2-Lithio-3,3-dimethyl-2-oxazolinyloxirane: Carbanion or azaenolate? Structure, configurational stability, and stereodynamics in solution

Article abstract of DOI:10.1021/jo801646e

(Figure Presented) Chemical studies have shown that, in ethereal solvents, oxiranyllithium Li-1 is configurationally unstable either on the macroscopic (Hoffmann's text) or microscopic time scale at low temperature (175 K). An optically pure sample of oxazolinyloxirane 1, once deprotonated, racemized within 1 min at -130°C in THF/Et₂O (3:2) (t_1/2 = 6.05 s); the application of the Eyring equation suggested a barrier to inversion for Li-1 of 8.8 kcal/mol at -130°C. Despite this, Li-1 exhibited an unusual thermal stability undergoing a successfully deuterium incorporation (>98%) also at 25°C with a little decomposition. The structure, configurational stability, and stereodynamics in solution of α-lithiated oxazolinyloxirane Li-1 have been also synergically investigated by means of in situ IR and NMR spectroscopy. IR spectroscopic studies showed that lithiation of 1 is complete at -98°C within 1 min and is accompanied by a decrease of the C=N wavenumber by only 60 cm^-1, so supporting the idea that the structure of Li-1 may be more similar to that of an organolithium rather than an azaenolate . In addition to this, multinuclear magnetic resonance studies suggested that at least in a range of concentration of 0.08-0.3 M, Li-1 mainly exists in THF as a monomelic η³-aza-allyl coordinated species rapidly equilibrating, on the NMR time scale, with a complex mixture of diastereomeric oxazoline-bridged dimeric species variously intraaggregated. An exchange mechanism by which monomers would interchange their Li atoms via one of the above dimeric species and which may be responsible for the fast racemization Li-1 undergoes as soon as is generated has been proposed.

Full text of DOI:10.1021/jo801646e

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane: Carbanion or Azaenolate?
Structure, Configurational Stability, and Stereodynamics in Solution
Vito Capriati,* Saverio Florio,* Renzo Luisi, Filippo Maria Perna, and Agnese Spina
Dipartimento Farmaco-Chimico, UniVersita` di Bari, Consorzio InteruniVersitario Nazionale Metodologie e
Processi InnoVatiVi di Sintesi C.I.N.M.P.I.S., Via E. Orabona 4, I-70125 Bari, Italy
capriati@farmchim.uniba.it; florio@farmchim.uniba.it
ReceiVed July 24, 2008
Chemical studies have shown that, in ethereal solvents, oxiranyllithium Li-1 is configurationally unstable
either on the macroscopic (Hoffmann’s text) or microscopic time scale at low temperature (175 K). An optically
pure sample of oxazolinyloxirane 1, once deprotonated, racemized within 1 min at -130 °C in THF/Et2O
(3:2) (t1/2 ) 6.05 s); the application of the Eyring equation suggested a barrier to inversion for Li-1 of 8.8
kcal/mol at -130 °C. Despite this, Li-1 exhibited an unusual thermal stability undergoing a successfully
deuterium incorporation (>98%) also at 25 °C with a little decomposition. The structure, configurational
stability, and stereodynamics in solution of R-lithiated oxazolinyloxirane Li-1 have been also synergically
investigated by means of in situ IR and NMR spectroscopy. IR spectroscopic studies showed that lithiation of
1 is complete at -98 °C within 1 min and is accompanied by a decrease of the CdN wavenumber by only
60 cm-1, so supporting the idea that the structure of Li-1 may be more similar to that of an “organolithium”
rather than an “azaenolate”. In addition to this, multinuclear magnetic resonance studies suggested that at
least in a range of concentration of 0.08-0.3 M, Li-1 mainly exists in THF as a monomeric η3-aza-allyl
coordinated species rapidly equilibrating, on the NMR time scale, with a complex mixture of diastereomeric
oxazoline-bridged dimeric species variously intraaggregated. An exchange mechanism by which monomers
would interchange their Li atoms via one of the above dimeric species and which may be responsible for the
fast racemization Li-1 undergoes as soon as is generated has been proposed.
9552 J. Org. Chem. 2008, 73, 9552–9564
10.1021/jo801646e CCC: $40.75 2008 American Chemical Society
Published on Web 11/12/2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
Introduction
controversial.1j,8 Generally, placing the anionic center in a small
ring or coordinating with lithium gives an inversion rate low
enough for the process to be observable by dynamic nuclear
magnetic resonance spectroscopy.9 However, when carbanions
are “stabilized” by π-electron acceptor groups, the stereochem-
ical outcome of their formation and protonation is strongly
influenced by the solvent and the cation. For example, the
potassium and cesium salts of 7-phenylnorbornyl anion are
consistent with a symmetrical planar benzylic carbanion,
whereas the same anion with Li+ is pyramidal; the hypothesis
drawn by the authors concerning a possible lithium-induced
pyramidalization is amazing as well. 10 The first enantioenriched
metalated cyclopropyl nitrile possessing macroscopic configu-
rational stability has been also recently described.11 Our concern
in the oxiranyllithium field led us to ascertain that some
“stabilized” lithiated oxiranes, such as R-lithiated styrene
oxides12,1g and R-lithiated oxazolinyloxiranes,2b,13 do exhibit a
singular “carbanionic/carbenoid switch” and a different thermal
and configurational stability according to the oxiranyl skeleton,
the employed experimental conditions, and the nature of the
electrophile they react with.12c R-Lithiated oxazolinyloxiranes,
which react smoothly with many electrophiles, are known to
be “configurationally unstable” on the macroscopic time scale
at low temperature.2b,13a,b In this paper, we wish to report the
results of a combination of a multinuclear magnetic resonance
study with an in situ IR investigation to elucidate the nature,
the structure, and the stereodynamics, as well as the reasons
affecting the configurational instability, of an R-lithiated ox-
azolinyloxirane such as Li-1 employed as a model system
(Scheme 1).
Oxiranyllithiums are growingly recognized as being extremely
versatile synthetic intermediates in organic synthesis, and their
reactivity has been widely reviewed and continues to be
updated.1 However, the lack of structural knowledge of these
lithiated oxiranes is in sharp contrast to their widespread use.
In fact, until recently, just a few computational studies have
been published on these intermediates,2 a few detailed rate
studies of LiTMP- and LDA-mediated lithiation of epoxides
are available and have been reported by Collum’s group,3 and
to the best of our knowledge, only one example of a NMR
investigation concerning a R-lithiated triphenylsilyl-substituted
oxirane has been mentioned in a thesis dissertation.4 Moreover,
no single-crystal X-ray structures, which might have provided
hints at least concerning their static structure, have ever been
published,5 and just an X-ray structure determination concerning
a lithiated cyclopropane, such as 1-cyano-2,2-dimethylcyclo-
propyllithium, which would be the closest compound to a
lithiated oxirane, has been reported by Boche.6 One reason may
be that, in many cases, these oxiranyllithiums behave really as
“fleeting intermediates” with a lifetime too short to preclude
all long-term NMR spectroscopic investigations.1f In addition,
the well-known carbenoid nature of R-oxygen-substituted or-
ganolithium compounds,7 which manifest with the “spontane-
ous” tendency they have to give “reductive alkylation” and/or
“homocoupling” reactions, needs a careful choice of the
experimental conditions, which must be tailored to the properly
substituted epoxides under investigation in order to minimize
their electrophilic character. In fact, sometimes the best
optimized conditions described in the literature to successfully
generate these anions and trap them with electrophiles make
use of very low temperatures (e-78 °C) and of nonpolar
solvents (such as toluene, hexane);1h other times, the employ-
ment of polar solvents (such as THF, Et2O) is the winning
choice.1g The “need” of using ligands (such as TMEDA) to
increase the lifetime of these reactive species in solution is also
Results and Discussion
Chemical Studies. Lithiated oxazolinyloxirane Li-1 could
be successfully generated by treating 1 with 1.2 equiv of s-BuLi
in a donor solvent, such as THF, at -78 °C in the presence or
absence of TMEDA (Scheme 1). At this temperature, oxira-
nyllithium Li-1 proved to be thermally stable over a period of
several hours; indeed, it could be quenched with CH3OD as
1-D, almost quantitatively and without decomposition, even after
3 h (95% D, >98% yield).14 The thermal stability of Li-1 slowly
decreased with time upon warming up the reaction mixture in
THF to room temperature, the ketone 3 being the only product
deriving from a sort of isomerization reaction probably involving
(1) (a) Satoh, T. Chem. ReV. 1996, 96, 3303–3325. (b) Mori, Y. ReV.
