (2013)
The conclusion for the reaction under our standard catalytic conditions would be that
the (2 þ 2) cycloadduct is slowly formed, but is not opening up to a linear intermediate
to give the nitro aldehyde; in this case, the cyclobutane is a dead-end trap for the
pyrrolidine catalyst.
2
.1.3. Equilibrium between Cyclobutane 4a and Oxazine Oxide Derivative 5a. In an
elaborate NMR analysis, we have found that the reaction between the propanal
enamine and nitrostyrene in (D )benzene leads to an equilibrating mixture of the
6
cyclobutane 4a and the oxazine N-oxide 5a (4 :1 ratio at room temperature), which is
converted to the nitro enamine 6a within 15 h; in CH Cl solution the open-chain
2
2
compound is formed within 10 min (Scheme 6) [1]. This was recently confirmed by an
independent investigation by Pihko and co-workers (propanal/nitrostyrene/1a 1:1:2,
21
(
D )toluene, room temperature, complete conversion in 24 h) ). A reasonable
8
mechanism for the equilibration 4a > 5a and for the conversion 4a/5a ! 6a would be
22
formation of the zwitterion and an intra- ) or intermolecular H-shift, which was
confirmed by the D-labeling experiment outlined in Scheme 6,b.
2
.1.4. Acid Hydrolysis of the Nitro Enamine 6a to the Nitro Aldehyde 7a. Under the
conditions indicated in Scheme 6,a, we could prepare rather pure samples of the nitro
enamine 6a (84% on an 0.5-g scale), which we used to probe the diastereoselectivity of
the acidic hydrolysis to the nitro aldehyde 7a under two sets of conditions (Scheme 7):
similar to the catalytic reaction [14], a 0.05m solution of 6a, 1 equiv. 4-nitrophenol, and
1
equiv. H O in benzene was kept at room temperature for up to 40 h, and similar to the
2
classical workup of stoichiometric reactions of enamines with nitro olefins [8], 6a was
dissolved in EtOH and treated with H O/HCl. As can be seen from the data in the table
2
of Scheme 7, the selectivities determined by NMR analysis are poor (ꢁ 3 :1), as
compared to those of the corresponding catalytic reaction providing the diastereo-
isomer 7a with a dr of 15 :1 [14]. Bur e` s et al. [17a] have performed a kinetic analysis of
the reverse reactions (7a þ 1a > 6a þ H O and epi-7a þ 1a > 6a þ H O) from which
2
2
they calculated a protonation selectivity of ꢂ 39 :1; they concluded that the
configuration of the center bearing the Me group is not set in the coupling step of
the trigonal centers of enamine and nitro olefin, but in the nitro enamine protonation
step and modified the catalytic cycle accordingly. We are unable to interpret the
discrepancies between these experimental results, i.e., the diastereoselectivities of the
catalytic reaction without added acid (14 :1 [14]) or with 4-nitrophenol (15 :1 [14]), the
stoichiometric reaction with 4-nitrophenol (3 :1) or with HCl (1.5 :1) (Scheme 7), and
the kinetics of equilibration between the diastereoisomers 7a and epi-7a with 1a (39 :1
23
[
17a]); an expert physical-chemical analysis is required ).
2
.1.5. Reactions of Enamines from 2-Alkoxyacetaldehyde with 3-Nitroacrylate ( ¼
(
E)-3-Nitroprop-2-enoate) and 2-Acetamido-nitroethene. These reactants have been
used for syntheses of oseltamivir, the active agent of the antiviral drug Tamiflu [21 –
ꢀ
21)
See Fig. 2 in [12] (received June 20, revised October 10, and published online November 11, 2012);
cf. also [17]; [1] was received May 31, and published online July 11, 2012.
See also Scheme 1 in [1].
2
2
2
3
)
)
Protonation, deprotonation, and iminium-ion formation are involved in the reaction between a sec-
amine and an aldehyde, as well as in the reverse reaction, the hydrolysis of an enamine; in the
þ
þ
þ
reactions discussed here, the protonating species may be H O , EtOH , 4-nitrophenol, R*NH .
3
2
2
2

Helvetica Chimica Acta – Vol. 96 (2013)
811
Scheme 6. Equilibrium between the Nitrocyclobutane 4a and Its Six-Membered-Ring Isomer 5a, and
Rearrangement to the Nitro Enamine 6a. The NMR spectra for the detection of this process are shown
and the detailed analysis is described in [1]. a) Slow ring opening in benzene, fast ring opening in CH Cl
2
2
with the corresponding zwitterion as presumed intermediate. b) With D-labeled enamine, the label ends
up in the a-NO position. For a possible intramolecular H-shift converting the zwitterion to the nitro
2
enamine, see Scheme 1 in [1]. For a preparation of the dideutero-propanal, see [20].
2
3]. Before turning to the stoichiometric reactions of the corresponding isopentyloxy
enamine, a new type of resting state of the catalyst 1a is discussed.
.1.5.1. The Michael Adduct of 1a to Nitroacrylate. The nitroacrylate is a much more
reactive Michael acceptor than nitrostyrenes or aliphatic nitro olefins. Indeed, when the
catalystꢂ 1a was added to a tert-butyl-3-nitroacrylate solution in C D or CD Cl , the
2
ꢁ
6
6
2
2

8
12
Helvetica Chimica Acta – Vol. 96 (2013)
Scheme 7. Poorly Stereoselective Hydrolysis of the Nitro Enamine 6a to the Nitro Aldehyde 7a.
Conditions A as for the catalytic reaction [14]; conditions B are typical for the workup of stoichiometric
reactions of enamines with nitro olefins [8].
24
adduct 8 was formed (Scheme 8); in CH Cl this took just a few minutes ). The
2
2
reaction was reversible, i.e., the adduct 8 is a resting state of the catalyst: addition of 3-
methylbutanal led to the cyclobutane 4g, and then to the nitro enamine 6g, and finally
Scheme 8. Addition of ꢁCatalystꢂ 1a to Nitroacrylate and Reaction of the Adduct 8 with 3-Methylbutanal
to Eventually Produce the Formyl Ester 7g (up to 72%; for details, see Table 7 in the Exper. Part)
24)
With nitrostyrene, no reaction takes place (in C D ) in 24 h; Blackmond and co-workers mention
6 6
<
5% of adduct formation, without giving experimental evidence [16]. With ethyl (E)-3-nitrobut-2-
enoate, there is 71% conversion to the corresponding adduct in 20 h (C D , room temperature).
6
6