Heteroatom. Chem. 1997, 17, 183–211. (c) Hodgson, D. M.; Gras, E. Synthesis
2002, 12, 1625–1642. (d) For a special issue on oxiranyl and aziridinyl anions,
see: Florio, S. Tetrahedron 2003, 59, 9693–9684. (e) Hodgson, D. M.; Gras, E.
In Topics in Organometallic Chemistry; Hodgson, D. M., Ed.; Springer: Berlin,
2003, p 217. (f) Chemla, F.; Vranchen, E. In The Chemistry of Organolithium
Compounds; Rappoport, Z., Marek, I., Eds.; John Wiley & Sons: New York,
2004; Vol. 2, Chapter 18, p 1165. (g) For a recent account on R-lithiated
aryloxiranes, see: Capriati, V.; Florio, S.; Luisi, R. Synlett 2005, 9, 1359–1369.
(h) Hodgson, D. M.; Bray, C. D. In Aziridine and Epoxides in Organic Synthesis;
Yudin, A. K., Ed.; Wiley-VCH Verlag GmbH & Co. KgaA: Weinheim, 2006;
Chapter 5, p 145. (i) Padwa, A.; Murphree, S. ARKIVOC 2006, (iii), 6–33. (j)
For a recent review on R-lithiated-R-substituted oxiranes, see: Capriati, V.; Florio,
S.; Luisi, R. Chem. ReV. 2008, 108, 1918–1942.
(2) (a) Boche, G.; Bosold, F. B.; Lohrenz, J. C. W.; Opel, A.; Zulauf, P.
Chem. Ber. 1993, 126, 1873–1885. (b) Abbotto, A.; Capriati, V.; Degennaro,
L.; Florio, S.; Luisi, R.; Pierrot, M.; Salomone, A. J. Org. Chem. 2001, 66,
3049–3058. (c) Pratt, L. M.; Ramachandran, B. J. Org. Chem. 2005, 70, 7238–
7242.
(3) (a) Wiedemann, S. H.; Ram´ırez, A.; Collum, D. B. J. Am. Chem. Soc.
2003, 125, 15893–15901. (b) Ma, Y.; Collum, D. B. J. Am. Chem. Soc. 2007,
129, 14818–14825.
(8) For an enlightening account about the role TMEDA may have in
organolithium chemistry, see: Collum, D. Acc. Chem. Res. 1992, 25, 448–454.
(9) For the structure and dynamics of 1-lithio-1-phenylcyclopropanes, see:
Hoell, D.; Schnieders, C.; Mu¨llen, K. Angew. Chem., Int. Ed. 1983, 22, 243–
244.
(10) People, P. R.; Grutzner, J. B. J. Am. Chem. Soc. 1980, 102, 4709–4715.
(11) Carlier, P. R.; Zhang, Y. Org. Lett. 2007, 9, 1319–1322.
(12) (a) Capriati, V.; Florio, S.; Luisi, R.; Salomone, A. Org. Lett. 2002, 4,
2445–2448. (b) Capriati, V.; Florio, S.; Luisi, R.; Nuzzo, I. J. Org. Chem. 2004,
69, 3330–3335. (c) Capriati, V.; Florio, S.; Luisi, R.; Perna, F. M.; Barluenga,
J. J. Org. Chem. 2005, 70, 5852–5858. (d) Capriati, V.; Florio, S.; Luisi, R.;
Salomone, A.; Cuocci, C. Org. Lett. 2006, 8, 3923–3926. (e) Capriati, V.;
Degennaro, L.; Florio, S.; Luisi, R.; Punzi, P. Org. Lett. 2006, 8, 4803–4806.
(13) (a) Luisi, R.; Capriati, V.; Carlucci, C.; Degennaro, L.; Florio, S.
Tetrahedron 2003, 59, 9707–9712. (b) Luisi, R.; Capriati, V.; Degennaro, L.;
Florio, S. Org. Lett. 2003, 5, 2723–2726. (c) Florio, S.; Perna, F. M.; Luisi, R.;
Barluenga, J.; Fan˜ana´s, F. J.; Rodr´ıguez, F. J. Org. Chem. 2004, 69, 9204–
9207.
(4) Opel, A. Dissertation, Universita¨t Marburg, 1993.
(5) Only the isolation and crystal structure of a presumed intermediate formed
in the reaction of an organolithium compound with an epoxide has been reported
until to date: Harder, S.; Boersma, J.; Brandsma, L.; Kanters, J. A.; Duisenberg,
J. M.; van Lenthe, J. H. Organometallics 1990, 9, 511–516.
(6) The tetrahedral configuration of the anionic carbon atom seen in the solid
state for the R-cyanocyclopropyl anion resulted to be in agreement with
calculations and also nicely confirmed the results of H/D exchange experiments
performed with cyclopropylnitriles in solution: Boche, G.; Harms, K.; Marsch,
M. J. Am. Chem. Soc. 1988, 110, 6925–6926.
(14) In the lithiation of oxazolinyloxirane 1, TMEDA does not seem to play
a crucial role in accelerating oxiranyllithium reaction rate as well as in improving
product yields. Indeed, as has been pointed out by Collum,8 “TMEDA appears
to manifest a highly substrate dependent affinity for lithium”, and in the case of
1, the results obtained from the described chemical studies in the presence or
absence of TMEDA were the same.
(7) Boche, G.; Lohrenz, J. C. W. Chem. ReV. 2001, 101, 697–756.
J. Org. Chem. Vol. 73, No. 24, 2008 9553

Capriati et al.
SCHEME 1
SCHEME 2
the enolate 2 (Scheme 1).15 An E1-like mechanism involving
the unusual Cꢀ-O bond cleavage (with respect to the oxazoline
moiety)16 in the rate-determining step, which would permit the
ideal alignment of the orbitals to be involved in the C-C double
bond formation of the enolate, may be operating. However, the
electrocyclic R-ring-opening mechanism, which has been claimed
by several authors1f in the case of other oxiranyllithiums, cannot
be ruled out, at least at present. It is worth noting that, upon
warming the reaction mixture containing the oxiranyllithium
Li-1 to room temperature, the above isomerization reaction takes
place very slowly; indeed, we succeeded in trapping Li-1 with
CH3OD also at 25 °C in 85% yield (>98% D) and 65% yield
(>98% D) starting from 0 (50 min as the reaction time) and
-78 °C (6 h as the whole reaction time), respectively. To the
best of our knowledge, this should be the first example of an
oxiranyllithium showing such an unusual thermal stability.17 It
might be useful to point out that oxiranyllithium Li-1 exhibited
a carbene-like reactivity mainly in a nonpolar solvent. In fact,
running the deprotonation of 1 in toluene, the carbenoid
character of Li-1 became favored with time, also at -78 °C,
and the formation of diastereomeric enediols 4 and alkene 5
(particularly using excess s-BuLi),18 as the result of a
“homocoupling”19a and a “reductive alkylation” reaction,19b,c
respectively, competes strongly.
stability of a chiral organolithium, on the time scale of its
addition to an electrophile, can be gathered, in principle, by
the Hoffmann test,20 which is based on kinetic resolution. To
this end, first, racemic oxiranyllithium Li-1 was allowed to react
at -98 °C with the racemate of N,N-dibenzylphenylalaninal 621
(Scheme 2), which generally shows good reactivity toward a
wide range of organolithium reagents. In most cases, starting
from a racemic mixture of the chiral organolithium, the
(15) For an oxiranyllithium-based synthesis of R-keto-2-oxazolines, see:
Capriati, V.; Florio, S.; Luisi, R.; Russo, V.; Salomone, A. Tetrahedron Lett.
2000, 41, 8835–8838.
(16) For other examples of Cꢀ-O bond cleavage in metalated oxiranes, see:
Kasatkin, A.; Whitby, R. J. Tetrahedron 2003, 59, 9857–9864.
(17) Generally, the presence of an electron-withdrawing group on the lithiated
carbon is able to increase the anion lifetime (ref 1j). However, the factors
responsible for tuning the reactivity of an oxiranyllithium towards a carbanion
or a carbene-type behaviour are still an open question.
(18) Performing the lithiation of 1 in toluene in the presence of an excess
s-BuLi, oxiranyllithium Li-1 completely converted into an equimolar mixture
of diastereomeric enediols 5 (dr 1:1) and alkene 6 (both not isolated), which
were the only two products that could be checked in the crude by 1H and 13C
NMR and GC-MS.
(19) (a) Lohse, P.; Loner, H.; Acklin, P.; Sternfeld, F.; Pfaltz, A. Tetrahedron
Lett. 1991, 32, 615–618. (b) Doris, E.; Dechoux, L.; Mioskowski, C. Tetrahedron
Lett. 1994, 35, 7943. (c) Doris, E.; Dechoux, L.; Mioskowski, C. Synlett 1998,
337.
(20) (a) Hirsch, R.; Hoffmann, R. W. Chem. Ber. 1992, 125, 975–982. (b)
Hoffmann, R. W.; Julius, M.; Ruhland, T.; Frenzen, G. Tetrahedron 1994, 50,
6049–6060.
(21) Reetz, T. M.; Drewes, M. W.; Schwickardi, R. Org. Synth. 1999, 76,
110–116.
Configurational Stability: Hoffmann’s Test and Estimation
of the Racemization Rate. Information about the configurational
9554 J. Org. Chem. Vol. 73, No. 24, 2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
FIGURE 1. First (a) and second experiment (b): integration of 13C NMR signals, relative to the oxazoline methylenic carbons, each peak being
representative of a single diastereomer.
asymmetric induction from the resident chirality center in 6 is
high enough to favor the formation of only two adducts.
Unfortunately, in the case of oxiranyllithium Li-1, the asym-
metric induction was poor, and a mixture of the four possible
adducts 7a-d formed (Scheme 2, 90% overall yield). The
diastereomeric ratio was accurately determined from 13C NMR
spectrum on the crude reaction mixture to be 49:19:19:13
(Figure 1a). Since none of the pairwise combinations of these
numbers may lead to 50:50, kinetic resolution would be at least
1.6:1. For this reason, there is no need to assign the relative
stereochemistry to the individual diastereomers. Separation of
the four diastereomeric adducts 7a-d by column chromatog-
raphy (see Experimental Section) confirmed their structural
assignment. In a second experiment, the racemic oxiranyllithium
Li-1 was allowed to react with the enantiomerically pure
aldehyde 6. The ratio of diastereomers, again determined by
13C NMR spectrum of the crude reaction mixture, was 50:20:
FIGURE 2. Estimation of the racemization half-life of the oxiranyl-
lithium Li-1 derived from (-)-1.
19:11 (Figure 1b). The almost identical diastereomeric ratio at
conversions of >50% in the two experiments would indicate
configurational lability of oxiranyllithium Li-1 on the time scale
set by the rate of its addition to the aldehyde, that is, within 5
min at temperatures as low as -100 °C. In other words,
racemization of oxiranyllithium Li-1, under the above condi-
tions, occurs faster than its addition to the electrophile.
SCHEME 3
With the aim of calculating the rate of racemization of Li-1
at a certain temperature, a chromatographic separation of a
racemic mixture of 1 was preliminary set up on an analytical
chiral stationary HPLC phase and successfully carried out on
the same preparative chiral stationary HPLC phase (see Ex-
perimental Section); the individual enantiomers were each
obtained with ee g 99.5%. A deprotonation (1.2 equiv s-BuLi)/
deuteration (CH3OD) experiment performed on (-)-122 in THF
at -98 °C showed that just within 10 s the deuteration was
almost quantitative (>98% D) but 1-D was recovered com-
pletely racemic. These data are consistent with the idea that
oxiranyllithium Li-1 tends to undergo a very quick racemization
soon after its generation, under the above conditions, and is
configurationally unstable also on the microscopic time scale
at -98 °C. However, when the deprotonation was carried out
with 1.5 equiv of s-BuLi at -130 °C using a mixture of 3:2
THF/Et2O as the solvent system (liquid N2-THF/Et2O 3:2 being
the cooling bath) followed by a CH3OD quench, 1-D was
recovered with 22% ee (84% D), 11% ee (85% D), and 7% ee
(95% D) after a 5, 10, and 15 s deprotonation time, respectively
(Figure 2). By plotting ln(ee/100) vs time, it was possible to
estimate the rate of racemization23 (Figure 2). This plot showed
a roughly good linear relationship (R2 ) 0.9854), and the slope,
corresponding to -krac, indicates an estimated racemization t1/2
of 6.05 s at -130 °C. Application of the Eyring equation24 to
the estimated krac for the oxiranyllithium Li-1 derived from (-)-1
indicates a barrier to inversion (∆Gqrac) of 8.8 kcal/mol at -130
°C in THF/Et2O (3:2).
The structure and the stereodynamics of oxiranyllithium Li-1
in solution were synergically investigated by means of IR and
NMR spectroscopic measurements.
IR Spectroscopic Studies. A first key feature about the
structure of an oxiranyllithium such as Li-1 that needs to be
(23) Kawabata, T.; Suzuki, H.; Nagae, Y.; Fuji, K. Angew. Chem., Int. Ed.
2000, 39, 2155–2157. See also ref11.
(22) The absolute configuration to enantiomers (-)- and (+)-1 could not be
assigned to date.
(24) Activation free energy is calculated as ∆Gqrac ) 4.576T[10.319 + log-
(t/krac)]: Eyring; H., Chem. ReV. 1935, 17, 65–77.
J. Org. Chem. Vol. 73, No. 24, 2008 9555

Capriati et al.
FIGURE 3. Progression of the reaction between oxazolinyloxirane 1 and 1 equiv of s-BuLi in the presence of 1 equiv of TMEDA monitored by
ReactIR. Spectra displayed in 3D view: overlapped regions 1600-1700 cm-1
.
FIGURE 4. Reaction profile of the lithiation of 1 to give Li-1 obtained by plotting the intensities of the iminic absorbance at 1660, 1678, and 1600
cm-1 against the time.
answered is the following: Is Li-1 in solution an azaenolate or
a true organolithium (Scheme 3)?
iminic group relative to the two equilibrium geometries and a
∆E ) 0.25 kcal/mol (see Supporting Information). Addition of
1 equiv of s-BuLi after 21 min resulted in a prompt and complete
disappearance of the above bands (within 1 min in THF and in
the presence of TMEDA; see Figure 4), whereas a new
absorption emerged at 1600 cm-1, which was assigned to the
lithiated species Li-1 (Figures 3 and 4). The fact that this new
stretching frequency of the iminic group was shifted by only
60 cm-1 to lower wavenumbers is indicative of a weakening of
the CdN bond probably caused by intramolecular coordination
of the iminic nitrogen to lithium. A shift to lower wavenumbers
of 34 cm-1 and a stretching frequency of 1607.9 cm-1 has been
also reported by Meyers et al.25 for formamidines upon
R-lithiation as a consequence of a strong complexation of the
iminic nitrogen with lithium. On the other hand, in the case of
alkylidene epoxides (such as 1-tert-butylallene oxide), a stretch-
We decided to use in situ IR spectroscopy to obtain hints
about the force constant of the oxazoline iminic moiety adjacent
to the oxiranyl carbanion. The spectra were recorded with a
ReactIR 4000, and the experiments were monitored under inert
atmosphere as follows: (1) first the THF was introduced and
then the epoxide; (2) 1 equiv of TMEDA was added, and the
reaction mixture was cooled to -98 °C; (3) 1.0 equiv of a
commercial solution of s-BuLi was introduced all at once; (4)
the reaction was finally quenched with CH3OD. As it can be
noted in Figure 3, the IR spectrum of the starting oxazoliny-
loxirane 1 displays two bands for the stretching frequency of
the oxazoline iminic group at 1660 and 1678 cm-1, due to two
different conformations of the CdN moiety in solution. This
was proved by ab initio geometry optimization of the oxazoli-
nyloxirane 1 at the B3LYP/6-31 G* level of theory, which
resulted in two local minimum-energy structures, each charac-
terized by zero imaginary frequencies with a difference of ∆υ
) 13 cm-1 for the two stretching frequencies of the oxazoline
(25) Meyers, A. I.; Rieker, W. F.; Fuentes, L. M. J. Am. Chem. Soc. 1983,
105, 2082–2083.