Helvetica Chimica Acta – Vol. 96 (2013)
813
to the corresponding nitro aldehyde (7g in Scheme 8) over a period of hours and days
NMR analysis; for details see Table 7 in Exper. Part). To consider the importance of
(
this resting state 8 of the catalyst 1a, we have to remember a) that there is a large excess
of the nitro olefin vs. sec-amine under catalytic conditions (shifting an equilibrium
towards adduct 8), b) that there was always an acid co-catalyst necessary to render the
catalytic reaction effective, and acid will catalyze the elimination 8 ! nitroacrylate þ
1a, and c) that this conjugate addition was much slower with the less reactive nitro
olefins.
.1.5.2. Reactions of (E)/(Z)-Pentyloxy enamines with Functionalized (E)- and (Z)-
2
25
Nitro Olefins. When the pent-3-yloxy aldehyde 9 ) shown in Scheme 9 was mixed with
the pyrrolidine 1b in various dry solvents in the presence of 3-ꢄ MS, a mixture (Z)/
(
E)-enamine 10a was formed. The resulting solution in C D of the two isomers
6 6
contained an excess (up to 1.8 :1) of the (Z)-form. For a theoretical analysis of this type
Scheme 9. Preparation of Solutions of (Z)/(E)-Mixtures of Enamine 10a from Aldehyde 9. a) Except in
CDCl , the (Z)-isomer prevails slightly. There is essentially no temperature effect and only a small effect
3
of 4-nitrophenol on the (Z)/(E)-ratio. In an EXSY-NMR spectrum, there is no cross-peak between the
signals of the two isomers. We assume that the ratios are thermodynamical values, in agreement with
some of the calculated energy differences between (E)- and (Z)-forms of alkoxy enamines (see the
theoretical investigations of enamines of type 10 described in Sect. 2.1.5.3 (Fig. 8 and Table 2)). b) Under
the typical catalytic conditions, there is always an acid co-catalyst present [21 – 23], which is expected to
equilibrate the (E)- and (Z)-alkoxy enamines!
25)
For organocatalytic Michael additions of 2-(trimethylsilyloxy)acetaldehyde to nitro olefins, see [24].

8
14
Helvetica Chimica Acta – Vol. 96 (2013)
of an alkoxy enamine, which at the same time is an amino-enol ether, a highly electron-
rich system [25a – c], see Sect. 2.1.5.3.
Upon addition of an equimolar amount of t-butyl 3-nitroacrylate to the solution in
C D of the alkoxy enamine 10a ((Z)/(E) ca. 1.5 :1), the highly electrophilic nitro
6
6
compound disappeared completely and immediately (Fig. 6). The two stereoisomeric
cyclobutanes 4i (formally derived from (Z)-10a) and 4h (from (E)-10a) were formed in
a ratio of ca. 1:2, and very little nitro enamine 6h was detected in the initial NMR
analysis. Over a period of ca. 50 min, the fraction of the cyclobutane (4i), derived from
(
Z)-enamine, decreased, and the ratio of the five components of the mixture remained
more or less constant (an equilibrium situation?), with the (Z)/(E)-enamine ratio still
being close to 1.5 :1, and the ratio all-trans/cis,cis,trans,trans-cyclobutane, i.e. 4h/4i,
being ca. 8 :1; the nitro enamine (6h) concentration had somewhat increased at this
point, more or less at the expense of the minor cyclobutane 4i. Four points are worth
emphasizing: a) in the instantaneous (ꢁ 5 min) reaction of the electron-rich alkoxy
enamine with the electron-poor nitroacrylate, both enamines reacted, but there was
twice as much product (i.e., 4h) derived from the minor enamine isomer ((E)-10a); b)
the concentration of the all-trans-cyclobutane, 4h, was constant after the first ꢁflashꢂ
reaction; c) the corresponding catalytic reaction gave best results when carried out in
toluene and in the presence of ClCH CO H (yield of nitro aldehyde > 99%, dr 7.8 :1, er
2
2
98.5 :1.5) [14][21]; d) an epimerization on the RO-substituted center of 4i to give 4h
would require dissociation back to the cis-enamine and cis/trans-isomerization thereof
(cf. Scheme 9).
In any case, it looks as if the success of the catalytic reaction is the result of a complex
series of steps involving various species, possibly including the resting state 8
(
(
Scheme 8), and not just a simple preference of the transition state/zwitterion H
bottom part of Fig. 6) ) ).
26 27
The stoichiometric reaction between the (E)/(Z)-enamines 10a and acetamido-
nitro-ethene took an even more surprising course (Fig. 7). The nitro compound is less
reactive than nitroacrylate; (Z)-enamine 10a reacts faster than the (E)-enamine.
Besides the open-chain nitro enamine 6j only one cyclobutane, the ꢁall wrongꢂ all-trans-
compound 4j was formed, and not the cis-trans-cis-trans-isomer expected from the
trajectory and zwitterion/transition state I (Fig. 7). Again, a more or less stable
composition of the components resulted from this stoichiometric reaction after ca. 1 h.
We wondered whether the acetamido-nitroethene really has the cis-configuration,
stabilized by an intramolecular H-bond, as generally assumed, and recorded its NMR
spectrum in carefully neutralized CDCl and in a series of solvents containing H-bond
3
acceptor groups, all the way to (D )DMSO (the solvent of choice for breaking intra-
6
28
and intermolecular H-bonds, by being an excellent H-bond acceptor) ). The results
26)
... as seemed to be the case when 2-alkyl- and 2-aryloxy-acetaldehydes were added to nitro olefins
including ethyl nitroacrylate) in a less diastereoselective reaction (catalyst 1a, toluene, room
temperature, 55 or 96 h) [14][23b]; see, however, the reaction in DMSO and in CHCl /H O [23a].
(
3
2
27)
28)
In the catalytic reaction, it is important to perform the workup at the right point in time to avoid
erosion of the diastereoselectivity [21].
This is why we have expressed doubt [25k] about the stereodirecting role of the carboxylic acid
group by H-bonding in proline catalysis of reactions carried out in DMSO [6][26].