9556 J. Org. Chem. Vol. 73, No. 24, 2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
FIGURE 5. Progression of the lithiation reaction versus absorbance in 2D. Black: starting oxazoliniloxirane 1. Blue: after addition of s-BuLi. Red:
after quenching and warming up.
ing frequency of 1780 cm-1 was recorded instead.26 On the
basis of the above results, it is reasonable to assume, in the
first place, that imine chelation in Li-1, in solution, may be more
favored than ether chelation27 and that the structure of Li-1
should be more similar to that of an oxiranyllithium rather than
an azaenolate. After 84 acquisitions (almost 80 min), the reaction
mixture was quenched with an excess CH3OD, the cooling bath
was taken away, and the system warmed to room temperature.
The IR spectrum showed a disappearance of the 1600 cm-1
band, whereas the original 1660 and 1678 cm-1 bands came
back to a certain amount (Figure 5). The oxazolinyloxirane 1
was quantitatively recovered and was 98% D. Now, the question
is which type of intramolecular coordination may be favored.
Can the in-plane lone pair on oxazoline nitrogen efficiently
coordinate lithium? Which should be the preferred aggregation
state of Li-1 in solution, and in which way such a “configura-
tional instability” may be rationalized? To answer these ques-
tions, a multinuclear magnetic resonance investigation was
performed.
abundance at 170 K, all signals were very broad, and apparently
at least three sets of signals (one being stronger than the other
two) were detected for some carbon atoms, indicative of the
presence of more than one species in solution (Figure 6b). The
7Li NMR spectrum exhibited, within 1 ppm, besides the main
signal centered at δ 0.77, up to eight peaks, all related to
nonequivalent lithiums, either as shoulders of the previous one
at δ 0.77 or close to the signals at δ 0.44 and 0.35 (Figure 6c).
This fact supports the hypothesis that several species differently
aggregated seem really to coexist at this temperature and
concentration in solution (vide infra). Most diagnostic and
interesting for the structure and the dynamic behavior of Li-1
were in particular, in the 13C NMR spectrum, the shifts and
multiplicities of the oxazoline iminic carbon, the oxiranyl
carbanion, and the two geminal oxiranyl methyl groups. The
C9 oxazoline carbon in Li-1 consists of three signals in the
range 178.0-181.0 ppm (peaks at δ 180.2, 179.6, 178.7; see
numbering in Figure 6), the stronger and the broader one being
that falling at 179.6 ppm. This carbon experienced a low-field
shift of more than 17 ppm further to lithiation (Figure 6a,b). In
the case of ortho-lithiated phenyloxazolines, where a chelation
between the nitrogen lone pair and the lithium takes place, the
iminic carbon was downfield shifted only 10.5 and 11.8 ppm
for the monomer and the bridged dimer, respectively.27 Thus,
in our case, the greater downfield shift observed may suggest a
strong decrease of the π-electron density occurring at the CdN
bond. One possibility is that lithium occupies a η3-aza-allyl
position (Figure 6). This interpretation is supported, inter alia,
by the nonequivalence of the two oxazoline geminal methyls
C3,C4 in lithiated species, as one would be expect for a “locked”
geometry such as that. These geminal methyls, likewise to the
oxazoline iminic carbon, were downfield shifted showing a
complex pattern between 30.0 and 32.0 ppm, whereas in the
starting oxazolinyloxirane they were isochronous, falling at 28.7
ppm (Figure 6a). High level ab initio calculations performed at
the DFT level on two diastereomeric ꢀ-phenyl-substituted
R-lithiated oxazolinyloxiranes2b suggested a distorted sp3 ar-
rangement for the lithium cation-bearing carbanionic carbon
NMR Spectroscopic Studies: 13C and 7Li NMR Investigation
of 0.2 M Li-1 in THF-d8 at 170 K. As a first step, in two separate
runs, a 0.2 M sample of 1 was treated with 0.8 and 2.0 equiv
of a commercial solution of cyclohexane-free s-BuLi in THF-
d8 at 170 K. The two resulting 13C NMR spectra were identical
(apart from additional signals deriving from excess of 1 or
excess organolithium), indicating that no mixed aggregates
between Li-1 and excess s-BuLi formed. Therefore, these spectra
have to be ascribed only to the oxiranyllithium Li-1 either
because several preparations were fully reproducible or because
a spectroscopic analysis (1H, 13C NMR) of the hydrolysis
products of Li-1 (once quenched with CH3OD) showed the
exclusive presence of 1-D, so ruling out the presence of products
deriving from side reaction of the oxiranyl anion. Running a
13C NMR spectrum of Li-1 (0.2 M, 1 equiv s-BuLi) at natural
(26) Chan, T. H.; Ong, B. S. J. Org. Chem. 1978, 43, 2994–3001.
(27) Also in the case of ortho-lithiated phenyloxazolines, nitrogen chelation
has been reported to be stronger than oxygen chelation; see: Jantzi, K. L.; Guzei,
I. A.; Reich, H. J. Organometallics 2006, 70, 3375–3382.
J. Org. Chem. Vol. 73, No. 24, 2008 9557

Capriati et al.
7
FIGURE 6. (a) 13C NMR spectrum of 0.2 M 1 in THF-d8 at 170 K. (b) 13C NMR spectrum of 0.2 M Li-1 in THF-d8 at 170 K. (c) Li NMR
spectrum of 0.2 M Li-1 in THF-d8 at 170 K. All spectra were measured at natural abundance lithium and carbon. Assignments were confirmed by
DEPT 135 experiments. Cy represents cyclohexane.
atom, probably in order to allow a stabilizing intramolecular
coordination between lithium, the oxazoline nitrogen, and
π-bond, that is, a trihapto η3-aza-allyl coordination. The above
spectroscopic evidence seem therefore to fit very well compu-
tational predictions obtained from similar lithiated systems.
Upon lithiation, the oxiranyl R-carbon (C5) was downfield
shifted of about 36.5 ppm, and its signal appeared as a featurless
oxiranyl methyl groups (C1,C2) which were nonequivalent in
oxazolinyloxirane 1 (18.5 and 24.3 ppm), merged into a broad
peak in lithiated species Li-1, exhibiting an average chemical
shift value at δ 22.1 (Figure 6b); it indeed corresponds to two
overlapping broad signals at δ 22.09 and 22.11. This fact may
be consistent with a nonprivileged fixation of the lithium on
one of the two sides of the plane of the oxirane ring and also
suggests either a rapidly inverting tetrahedral or a planar
configuration of the corresponding R-lithiated oxiranyllithium,
both implicating an enantiomeric conversion; in other words,
the oxiranyl R-carbanionic carbon atom behaves as a prochiral
entity. On the basis of the above considerations, an inverting
tetrahedral configuration may be most probably involved.
Effect of Temperature. Cooling down the above 0.2 M
sample of Li-1 to 160 K, near to the freezing point of the THF
solution, produces broadening for the various carbon resonances
in the corresponding 13C NMR spectra, but we were still unable
either to decoalesce signals representative of the “individual
species” in dynamic equilibrium in solution or to obtain Li-C
coupling information that would be indicative for aggregation
states of Li-1 (Figure 7).29 Similarly, after cooling the sample,
7
lump not revealing any line splitting due to Li coupling. This
deshielding has not necessarily to be interpreted in terms of a
rehybridization of the anionic carbon atom but may be just the
consequence of a “carbenoid effect”; in other terms, as pointed
out by Boche,7 “the deshielding of the 13C carbenoid atom is
not only due to the isotropic chemical shifts σ but also due to
the dia- or paramagnetic contributions of the various localized
molecular orbitals”.28 Being R-lithiated ether “carbenoids”, the
deshielding of the lithiated 13C atoms is an indication of their
carbenoid character, that is, of their electrophilic reactivity. In
this view, the carbenoid character of R-lithiated oxazolinylox-
irane Li-1 is almost as strong as that of R-lithiated triphenyl-
sililoxirane (∆δ ) 36.9).4,7 Interestingly, the two geminal
(28) This is a well-studied effect in PhLi where the unusual downfield shift
of the ipso carbon is presumably due to the large paramagnetic term, σpara, of
the shielding constant; see: (a) Bauer, W.; Winchester, W. R.; Schleyer, P. v. R.
Organometallics 1987, 6, 2371–2739. (b) Reich, H. J.; Green, D. P.; Medina,
(29) Lowering further the temperature of a 0.2 M sample of Li-1 to 150 K
with a 3:2 THF-d8/Et2O-d10 mixture as the solvent system led to a considerable
broadening of the 13C signals with both a poor resolution and a poor signal-to-
noise ratio.
¨
M. A.; Goldenberg, W. S.; Gudmundsson, B. O.; Dykstra, R. R.; Phillips, N. H.
J. Am. Chem. Soc. 1998, 120, 7201–7210.
9558 J. Org. Chem. Vol. 73, No. 24, 2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
FIGURE 7. Variable-temperature 13C and 7Li NMR spectra of 0.2 M Li-1 in THF-d8. All spectra were measured at natural abundance lithium and
carbon.
7
the relatively unresolved Li resonances (Figure 6c) could not
be further split, and particularly in the range of 0.60-0.90 ppm,
signals continued to be strongly overlapped. On the contrary,
when the sample is warmed to 243 K (Figure 7), signals of the
various carbons sharpen and 7Li NMR resonances tend to
undergo coalescence. Also in consideration of the fact that
chemical studies suggested a barrier to racemization of just 8.8
kcal/mol at -130 °C for Li-1, we interpret the above data in
terms of possible inter/intraaggregate exchanges occurring still
fast on the 13C NMR time scale even at a temperature as low
as 150 K.29 Above 233 K, the 13C NMR signals of both
carbanionic and iminic carbons are almost coalesced most
probably because a faster interaggregate exchange is now under
way. To get more insights about the nature of the above
aggregation states, we found it instructive to perform a variable
concentration study.
concentration and temperature, the most striking difference
concerned the Li NMR spectrum (Figure 8b) in which an
7
enhancement of the resonance intensities of both two average
broad peaks at δ 0.26 and 0.45 occurred, whereas the main peak
at δ 0.77 broadened at both sides. The sample was then diluted
with 0.8 mL of THF-d8 in order to obtain a 0.08 M solution
and run again at 203 K. Lower concentrations should lead, in
principle, to slower interaggregate exchange as well as to break
down higher aggregates into smaller ones. As is evident from
Figure 8c, the corresponding 13C NMR spectrum now was
apparently more simplified, in particular within the range of
30.0-32.0 ppm and around 70.0 ppm. Moreover, what was most
notable is that, upon dilution of the solution, the broad signal
related to the carbanionic carbon (C5) at δ 93.2 resolved into a
very broad quartet of almost 1:1:1:1 intensity (7Li I 3/2) (Figure
8c). A careful examination of its shape suggested that two
overlapping 1:1:1:1 quartets (1J13C-7Li ≈ 29.0 Hz for both) may
actually be involved instead; their poor resolution may probably
be related to fast quadrupole-induced relaxation of the 7Li
Variable-Concentration Study. A 0.3 M sample of Li-1 in
THF-d8 (prepared from 1.0 equiv of oxazolinyoxirane 1 and
1.0 equiv of s-BuLi) was run at 203 K. At this higher
J. Org. Chem. Vol. 73, No. 24, 2008 9559

Capriati et al.
FIGURE 8. (a) 13C and 7Li NMR spectra of 0.6 M Li-1 in THF-d8 at 203 K. (b) 13C and 7Li NMR spectra of 0.3 M Li-1 in THF-d8 at 203 K. (c)
13C and 7Li NMR spectra of 0.08 M Li-1 in THF-d8 at 203 K. (d) The 13C NMR signal of the iminic carbon C9 of Li-1 (0.08 M, THF-d8, 203 K).
7
7
(e) The 13C{1H, Li} NMR signal of the iminic carbon C9 of Li-1 (0.08 M, THF-d8, 203 K). (f) The 13C{1H, Li} NMR signal of the lithiated
carbon C5 of Li-1 (0.08 M, THF-d8, 203 K). The insert in (f) is a Gaussian enhanced spectrum (LB ) -1). All spectra were measured at natural
abundance lithium and carbon.
nucleus.30 To be sure that the fine structure exhibited by the
peak at δ 93.2 under dilution conditions was due only to 13C,7Li
couplings, a triple-resonance experiment, 13C{1H,7Li}, was run
on the same sample: the two overlapped broad quartets collapsed
to give two overlapped broad singlets (Figure 8f). Interestingly,
having a carbon directly coupled to only one lithium atom, such
as it would occur either in a monomer intramolecularly
η3-coordinated (η3-M) or in an oxazoline-bridged dimer (D) in
which each lithium atom of a unit interacts with the π iminic
system of another unit (Scheme 4). It is worth noting that, after
7
7
the Li decoupling also affected the line width of the iminic
the sample was diluted from 0.3 to 0.08 M, in the Li NMR
signal (C9): in fact, the line width at half-height changed from
a value of ∆υ1/2 ) 68 Hz to a value of ∆υ1/2 ) 37 Hz before
and after 7Li decoupling, respectively (Figures 8d,e). This fact
is suggestive of a coupling occurring between the iminic carbon
spectrum, peaks at δ 0.26, 0.45, 0.71 underwent a decrease in
their resonance intensities with respect to the main one at δ
0.77, which was also simplified at the left side between 0.82
and 0.86 ppm (Figure 8c). These findings suggest that all of
the species related to the above peaks should be “more
aggregated” than that at δ 0.77, which was tentatively assigned
to the η3-monomer. Unfortunately, the strong overlapping
occurring between peaks in both 13C and 7Li NMR spectra made
a correlation between concentrations and integrals impractical.
Now, the question is what about the nature of these several
“more aggregated” species coexisting in equilibrium in solution
with the monomer? In consideration of the fact that variable
temperature studies suggested that intramolecular association
may be probably playing an important role in dynamic equilibria
7
and Li as well, so supporting the hypothesis that an intra/
intermolecular lithium π-coordination may be actually under
way for the aggregates of Li-1. Indeed, in Stucky’s view,31 these
organolithium π-complexes may be interpreted in terms of
“multicenter covalent interactions between the unoccupied
atomic orbitals of lithium and the appropriate occupied molec-
ular orbitals of the hydrocarbon anion”, even if ionic interactions
are thought to dominate.32 These results are suggestive of the
possible coexistence of at least two type of aggregates each
7
(30) (a) Fraenkel, G.; Subramanian, C.; Chow, A. J. Am. Chem. Soc. 1995,
117, 6300–6307. (b) Fraenkel, G.; Fraenkel, A. M.; Geckle, M. J.; Schloss, F.
J. Am. Chem. Soc. 1979, 101, 4745–4747.
under way and in order to justify the complexity of Li NMR
spectra, we interpret the broad quartet (Figure 8b) related to a
(31) Stucky, G. In Polyamine-Chelated Alkali Metal Compounds; Langer,
A. W., Ed.; American Chemical Society: Washington, DC, 1974; Adv. Chem.
Ser. No. 130, Chapter 3.
(32) Setzer, W.; Schleyer, P. v. R. AdV. Organomet. Chem. 1985, 24, 353–
451.
9560 J. Org. Chem. Vol. 73, No. 24, 2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
SCHEME 4
plausible oxazoline-bridged dimer as the average signal “rep-
resentative” of a mixture of dimers, similar to D, but differently
intraaggregated, each in a fast equilibrium with the monomeric
species. One or both aza-allyl monomeric units may, indeed,
benefit from an additional intramolecular η3-coordination to give
a mono- (MD, Scheme 4) or a bis-symmetrical (BD, Scheme
4) double-bridging arrangement with the two lithium atoms,
respectively.33 In addition, each of the above dimers may be
present in solution as a mixture of homochiral or heterochiral
diastereomers. Being 7Li nuclei homotopic in homochiral dimers
(R*,R*)-D and (R*,R*)-BD but enantiotopic in heterochiral
dimers (R*,S*)-D and (R*,S*)-BD, these aggregates should each
four other peaks on both side of the average signal at δ 22.1
(C1,C2 resonances) together with a new carbanionic carbon,
as a broad peak at δ 94.5, and other sets of iminic carbons more
downfield shifted than 180 ppm could be detected. These results
are suggestive of the possible formation of higher aggregated
oligomers generated, most probably, at the expense of the
aforementioned dimers, so supporting the hypothesis that
intermolecular as well as intramolecular associations are the key
aspects of the reactivity of Li-1. However, this new “picture”
occurring at a higher concentration was not further investigated.