Helvetica Chimica Acta – Vol. 96 (2013)
815
Fig. 6. Stoichiometric reaction of the (Z/E)-enamine 10a with (E)-nitroacrylate in benzene, NMR analysis
of the reaction mixture, and model for the transition state/zwitterion H with least steric hindrance between
the substituents ((E)-enamine/(E)-nitroacrylate). The approach H would lead to the all-trans-cyclo-
butane 4h. Epimerization 4i ! 4h would require dissociation to the starting materials (through a
zwitterion) and (Z/E)-10a isomerization. From the concentration traces, it looks like 6h is actually
formed from 4i, rather than from 4h under these conditions. Note that in our (Y. H. ) published [21]
oseltamivir synthesis ent-1a was used as catalyst (solvent toluene, co-catalyst ClCH CO H).
2
2

8
16
Helvetica Chimica Acta – Vol. 96 (2013)
Fig. 7. Stoichiometric reaction of the (Z/E)-enamine 10a with (Z)-acetamido-nitroethene in benzene,
NMR analysis of the reaction mixture, and model for the transition state/zwitterion I with least steric
hindrance between the substituents ((Z)-enamine/(Z)-nitro olefin). The approach I would lead to a cis-
trans-cis-trans-cyclobutane, which is not detected. Note that the (Re/Re)-coupling (see J) of (E)-
enamine with (E)-nitro olefin (cf. Table 1) would give a diastereoisomer of 4j, in which the absolute
configurations of all four stereogenic centers on the four-membered ring are reversed: epi-ent-4j
(
epimeric at C(2) of the pyrrolidine ring); by the NMR analysis, we could not reliably differentiate
between 4j and epi-ent-4j, for instance, from the NOE between the H-atom in a-position to NO and the
2
HꢀC(2) of the pyrrolidine ring (in some of the cyclobutanes 4, this NOE is strong, in others it is absent).
The epi-ent-4j cyclobutane would, of course, eventually lead to ent-oseltamivir! Note that, in the
oseltamivir synthesis of Ma and co-workers [22], catalyst 1, with R Si ¼ Me Si and naphthalenyl instead
3
3
of Ph groups on the prolinol ether unit, was used. In our (Y. H. ) recent unpublished work, the catalyst 1b
turned out to be superior to 1a.

Helvetica Chimica Acta – Vol. 96 (2013)
817
are compiled in Table 1: in ꢁneutralꢂ CHCl 29) the cis-isomer prevailed, in DMSO ) the
30
3
trans-isomer was in excess. This means that the involvement of the trans-acetamido-
nitroethene has to be considered in the oseltamivir syntheses using this nitro olefin
31
derivative [21 – 23] ). However, as shown in Fig. 7 (bottom part), coupling J of (E)-
enamine 10a with the (E)-acetamido-nitroethene would eventually lead to ent-
oseltamivir.
Table 1. (E)/(Z) Ratio [%] of Acetamido-nitroethene in Different Solvents as Determined by NMR
Spectroscopy
Solvents
CDCl washed with basic Al O
> 99
80
18
15
11
7
< 1
20
82
85
89
93
3
2
3
(
D )THF
8
CD OD
3
(
(
(
D )Pyridine
5
D )DMF
7
D )DMSO
6
Obviously, an even more complex, most puzzling sequence of events is likely to be
involved in the case of the successful catalytic application of isopentyloxy-acetaldehyde
addition to acetamido-nitroethene (see Fig. 7) than in the case of addition to the
nitroacrylate.
2
.1.5.3. DFT- and MP2-Calculations of the Structures of Alkoxy enamines. The
experimental observation of a (Z/E)-ratio close to 1:1 of the enamine 10a from
32a
diphenyl-prolinol Si ether 1a and 2-(pentyloxy)acetaldehyde prompted us ) to carry
out a computational investigation of alkoxy enamines, to evaluate whether the almost
32b
equal stability of these (Z)- and (E)-isomers is found for the isolated species ).
Computational Methodology. For all quantum-chemical calculations, we employed
the Turbomole 6.4 suite of programs [25d]. All density-functional theory (DFT)
calculations applied the BP86 density functional [25e,f]. To study the method
dependence of the DFT results obtained, we compared with second-order Møller-
Plesset perturbation theory (MP2) results for selected molecules. For large molecules,
MP2 calculations became less and less feasible so that the empirical D3 corrections by
29)
30)
31)
We thank Dr. J. Durmis [23a,c] for a personal communication about a 10:1 cis/trans ratio of
acetamido-nitroethene in CDCl3.
In fact, the catalytic reaction used in a recent oseltamivir synthesis [23a] occurred with best
diastereoselectivity in DMSO.
The acid co-catalyst present in all catalytic coupling reactions between 2-alkoxyacetaldehydes and
acetamido-nitroethene will thus not only be able to catalyze (Z)/(E)-enamine (cf. Scheme 9,b)) but
might also cause (Z)/(E)-nitro olefin equilibration.
32)
a
) Anti-intuitive ꢁcis-effectsꢂ have always been intriguing for organic chemists [25a – c]. b) For a
comprehensive review (95 pages!) entitled ꢁQuantum Mechanical Investigations of Organocatalysis:
Mechanisms, Reactivities, Selectivitiesꢂ, see [7] (Houk and co-workers).