Therefore, we conclude that the Li-1 species may be present,
at least in a range of concentration of 0.08-0.3 M, mainly as
a η3-M monomer in fast equilibrium, on the NMR time scale,
with a complex mixture of variously intraaggregated diastere-
omeric oxazoline-bridged dimers (D, MD and BD) (Scheme
4). However, the higher the concentration is, the more competi-
tive is the formation of higher aggregates.
7
exhibit a different Li signal. On the other hand, four distinct
7Li signals would be expected for the two homochiral and
heterochiral dimers (R*,R*)-MD and (R*,S*)-MD, as each does
have two diastereotopic 7Li nuclei. In principle, also considering
the η3-M monomer, up to nine signals may be observed in the
corresponding 7Li NMR spectrum. Actually, the 7Li NMR
spectrum of a 0.2 M sample of Li-1 consists, at 160 K, of just
nine peaks partially overlapped (Figure 7), even if the coexist-
ence of other aggregates at this higher concentration cannot be
completely discarded owing to the presence, in the 13C NMR
spectrum, of minor species. In order to better investigate this
aspect, we decided to check a sample of Li-1 at a higher
concentration than 0.3 M. Surprisingly, at a 0.6 M concentration,
Interestingly, a recent paper34 claimed an inverse dependence
1
of JC,Li on the sum of the numbers of carbanion centers and
coordinating electron-pair donating ligands, that is, a strong
1
correlation between J13C,7Li and the microsolvation of an
organolithium compound. Having detected a similar value (ca.
1
29 Hz) for the J13C,7Li coupling constant relative to the two
type of aggregates mainly present at higher dilution (the η3-
monomer and the oxazoline-bridged dimer, each having a
carbanionic carbon coupled to only one lithium), it is reasonable
that the THF-microsolvation for both may be the same.
Proposed Racemization Mechanism for Li-1. On the basis
of the above spectroscopic results, the configurational instability
experimentally ascertained for oxiranyllithium Li-1 may be
tentatively interpretated as follows. At least in a range of
concentration of 0.08-0.3 M, a multinuclear magnetic resonance
investigation suggests that oxiranyllithium Li-1 would mainly
7
both Li and 13C NMR spectra of Li-1 changed drastically
(Figure 8a). In the 7Li NMR spectrum, the previous broad peaks
at δ 0.26, 0.45, 0.71 (Figure 8b) did not increase their resonance
intensity but instead, together with other smaller ones around
0.86-0.90 ppm, collapsed in favor of two broad shoulders at δ
0.61 and 0.83 on both side of the main broad peak, whose
resonance appeared to be highfield shifted by 0.02 ppm (δ 0.75).
In the 13C NMR spectrum, more complex patterns are discernible
around the previous main average signals. In particular, at least
(34) According to an empirical expression proposed by the authors, the
possibility of determining the unknown microsolvation number for the aggrega-
tion state of an organolithium compound from 1JCLi requires knowledge not only
of its aggregation state but also of a sensitivity factor related to the nature of the
organolithium. However, for R-oxygen-substituted organolithium compounds,
sensitivity factors have not been determined: Knorr, R.; Menke, T.; Ferchland,
K.; Mehlsta¨uble, J.; Stephenson, D. S. J. A. Chem. Soc. 2008, 130, 14179–
14188.
(33) For examples of other mono and double lithium(η3-aza-allyl)-type
complexes, see: (a) Colgan, D.; Papasergio, R. I.; Raston, C. L.; White, A. H.
J. Chem. Soc., Chem. Commun. 1984, 1708–1710. (b) Hacker, R.; Schleyer,
P. V. R.; Reber, G.; Mu¨ller, G.; Brandsma, L. J. Organomet. Chem. 1986, 316,
C4–C8. The case of a dimer containing two diazallyl moieties, one of which
acting as a double bridge of two lithiums, has been reported: (c) Eisen, M. S.;
Kapon, M. J. Chem. Soc., Dalton Trans. 1994, 3507–3510.
J. Org. Chem. Vol. 73, No. 24, 2008 9561

Capriati et al.
SCHEME 5. Indirect Dynamic Interconversion between Two Lithiated η3-Aza-allyl Enantiomeric Monomers
exist in a THF solution as two η3-aza-allyl enantiomeric
monomers [(R)-η3-M and (S)-η3-M, Scheme 5] rapidly inter-
converting on the NMR time scale and each one in equilibrium
with a complex mixture of diastereomeric oxazoline-bridged
dimeric aggregates (D, MD, BD, Scheme 4). Their intercon-
version may be mediated by just one of the above detectable
dimeric intermediates through a possible inversion transition
state (TS) within which the two enantiomeric monomeric units
would interchange their Li atoms in such a way that each lithium
atom of a monomer would produce a SE2-like rear-side attack
on the carbanionic carbon atom of the other with the final result
of an inversion of configuration (Scheme 5).35 Among the
various dimeric aggregates, one may speculate that a dimer such
as D could be a good candidate in promoting the above
racemization because, owing to its minor intramolecular as-
sociation, a fewer number of bonds would need to be broken
in the transition state on the way to inversion, so that, further
to the SE2 attack, an intermolecular π-deaggregation would be
only followed by a new intramolecular η3-coordination for each
unit. Starting from a pure enantiomer, such as (R)-η3-M, a
homochiral dimer, such as (R,R)-D, would be most probably
involved in the first equilibration to give, through the transition
state TS, the enantiomer (S)-η3-M. The latter would then
equilibrate with the enantiomeric dimer (S,S)-D, which through
the inversion transition state TS′ would furnish again the
enantiomer (R)-η3-M.36
a barrier to inversion for Li-1 of 8.8 kcal/mol at -130 °C.
Notwithstanding its configurational instability, oxiranyllithium
Li-1 exhibited an unusual thermal stability; in fact, it underwent
a successful deuterium incorporation (>98% D, 85% yield) even
at 25 °C, the main side reaction being in THF a ring-opening
reaction to give the keto-oxazoline 3, in particular upon warming
the reaction mixture to room temperature. Both in situ IR and
NMR spectroscopy were employed to investigate the structure
and dynamics of Li-1 in solution. IR-spectroscopic studies
showed that lithiation of 1 is complete at -98 °C within 1 min
and is accompanied by a decrease of the CdN wavenumber by
only 60 cm-1, so supporting the idea that lithiated compound
Li-1 mainly exists in a THF solution and at low temperature
(-98 °C) not as an azaenolate but as a true organolithium. The
strong bias Li-1 has to give rise to both intramolecular as well
as intermolecular trihapto coordinations is the key point of its
reactivity. A multinuclear magnetic resonance study suggested
that several η3-aza-allyl coordinated species might be most
probably involved in dynamic equilibria in a THF solution at
170 K.37 Dilution as well as variable-temperature studies also
suggested that Li-1 would be mainly present in a THF solution,
at least in a range of concentration of 0.08-0.3 M, as a η3-
aza-allyl monomer in fast equilibrium, on the NMR time scale,
with a complex mixture of diastereomeric oxazoline-bridged
dimeric species variously intraaggregated. At higher concentra-
tion close to 0.6 M, Li-1 may organize in higher oligomeric
aggregates. An exchange mechanism, responsible for the fast
racemization Li-1 undergoes on the NMR time scale once
generated, has been proposed. Powerful NMR techniques, such
as the employment of diffusion-ordered NMR spectroscopy
(DOSY), together with ab initio calculations and the use of
labeled reagents, seem to be useful and complementary experi-
mental approaches for a complete elucidation of the solution
structure of Li-1 at different levels of concentration; results will
be reported in due course.
Conclusions
In summary, R-lithiated oxazolinyloxirane Li-1 was proved
to be configurationally unstable in ethereal solvents on either
the macroscopic or the microscopic time scale at -98 °C. Under
the conditions of the Hoffmann test, which is based on kinetic
resolution, Li-1 underwent enantiomer equilibration at -98 °C
with a rate comparable to that of its addition to N,N-diben-
zylphenylalaninal. In addition, an optically pure sample of 1,
once deprotonated, racemized soon after its generation at -98
°C in THF and within 1 min at -130 °C in THF/Et2O (3:2)
(t1/2 ) 6.05 s); the application of the Eyring equation suggested
Experimental Section
General. See Supporting Information.
In Situ IR Spectroscopy. In situ IR experiments were conducted
with a ReactIR 4000 spectrometer equipped with K6 conduit and
5/8′′ DiComp insertion probe (ASI Comp Technology, Mettler
Toledo). The data points displayed per spectrum were collected
with 45 scans every 30 s, resolution 8 cm-1. The background
spectrum was recorded in neat THF. The reaction was carried out
(35) Schlosser, M. In Organoalkali Chemistry, Organometallics in Synthesis
A Manual, 2nd ed.; Schlosser, M., Ed.; Wiley & Sons: New York 2004; p 32.
(36) It is worth noting that in the case of 6Li13CH3 and 6Li13CCH2SC6H5, an
exchange mechanism by which monomers would interchange C and Li atoms
via a nonobserved dimeric intermediate has been also proposed: Heinzer, J.;
Oth, J. F. M.; Seebach, D. HelV. Chim. Acta 1985, 68, 1848–1862.
9562 J. Org. Chem. Vol. 73, No. 24, 2008

2-Lithio-3,3-dimethyl-2-oxazolinyloxirane
as follows: the flask, fitted with a magnetic stirring bar and under
N2 atmosphere, was charged with THF (4 mL), and oxazolinylox-
irane 1 (169 mg, 1 mmol) was introduced after 11 acquisitions.
TMEDA was added after the 20th acquisition (0.150 mL, 1 mmol,
1 equiv). After 24 acquisitions, the reaction mixture was cooled to
-98 °C (liquid N2/MeOH bath), and after the 43th (21 min elapsed)
s-BuLi was added (0.77 mL, 1 mmol) at the same temperature.
The mixture was finally quenched with excess MeOD after 84
acquisition (almost 80 min as the whole reaction time) and warmed
to room temperature.
NMR Spectroscopy: General Information. NMR experiments
were performed in Wilmad tubes (5 mm) fitted with a Wilmad/
Omnifit Teflon valve assembly (OFV) and a Teflon/silicon septum.
All NMR spectra were acquired using nonspinning 5 mm samples
with deuterium field-frequency locking on a spectrometer equipped
with a direct custom built 5 mm 1H, 7Li-31P, 13C triple-resonance
probe-head including a z-gradient coil at the following frequencies:
599.944 MHz (1H), 150.856 MHz (13C), and 233.161 MHz (7Li).
13C NMR spectra were referenced internally to the C-O carbon of
THF-d8 (δ 67.45). Exponential multiplication (LB) of 2-6 Hz was
applied to 13C spectra. Quantitative 13C NMR measurements were
performed according to the inverse gated 1H-decoupling technique38
with the composite pulse decoupling applied only during the short
acquisition time and not during the delay (d1 ) 10 s), which was
at least 10 times the acquisition time. The typical pulse widths for
90° pulses (µs) and attenuation levels (dB, in brackets) were the
following: 13C, 12.50 (3.50). 7Li spectra were referenced externally
to 0.3 M LiCl in MeOH-d4 (δ ) 0.0). The typical pulse widths for
90° pulses (µs) and attenuation levels (dB, in brackets) were the
following: 7Li, 15 (1.50); 7Li-dec, 100 (18). The probe temperature
was calibrated using a methanol thermometer.
to room temperature, diluted first with 10 mL of Et2O and then
with 10 mL of brine, and the two phases were separated. The
aqueous phase was extracted with Et2O (3 × 10 mL), and the
combined organic phases were dried with Na2SO4 and concentrated
under reduced pressure. Flash chromatography on silica gel
(petroleum ether/Et2O 8/2) afforded deuterated oxirane 1-D (160
mg, 94% yield, >95% D).
1-(4,4-Dimethyl-2-oxazolin-2-yl)-2-methylpropan-1-one (3). To
a solution of 1 (84.5 mg, 0.5 mmol) in 3 mL of THF precooled to
-78 °C, under N2 and magnetic stirring, was slowly added s-BuLi
(0.38 mL, 0.55 mmol, 1.3 M in cyclohexane). After stirring for 15
min at -78 °C, the reaction mixture was allowed to warm to room
temperature and after 10 h was quenched with saturated aqueous
NH4Cl and extracted with Et2O (3 × 10 mL); the combined organic
phases were dried with Na2SO4 and concentrated under reduced
pressure. Flash chromatography on silica gel (petroleum ether/Et2O
9/1) afforded keto-oxazoline 3 (65 mg, 77% yield) as a colorless
oil. 1H (600 MHz, CDCl3): δ 1.28 (d, J ) 6.2 Hz, 6 H), 1.43 (s, 6
H), 3.52 (eptet, J ) 6.2 Hz, 1 H), 4.14 (s, 2 H). This compound
could be also efficiently prepared in 80% yield by lithiation (n-
BuLi/TMEDA, Et2O, -100 °C)-rearrangement of oxazolinylox-
irane 1, as reported.15
Reaction of Oxiranyllithium Li-1 with N,N-Dibenzylphenyla-
laninal 6. To a precooled solution (-98 °C, liquid N2/MeOH bath)
of 169 mg (1.0 mmol) of oxazolinyloxirane 1 in 5.0 mL of THF
was added 0.85 mL of s-BuLi (1.1 mmol, 1.3 M in cyclohexane)
under N2. After stirring for 5 min at -98 °C, a solution of 394 mg
(1.2 mmol) of racemic 2-(N,N-dibenzylamino)-3-phenylpropanal 6
in 3.0 mL of THF was added dropwise. The reaction mixture was
then stirred for additional 20 min at -98 °C and finally quenched
with 20 mL of saturated aqueous NH4Cl solution. After warming
to room temperature, the mixture was diluted with 10.0 mL of Et2O,
and the two phases were separated. The aqueous phase was
extracted with Et2O (3 × 20 mL), and the combined organic phases
were dried with Na2SO4 and concentrated under reduced pressure
NMR Spectroscopy: Representative Procedure for the Prepara-
tion of 0.2 M Sample of Oxiranyllithium Li-1 at -78 °C. An
appropriate amount of s-BuLi (77 µL, 0.1 mmol, 1.3 M in
cyclohexane), first filtered over celite, was concentrated under
reduced pressure in an NMR tube (assembled as described and
previously evacuated and purged with argon) and the residual oil
dissolved in THF-d8 (0.3 mL) at -78 °C (acetone/dry ice bath). In
a separate small flask, a weighed amount (17 mg, 0.1 mmol) of 1
was solubilized in THF-d8 (0.2 mL), and the resulting solution was
then slowly added by a gastight syringe to the solution of s-BuLi.
The resulting reaction mixture was then shaken and immediately
put in the NMR probe precooled to -78 °C.
1
to give 548 mg of a slightly yellowish oil. H NMR of the crude
indicated the complete disappearance of the starting oxirane. 13C
NMR indicated the presence of four diastereomers (7a-d) in a
ratio of 49:19:19:13. Flash chromatography (silica gel) over a 15-
cm column (ꢁ 5 cm) with petroleum ether/AcOEt 85/15 as the
eluant furnished first the most abundant isomer (Rf 0.35), then the
second less abundant isomer (Rf 0.25), and finally a mixture of the
two least abundant isomers (Rf 0.19) which were further separated
on the same column with petroleum ether/AcOEt 90/10 as the eluant
(Rf 0.15 and 0.11).
1-(4,4-Dimethyl-2-oxazolin-2-yl)-2-methyl-1,2-epoxypropane (1).