8
18
Helvetica Chimica Acta – Vol. 96 (2013)
Grimme et al. [25g] have been used to include dispersion effects in the DFT/BP86
calculations. All energies have been obtained after full structure optimization with the
quantum-chemical method under consideration. Ahlrichsꢂ basis sets of triple-
(
TZVPP) and quadruple-zeta (def 2-QZVPP) quality, respectively, have been used
as implemented in Turbomole 6.4. In these basis sets, polarization functions were taken
from Dunningꢂs correlation consistent basis sets. The resolution-of-the-identity density-
fitting technique has been invoked for BP86 and MP2 with the corresponding auxiliary
basis sets from Turbomole 6.4 in order to accelerate the calculations. The orbitals have
been tightly converged until the change in relative electronic energy in the final self-
ꢀ8
consistent field iterations step was smaller than 10 Hartree. Moreover, the structure
optimization was carried out until the length of the geometry gradient was on the order
ꢀ
4
of 10 atomic units. In all DFT calculations, we employed a fine numerical integration
grid (ꢁm4ꢂ in Turbomole notation).
Conformational effects can blur the energetic ordering of different isomers if alkyl
substituents are present. To avoid such effects, we made sure that corresponding (Z)-
and (E)-isomers always feature similar conformations of the corresponding alkyl
substituents (see also below for a more detailed discussion of such effects). The
energies reported are differences of electronic energies. Hence, purely electronic
effects have been considered rather than zero-point-energy and temperature correc-
tions as well as entropy effects, which we have not taken into account.
As molecules for the calculations, we chose the simple, prototypical 1-(dimethyl-
amino)-2-methoxyethene (10b) and the diphenylprolinol trimethylsilyl ether-derived
33
1
-(benzyloxy)- ) and 1-(isopentyloxy)-2-pyrrolidinoethenes (10c and 10d, resp.). The
results are collected in Table 2 and Fig. 8.
Discussion. First of all, we find a more or less pronounced pyramidalization of the
N-atoms in all calculated structures. We use Dunitzꢂs definition for the degree of
pyramidalization D [ꢄ] as the distance of the N-atom from the plane spanned by its
three bonding partners (here C-atoms) [25h]. The D values (Table 2, last two columns)
34
are between 0.0 and 0.3 ꢄ ). For the generic small molecule 10b that carries only Me
groups as substituents, we found a BP86/TZVPP D value of 0.28 ꢄ for the (Z)-isomer,
whereas it was slightly larger (by 0.02 ꢄ) for the (E)-isomer. While these D values
increased in the MP2 calculations by 0.06 ꢄ, the difference between the (Z)- and (E)-
isomers remained the same, i.e., 0.02 ꢄ. For the isomers of 10c and 10d with bulky
substituents, this difference increased to values between 0.03 – 0.04 ꢄ (BP86/TZVPP).
Also for the large molecules with benzyl and isopentyl substituents, the (E)-isomer was
slightly more pyramidalized on N than the (Z)-isomer. It is interesting to note that the
pyramidalization in 10c and 10d with bulky substituents is reduced when one considers
attractive forces due to dispersion. As the BP86-D3/TZVPP results demonstrated, D
was reduced by 0.10 to 0.13 ꢄ in the sc-exo conformations (Table 2). This planarization
affects the (E)- and (Z)-isomers to the same extent. However, the situation changed
when the ap-conformations were investigated. Here, a reduced pyramidality of the N-
atom was observed (because the two bulky substituents at the two ends of the molecule
3
3
3
4
)
)
For a DFT calculation of this benzyloxy derivative, using B3LYP/6-311 þ G**, see [23b] and Table 2.
2 3
For an sp -hybridized N, D is 0 ꢄ; for an sp -hybridized N, D is 0.48 ꢄ (with C,N-bond lengths of
.45 ꢄ).
1

Helvetica Chimica Acta – Vol. 96 (2013)
819
Table 2. Relative Energies of (E)- and (Z)-Alkoxy-enamines Calculated by DFT and MP2. Method A: BP86/
TZVPP; Method B: BP86-D3/TZVPP; Method C: MP2/def2-QZVPP; Method D: B3LYP/6-311 þ G**. The
value obtained with Method D is taken from [23b]. Besides the generic (dimethylamino)-methoxy-ethene, only
the structures with the experimentally observed sc-exo- and ap-conformations (cf. [18b] and Fig. 4) of the
exocyclic bond of the pyrrolidine ring have been optimized.
Configuration
E)/(Z)
Method Energy
Pyramidalization
D [ꢄ] on N
[
kcal/mol]
(
(E)
(Z)
(E)
(Z)
A
C
0
0
þ 0.26 0.30
þ 0.25 0.36
0.28
0.34
A
B
þ 0.77
þ 2.79
0
0
0.27
0.18
0.24
0.11
A
B
D
0
0
0
þ 1.22 0.21
þ 0.95 0.02
þ 0.69
0.17
0.03
A
B
þ 2.64
þ 2.55
0
0
0.28
0.16
0.25
0.15
A
B
þ 1.90
0
0.21
0.17
0.05
0
þ 2.17 0.00
attract one another), which resulted in a significant structural distortion of each ap-
isomer. This distortion is facilitated by a phenyl ring that, in the ap-conformation, can
interact with the neighboring alkoxy substituent. However, because of the empirical
character of the D3 dispersion correction, this observation should not be over-
interpreted. In condensed phase, these interactions are likely to be reduced because of

8
20
Helvetica Chimica Acta – Vol. 96 (2013)
Fig. 8. BP86/TZVPP Structures of the (E)- and (Z)-alkoxy enamines 10b, 10c, and 10d with the
ꢁ
exocyclicꢂ bond angles C(3)ꢀC(2)ꢀC(substituent). The angles in parentheses are those obtained with
BP86-D3/TZVPP.

Helvetica Chimica Acta – Vol. 96 (2013)
821
additional interactions that become possible with solvent molecules. Still, it clearly
demonstrates that dispersion interactions of the bulky substituents can be more
pronounced in the ap-conformer when compared to the sc-exo conformer.
In conclusion, the N-pyramidalization of enamines, found in X-ray crystal
structures and discussed by Eschenmoser, Dunitz, and our groups [18b][25h – l], was
confirmed by the calculations. The direction of the pyramidalization was such that
electrophilic attack from the face anti to the large substituent in 10a would also be
stereoelectronically favored.
Another interesting observation is evident from an inspection of the ꢁexocyclicꢂ
bond angles in Fig. 8: in all cases, the (diphenyl)(trimethylsilyloxy)methyl group is in a
quasi-axial position on the puckered pyrrolidine ring, with a C(3)ꢀC(2)ꢀC(exo) bond
angle, which is substantially larger than the tetrahedral angle of 1098, i.e. 114.7 – 115.88
in the sc-conformations and 112.1 – 113.18 in the ap-conformations of 10c and 10d. This
leads to a closer proximity of the large substituent with the enamino double bond
(
increasing steric shielding for attack of an electrophile from the cis-face of the double
bond?).
We next discuss the relative energies of (Z)- and (E)-alkoxy-enamines 10 (see
Table 2). When we compared the electronic energies of the generic Me-substituted
compounds 10b, for which MP2 calculations with a sufficiently large basis set are
feasible, we found a rather small difference of 0.26 (BP86/TZVPP) and 0.25 kcal/mol
(
(
MP2/def 2-QZVPP), respectively, with the (E)-isomer being more stable than the
Z)-isomer. Importantly, the BP86 result deviated only by 0.01 kcal/mol from the MP2
energy difference, which is an indication that the BP86 functional can be taken as a
sufficiently accurate model for the description of the large derivatives.
However, two effects make the energetic comparison of the large derivatives more
complicated. One is that dispersion interactions can play a significant role because of
the bulky substituents. The other one is the multitude of stable conformers, which is
difficult to sample in a standard quantum-chemical approach. While we have already
considered the first issue above, we need to discuss the latter one in more detail,
especially for the isopentyl derivative 10d. All structure optimizations have been
carried out starting from corresponding (E)- and (Z)-isomeric structures with the same
conformation with respect to the isopentyl group. This procedure does hardly introduce
an energetic bias owing to the conformations of the isopentyl group, although it might
slightly affect the overall energy because of different long-range intramolecular
interactions. Only a rigorous sampling approach would allow us to identify the lowest-
energy structures of each conformer, which is beyond the scope of the present work.
Still, we can extract all relevant information from the calculations carried out for the
selected conformers under study. For example, in Fig. 8 one can clearly see that the
isopentyl conformation is different for the ap- and sc-exo-isomers. To investigate the
relevance of these different conformations, we also considered an isopentyl con-
formation for the ap-conformers that resembles the one of the sc-exo conformers. The
energy of the isomers was then reduced by ca. 2.4 kcal/mol so that all relative energies
of corresponding (E)- and (Z)-isomers and especially their energetical ordering
remain essentially the same. Hence, we use the energies of the structures depicted in
Fig. 8 in the following discussion, keeping in mind that lower-energy structures might
be obtained in a rigorous sampling approach.

8
22
Helvetica Chimica Acta – Vol. 96 (2013)
The relative electronic energies are collected in Table 2. Only for the ap-
conformers, we found the same energetic ordering of (E)- and (Z)-isomers as for
the Me-substituted generic structure 10b (vide supra). For the ap-conformer of the
benzyl derivative we found that the (Z)-isomer is higher in energy than the (E)-isomer
by þ 1.22 kcal/mol (þ0.95 kcal/mol). Here and in the following paragraph, the
energy difference given first was obtained with BP86/TZVPP, while the number in
parenthesis is the BP86-D3/TZVPP result. Surprisingly, the ap-(Z)-conformer of the
isopentyl derivative 10d was lower in energy by ꢀ 1.90 kcal/mol (þ2.17 kcal/mol),
while the inclusion of dispersion interactions in the structure optimization reversed
this order. For the sc-exo-isomers, the picture is without such ambiguity. The (Z)-
isomers of 10c and 10d were lower in energy. For the benzyl derivative 10c, we found
the (Z)-isomer to be ꢀ 0.77 kcal/mol (ꢀ2.79 kcal/mol) lower in energy than the (E)-
isomer, while it was ꢀ 2.64 kcal/mol (ꢀ2.55 kcal/mol) for the isopentyl derivative
10d.
In general, we found a comparatively small energetic effect of dispersion
interactions on the relative energies of (E)- and (Z)-isomers (with the sole exception
of the isopentyl ap-structure), which can be taken as an indication that the dispersive
attraction of the bulky substituents in both isomers is about the same, and hence does
not affect their energy difference.
It is noteworthy that we found the ap-(E)-conformers to be lower in energy than the
sc-exo-(E)-conformers of 10c and 10d, although the latter was usually found in X-ray
structures. For the isopentyl (E)-isomers, ap-conformer was lower in energy by
0.56 kcal/mol (BP86/TZVPP) than sc-exo-conformer, while the difference was even
larger in the case of the benzyl derivative, i.e., by 3.04 kcal/mol (BP86/TZVPP).
The general conclusion from these calculations is that the energy differences
between (E)- and (Z)-isomers of the isopentyloxy-diphenylprolinol-silyl ether
derivative 10d are quite larger (1.9 – 2.6 kcal/mol) than experimentally observed (ca.
1
:1; remember that a 10 :1 ratio corresponds to an energy difference of 1.4 kcal/mol at
room temperature). To understand this discrepancy, two explanations can be offered.
First of all, the accuracy of DFT could be limited, and a more accurate wave-function-
based calculation is desirable. To investigate this issue, we are currently optimizing the
structures with MP2/def 2-QZVPP. Moreover, the bulky substituents will interact with
solvent molecules in condensed phase, and this poses a rather complicated problem, in
which the moleculeꢂs solvent environment and its conformations need to be rigorously
sampled.
2
.2. The Dihydro-1,2-oxazine N-Oxides 5. The organocatalytic addition of
2
aldehydes to disubstituted nitro olefins, such as 1-phenyl-2-nitroprop-1-ene (3, R ¼
3
Ph, R ¼ Me; ꢁmethyl-nitrostyreneꢂ) or 1-nitrocyclopentene and 1-nitrocyclohexene,
with the catalyst 1 or other pyrrolidine derivatives has been found to be sluggish;
these reactions required a more acidic co-catalysts, CHCl or CH Cl instead of
3
2
2
toluene as solvent, and long reaction times. Some typical conditions, under which
this type of Michael addition can be carried out, are compiled in Table 3. Many authors
considered the disubstituted nitro olefins sterically hindered or demanding sub-
strates. This is in contrast to the results described in 1970 by Nielsen and Archibald
(see ref. in [8]), who found that a-substituted nitro olefins add to cycloalkanone-
derived, i.e., disubstituted, enamines to form isolable 5,6-dihydro-4H-oxazine 1,2-

Helvetica Chimica Acta – Vol. 96 (2013)
823
Table 3. Reported Additions of Aldehydes to Disubstituted Nitro Olefins 3, Catalyzed by Pyrrolidine Derivatives
Conditions
Nitro olefin 3
Ref.
Comments
5
mol-% 1a, 5 mol-% 4NO ꢀC H OH,
no reaction with
This work
2
6
4
toluene, r.t., up to 48 h
2
i
3
or with 3 R ¼ Pr, R ¼ Me
2
1
1
0 mol-% ent-1a, 60 mol-% PhCO H, H O, r.t.
no reaction with
[27]
[28]
2
2
0 mol-% 1a, 10 mol-% PhCO H, CH Cl , r.t., up to 48 h
reactions with
2
2
2
0 mol-% 1a, 40 mol-% 4-NO ꢀC H OH, CHCl , r.t., up to 65 h reactions with
,
2
6
4
3
1
i
not with R ¼ Pr
[12]
2
0 mol-% 1a, 10 mol-% 3-NO ꢀC H CO H, CH Cl ,
reactions with
[29a]
2
6
4
2
2
2
r.t., up to 54 h
1
1
5 mol-% proline, Na SO , DMSO, r.t., 90 min
reaction with
reaction with
[29b]
[30]
[31]
2
4
5 mol-% pyrrolidyl-pyrrolidine, CHCl , ꢀ 108, 7 d
3
H-Pro-Pro-d-Gln-OH or -Asn-OH (5 mol-%),
CHCl / PrOH 1 : 9, r.t., up to 3 d
reaction with
i
3
and nitrocyclohexene,
1
i
works with R ¼ Pr
35
oxides ) under conditions (hexane, 0 – 258), which do not indicate that the reaction
would be sluggish!
35)
These authors also make the statement: ꢁit has been found that a-substituted nitro olefins with
enamines from cycloalkanones of less than eight ring members, in polar or nonpolar aprotic solvents
(
hexane, MeCN, ether), lead to a zwitterion which undergoes intramolecular O-alkylation to a cyclic
nitro esterꢂ. The intermediacy of a zwitterion in such reactions has been questioned by a recent
calculation (see Footnote 4 and [12]), suggesting that oxazine derivatives, in contrast to
cyclobutanes, may be formed by a concerted (4 þ 2) cycloaddition of a nitro olefin to an enamine
(
see also Scheme 2).

8
24
Helvetica Chimica Acta – Vol. 96 (2013)
2
.2.1. Preparation and Identification of Oxazine Derivatives 5. We have now
studied the stoichiometric reactions of various 1a-derived enamines 2 (R Si ¼ Me Si)
3
3
2
3
with methyl-nitrostyrene (3, R ¼ Ph, R ¼ Me), 4-methyl-2-nitropent-2-ene (3,
2
i
3
R ¼ Pr, R ¼ Me), ethyl 3-nitrobut-2-enoate, and nitrocyclohexene. The reactions
were carried out as with the monosubstituted nitroethenes (cf. Scheme 4): to the
enamine solution in C D (NMR tube, in the presence of 4-ꢄ MS, room tempera-
6
6
ture) was added an equimolar amount of the nitro olefin, and the first spectrum
was recorded within ca. 10 min. The results are collected in Table 4. In all cases,
the disubstituted nitro olefins gave (4 þ 2) cycloadducts 5 (see formulae in Fig. 2 and
X-ray structures in Fig. 4). Compounds 5 were formed upon mixing of the compo-
i
1
i
nents (i.e., 5b, 5c, and 5h), except when the Pr-substituted enamine 2 (R ¼ Pr)
was employed (5d, 5e, and 5f took hours or days for high degrees of conversions).
To see whether the cycloadducts are in equilibrium with the precursors, we warmed
the solutions to 508, cooled them back down to ambient temperatures, and deter-
mined the ratios oxazine derivative 5/corresponding enamine 2 by suitable NMR-
i
peak integrations: equilibration could be detected with the Pr/Ph/Me-substi-
tuted derivative 5d and the bicyclic compound 5f (derived from nitrocyclo-
hexene).
Table 4. Cycloadditions of 1a-Derived Enamines 2 to Disubstituted Nitro Olefins 3 with Formation of Oxazine
Derivatives 5
5
R1 R2
R3 Yield [%]
Time
Comments
Ref. to corresp.
catal. reaction
5
5
5
b
Me Ph
nPr Ph
iPr Ph
Me > 90 (NMR),
ꢁ 10 min most stable of our compounds 5; [12]
no equilibration up to 508
ꢁ 10 min stable up to 508; no equilibration [28][31]
7
2 (recryst.)
Me > 90 (NMR),
3 (recryst.)
Me > 85 (NMR),
8 (recryst.)
c
6
d
2 h
highly moisture-sensitive
[31]
1
i
6
equilibrium 5d/2 (R ¼ Pr)
: 1 (r.t.), 1.5 : 1 (508)
slowest reaction
7
5
5
e
iPr iPr
Me 44 (after prep. TLC)
4 d
–
h
Pr CO Et Me > 90 (NMR)
ꢁ 10 min no equilibration; rearrange-
ment to nitro enamine 6s
within 3 d
[28]
2
iPr
ꢀ(CH ) ꢀ 50 (NMR)
16 h
equilibrium 5f/2 (R ¼ Pr)
: 1 (r.t.), 1 : 3 (508)
1
i
[29a]
5
f
2
4
1

Helvetica Chimica Acta – Vol. 96 (2013)
825
2
.2.2. Chemical Properties of the Oxazine Derivatives 5. To learn about the role,
which the (4 þ 2) cycloadducts 5 might play in catalytic reactions, in comparison with
the (2 þ 2) cycloadducts 4, we have investigated their properties, such as thermal
stability (Table 4), hydrolytic stability, rearrangement, and formation from open-chain
precursors (Scheme 10,a–f). As reported previously [1] (cf. Scheme 6), only the six-
membered-ring compound 5a, derived from a monosubstituted nitroethene, was shown
to be in equilibrium with the corresponding cyclobutane 4a (Scheme 10,a). In two cases,
there was reversible dissociation of the oxazine derivative (5d and 5f;b in Scheme 10).
Non-hydrolytic ring opening to a linear nitro enamine, 6s, and its reversal, 6r! 5b, as
well as hydrolyses to nitro aldehydes were observed or studied with the ethyl-ester
derivative 5h and the heterocycles 5b/d, as shown in Scheme 10,d and e, respectively. A
i
most peculiar derivative was the oxazine N-oxide 5d with 5-Pr, 3-Me, and 4-Ph
substitution: it was very sensitive to moisture, but could not be converted to a linear
derivative (nitro enamine or aldehyde) under the conditions tested by us, rather it fell
apart to the nitro olefin, aldehyde, and pyrrolidine 1a (the catalyst molecule; see
36
Scheme 10,f) ).
As with the cyclobutanes, there is a discrepancy between the ease, in most cases, of
cycloadduct formation and the failure of the catalytic reaction also with the oxazine
derivatives (cf. Tables 3 and 4). The successful catalysis with disubstituted nitro
olefins listed in Table 3 required stronger acidic conditions and, most remarkably,
chlorinated solvents (CH Cl and CHCl ), which are known to be solvents of choice
2
2
3
for many organic reactions involving cationic intermediates, notably iminium ions
37
[
32] ).
Referring to the prototypical reaction of propanal with the methylated nitrostyrene
2
3
(
3; R ¼ Ph, R ¼ Me), the experimental results presented in Tables 3 and 4, and
Scheme 10 are as follows: i) there was no catalytic reaction under our standard conditions
[
14] (5 mol-% 1a, 5 mol-% 4-nitrophenol, toluene, room temperature, 24 h); ii) ꢂ 68%
yield was obtained under the conditions of Pihko and co-workers for the catalytic
reaction [12] (10 mol-% 1a, 40 mol-% 4-nitrophenol, CHCl , room temperature, 30 h);
3
1
iii) upon mixing the corresponding enamine 2 (R ¼ Me) with the disubstituted nitro
olefin in toluene at room temperature, the oxazine derivative 5b was formed in ca. 5 min;
iv) the heterocycle 5b was stable in C D up to at least 508; v) with 10 equiv. H O and
6
6
2
10 mol-% 4-nitrophenol, this compound was stable for 24 h in benzene; vi) switching to
CH Cl as solvent, 10 equiv. of H O and the stronger acid PhCO H (1.1 equiv.; conditions
2
2
2
2
of Ma and co-workers for the catalytic reaction [28]) led to hydrolysis of 5b to the nitro
aldehyde with complete conversion in 24 h (Scheme 10,e); vii) deuterolysis of 5b (4-
NO C H OD, in CDCl ) furnished g-nitro aldehyde with > 50% D in the a- and < 10%
2
6
4
3
36)
37)
As far as we know, there has been only one report on an organocatalytic Michael addition of 3-
methylbutanal to disubstituted nitro olefins (by one of our groups (H. W. [31])).
An example relevant to the discussion of this subject is the reaction of nitrostyrene with the 1a-
1
derived enamine 2, R ¼ Me, R Si ¼ Me Si, to give the nitro enamine 6a: it took 15 h at room
3
3
temperature in toluene and ꢁ 5 min in CD Cl ; see Scheme 6, and also Footnote 12 and Scheme 6 in
2
2
[
1].

8
26
Helvetica Chimica Acta – Vol. 96 (2013)
Scheme 10. Conversions and Formations of Oxazin N-Oxide Derivatives 5
a) Equilibrium of 5a with 4a (C D , room temperature; Scheme 6) [1][12][17]. b) Equilibrium of
6
6
2
enamine and nitro olefin with 5 (see Table 3); for similar results with R ¼ 4-FꢀC H , see [12]. c) Ring
6
4
opening of 5h to nitro enamine 6s (C D , 3-ꢄ MS, room temperature, 3 days, > 99% conversion, 96%
6
6
yield; dr 95 : 5) þ 4% nitro aldehyde. d) Ring closure of nitro enamine 6r to 5b ((D )toluene, 40 mol-%
8
4
-NO ꢀC H OH, 3-ꢄ MS, room temperature, 24 h; 30% conversion); for a similar experiment, see
2
6
4
Supplementary Material in [12]. e) Hydrolysis of 5b (10 equiv. H O, C D , room temperature, 24 h: no
2
6
6
reaction; after addition of 10 mol-% 4-(NO )C H OH, r.t., 24 h: no reaction; with 10 equiv. H O,
2
6
4
2
1
.1 equiv. PhCO H, CD Cl , room temperature, 24 h: > 98% conversion to nitro aldehyde 7r of dr
2 2 2
6
6 :24 :8 :2); hydrolysis of 5c (10 equiv. H O, 1.1 equiv. PhCO H, CH Cl , room temperature, 24 h: 67%
2
2
2
2
nitro aldehyde after chromatography); deuterolysis of 5b (4 equiv. 4-NO ꢀC H OD, 1.5 equiv. D O,
2
6
4
2
CDCl , room temperature, 17 h) gives 41% g-nitro aldehyde with > 50% D in a- and < 10% in g-
3
position. f) Products formed from moisture-sensitive 5d in C D at room temperature on contact with air:
6
6
the oxazine derivative disassembled to its precursors; similarly, treatment of 5d with 10 equiv. of H O,
2
1
equiv. of PhCO H, CDCl , room temperature, leads to total destruction; no nitro aldehyde detected in
2 3
the product mixture. g) Conversion of a nitro aldehyde 7r (R ¼ Me) to nitro enamine 6r (7r/1a 1:1; 3-ꢄ
MS, (D )toluene, room temperature, ꢁ 10 min: > 90% conversion to 6r of high diastereoisomeric
8
purity); for a similar experiment with the nitro aldehyde lacking the Me group next to NO , see [17a].
2

Helvetica Chimica Acta – Vol. 96 (2013)
827
38
D in the g-position ); viii) treatment of the Me/Ph/Me-substituted nitro enamine 6r with
-nitrophenol (anhydrous conditions in toluene) caused cyclization to the oxazine
4
derivative 5b (Scheme 10,d; in an equilibration?).
Thus, at least in this case, it must be concluded for the catalytic reaction that the
plethora of conditions listed in Table 3 have, actually, not been elaborated to force the
intermediate enamine to add to the disubstituted nitro olefin, but to hydrolyze the
intermediate oxazine N-oxide derivative to the product nitro aldehyde. In general
terms, the exact route of formation of the (4 þ 2) cycloadducts 5 (zwitterion or
4
concerted one-step reaction )) and of their hydrolysis to nitro aldehydes (site of
5
38
protonation, direct acid-catalyzed hydrolysis ) ), or zwitterion) still need to be
discussed (see D ! F vs. D ! G ! F in Scheme 2).
3
. Intermediates 13 and 14 in Tripeptide-Catalyzed Michael Additions. – One of our
groups (H. W.) reported that tripeptides of the general type H-Pro-Pro-Xaa-X (with
Xaa bearing a CO H group; e.g., 11a in Fig. 9) are excellent catalysts (with loading of
2
ꢁ
1 mol-%) for enantioselective additions of aldehydes to monosubstituted nitro
39
olefins [37] ). Two such tripeptides have recently been found to catalyze aldehyde
addition to disubstituted nitro olefins [31]; the reactions were much slower and
required more catalyst (5 mol-%) than with monosubstituted nitro olefins. The
important role of the intramolecularly available CO H group was demonstrated by
2
employing corresponding methyl esters, which were much less active [19][37d]. Thus,
trapping of short-lived species, such as zwitterions, or a kind of intramolecular acid-
catalyzed hydrolysis of intermediates, such as cyclobutanes or oxazine derivatives, was
proposed to explain the catalytic activity of these peptidic carboxylic acids.
38)
Pihko and co-workers have recently proposed [12], based on DFT calculations, that the heterocycle
5
was protonated at C(3) stereoselectively, as indicated in i (cf. D ! F in Scheme 2); compounds 5
are cyclic nitronic acid esters, and, at the same time N/O-acetal derivatives. The same type of
reactivity, i.e., attack by electrophiles, was observed in the fluoride-catalyzed additions of silyl-
nitronates to aldehydes (see ii). The opposite reactivity, i.e., attack by nucleophiles on nitronate C-
atoms, was observed with organolithium compounds (cf. iii), in the Nef reaction (cf. iv), and with O-
silylated nitronate esters (cf. v), which are kind of iminium ions. The result of the deuterolysis
experiment (Scheme 10,e) is not compatible with protonation at C(3) of 5b.
39)
Simple oligoprolines have also been tested for, e.g., H-(Pro) -OH, H-(Pro) -OBn, H-
2
,3,4
3
3
(
Pro) NHBn, and H-(b h-Pro) -OH. They are quite active catalysts (short reaction time, good-to-
3 2
excellent diastereoselectivity), but the enantioselectivities, determined in some selected examples,
were poor: hitherto unpublished results (N. P., ETH Zꢀrich, 2012) and Universitꢅt Basel, 2009 [38].

8
28
Helvetica Chimica Acta – Vol. 96 (2013)
Fig. 9. Tripeptide derivatives 11, used as organocatalysts for Michael additions to mono- and disubstituted
nitro olefins [19][31][37][38], and/or identified or isolated enamines, as well as (2 þ 2) and (4 þ 2)
adducts 12 – 14, formed with 11c. The configuration of the cyclobutane stereogenic center carrying NO
2
and Me in 14c could not be safely deduced from the NMR spectra; it is tentatively assigned as (R), as
expected from the collapse of the corresponding zwitterion. The amido-carboxylic acid derivatives of
type 11a and analogs containing Asp or Asn residues, as well as epimers with reversed configuration of
one of the residues were also used or tested as catalysts. For disubstituted nitro olefins, H-Pro-Pro-d-Gln-
OH and H-Pro-Pro-Asn-OH turned out to give best results in catalytic reactions [31].
We have now looked for stable intermediates in stoichiometric reactions. To
prevent the action of CO H groups and to have better solubility in organic solvents, we
2
used compounds 11b and 11c with a methyl ester group in the side chain and a terminal
40
N-dodecylcarbamoyl group [19][37g] ) (Fig. 9). As aldehyde components, we chose
propanal and 3-methylbutanal, from which the enamines 12 formed immediately with
1
1c under our standard conditions (C D , 4-ꢄ MS, room temperature). The solution of
6 6
i
the Pr-substituted enamine 12b was combined with 4,4-dimethyl-1-nitrobut-1-ene, i.e.,
the Bu-substituted nitroethene, which had led to the most stable cyclobutane, 4q, with
t
the 1a-derived enamine. The corresponding cyclobutane 13 was indeed formed
40)
Experiments with 11b, which would have been better candidates for crystallizations of
intermediates, were not successful.