This compound was prepared according to the procedure previously
described.39 A chromatographic separation of the two enantiomers
was set up on an analytical chiral stationary HPLC phase (Chiral-
pack IA, hexane/i-PrOH 90/10 v/v, flow rate 1.00 mL/min, detector
UV at 220 nm, k1 ) 0.46 and R ) 1.60; see also Supporting
Information) and then quantitatively carried out on the same
preparative chiral stationary HPLC phase. First eluted enantiomer
Major diastereomer, presumably, (R*,S*,S*)-7a: white solid, mp
132-133 °C (Et2O), 44% isolated yield, Rf 0.35 (petroleum ether/
1
AcOEt 8/2). H NMR (400 MHz): δ 0.85 (s, 3 H), 1.17 (s, 3H),
1.39 (s, 3 H), 1.59 (s, 3 H), 2.94 (dd, J ) 14.3, 4.6 Hz, 1 H), 3.06
(dd, J ) 14.3, 7.9 Hz, 1 H), 3.37 (ddd, J ) 7.9, 6.1, 4.6 Hz, 1 H),
3.65 (d, J ) 13.9 Hz, 2 H), 3.70 and 3.81 (2 × d, AB system, J )
8.2 Hz, 2 H), 3.72 (d, J ) 13.9 Hz, 2 H), 4.01 (m, 1 H), 4.67 (br
s, exchanges with D2O, 1 H), 7.13-7.26 (m, 15 H). 13C NMR (100
MHz, DEPT): δ 20.2 (CH3), 20.8 (CH3), 27.6 (CH3), 28.8 (CH3),
32.9 (CH2), 53.8 (2 × CH2), 61.7 (CH), 63.1 (Cq), 63.2 (Cq), 67.7
(Cq), 71.0 (CH), 78.1 (CH2), 125.4 (CH), 126.4 (2 × CH), 127.9
(6 × CH), 128.9 (4 × CH), 129.5 (2 × CH), 139.8 (2 × Cq), 141.5
(+)-1: [R]20 ) +7.5 (c 1, CHCl3), ee >99.9%. Second eluted
D
enantiomer (-)-1: [R]20 ) -7.5 (c 1, CHCl3), ee ) 99.5%.
D
1-Deuterio-1-(4,4-dimethyl-2-oxazolin-2-yl)-2-methyl-1,2-epoxy-
propane (1-D). To a solution of 1 (169 mg, 1.0 mmol) in 5 mL of
THF, precooled to -78 °C under N2 and magnetic stirring, was
slowly added s-BuLi (0.85 mL, 1.1 mmol, 1.3 M in cyclohexane).
After stirring for 10 min at -78 °C, the reaction mixture was
quenched with 0.2 mL of CH3OD. The mixture was then allowed
(Cq), 163.0 (Cq). ESI-MS: 521 [M+ + 23 (Na)]. FT-IR (KBr, cm-1
)
3210, 3018, 2959, 1653 (CdN), 1601, 1493, 1363, 1294, 1070,
740, 698. Anal. Calcd for C32H38N2O3: C, 77.08; H, 7.68; N, 5.62.
Found: C, 76.73; H, 7.77; N, 5.36.
(37) One of the first 13C NMR studies of chiral R-lithiated-2-alkyloxazolines
was performed by Meyers et al., who demonstrated that in a THF solution, at
low temperature, the above metalated intermediates do exist as nonequilibrating
diastereomeric mixtures of azaenolates; see: Meyers, A. I.; Snyder, E. S.;
Ackerman, J. J. H. J. Am. Chem. Soc. 1978, 100, 8186–8189. It is also interesting
to note that for one of the above azaenolates, the 13C NMR chemical shift reported
for the C-2 oxazoline carbon was 168.5 ppm.
(38) Freeman, R.; Hill, H. D. W.; Kaptein, R. J. Magn. Reson. 1972, 7, 327–
329.
(39) Perna, F. M.; Capriati, V.; Florio, S.; Luisi, R. J. Org. Chem. 2002, 67,
8351–8359.
First minor diastereomer, presumably, (R*,R*,S*)-7b: white solid,
1
mp 124-126 °C (Et2O), 8%, Rf 0.11 (EP/AcOEt 9/1). H NMR
(400 MHz): δ 1.03 (s, 3 H), 1.18 (s, 3 H), 1.32 (s, 3 H), 1.43 (s,
3 H), 2.86 (dd, J ) 13.1, 1.9 Hz, 1 H), 3.14 (dd, J ) 13.1, 3.8 Hz,
1 H), 3.40 (m, 1 H), 3.66 (d, J ) 15.7, 2 H), 3.68 and 3.80 (2 ×
d, AB system, J ) 8.0 Hz, 2 H), 3.76 (d, J ) 7.5, 1 H), 4.10 (d,
J ) 15.7, 2 H), 4.77 (br s, exchanges with D2O, 1 H), 7.00-7.05
(m, 4 H), 7.10-7.16 (m, 5 H), 7.27-7.33 (m, 6 H). 13C NMR (100
J. Org. Chem. Vol. 73, No. 24, 2008 9563

Capriati et al.
MHz, DEPT): δ 20.9 (CH3), 22.2 (CH3), 28.1 (CH3), 28.3 (CH3),
33.5 (CH2), 53.7 (2 × CH2), 60.9 (CH), 63.5 (Cq), 64.6 (Cq), 67.6
(Cq), 71.8 (CH), 78.4 (CH2), 125.9 (CH), 127.1 (2 × CH), 128.3
(4 × CH), 128.4 (2 × CH), 129.1 (4 × CH), 129.8 (2 × CH),
138.9 (2 × Cq), 140.3 (Cq), 160.6 (Cq). ESI-MS: 521 [M+ + 23
(Na)]. FT-IR (KBr, cm-1) 3271, 3027, 2964, 2925, 1666 (CdN),
1495, 1454, 1365, 1122, 1028, 744, 698. Anal. Calcd for
C32H38N2O3: C, 77.08; H, 7.68; N, 5.62. Found: C, 76.91; H, 8.04;
N, 5.33.
IR (KBr, cm-1) 3272, 3062, 3028, 2966, 2926, 1668 (CdN), 1603,
1454, 1379, 1122, 1041, 744, 703. Anal. Calcd for C32H38N2O3:
C, 77.08; H, 7.68; N, 5.62. Found: C, 76.71; H, 7.99; N, 5.45.
Using optically active (S)-2-(N,N-dibenzylamino)-3-phenylpro-
panal 6, the four adducts 7a-d were obtained in almost the same
diastereomeric ratio (50:20:19:11 by 13C NMR) and, after isolation,
presented the following optical rotations: 7a [R]20 ) -49 (c 1,
D
CHCl3); 7b [R]20 ) -25 (c 1, CHCl3); 7c: [R]20 ) -82 (c 1,
D
CHCl3); 7d [R]20D ) +16 (c 1, CHCl3).
D
Second minor diastereomer, presumably, (S*,S*,S*)-7c: white
solid, mp 113-115 °C (Et2O), 17% isolated yield, Rf 0.25
Acknowledgment. This work was carried out under the
framework of the National Project “Stereoselezione in Sintesi
Organica. Metodologie ed Applicazioni” by the University of
Bari and by the Interuniversities Consortium C.I.N.M.P.I.S.
Mettler Toledo kindly provided the ReactIR spectrometer. We
also thank Professor Gideon Fraenkel (Ohio State University)
for his interest and helpful suggestions, Professor Francesco
Gasparrini (Universita` “La Sapienza”, Rome) for the chromato-
graphic separation of the two enantiomers of 1, Professor
Mohamed Amedjkouh (Go¨teborg University) who kindly sup-
ported preliminary NMR investigations on oxiranyllithium Li-
1, and an especially diligent reviewer for critical comments.
1
(petroleum ether/AcOEt 8/2). H NMR (400 MHz): δ 1.09 (s, 3
H), 1.14 (s, 3 H), 1.33 (s, 3 H), 1.56 (s, 3 H), 2.91 (dd, J ) 14.2,
10.9 Hz, 1 H), 3.07-3.15 (m, 6 H), 3.48 (d, J ) 10.1, 1 H), 3.73
(d, J ) 8.2, 1 H), 3.78 (d, J ) 8.0, 1 H), 4.70 (br s, exchanges
with D2O, 1 H), 7.18-7.41 (m, 15 H). 13C NMR (100 MHz, DEPT):
δ 19.6 (CH3), 21.2 (CH3), 27.8 (CH3), 28.4 (CH3), 33.7 (CH2), 53.4
(2 × CH2), 61.6 (CH), 62.6 (Cq), 63.2 (Cq), 67.0 (Cq), 71.4 (CH),
79.1 (CH2), 126.3 (CH), 127.1 (2 × CH), 128.3 (4 × CH), 128.4
(4 × CH), 129.1 (2 × CH), 130.0 (2 × CH), 138.5 (2 × Cq), 140.3
(Cq), 162.0 (Cq). ESI-MS: 521 [M+ + 23 (Na)]; FT-IR (KBr, cm-1
)
3352, 3027, 2966, 2927, 1663 (CdN), 1603, 1454, 1377, 1120,
1029, 750, 700. Anal. Calcd for C32H38N2O3: C, 77.08; H, 7.68;
N, 5.62. Found: C, 76.83; H, 8.01; N, 5.45.
Supporting Information Available: General (S2); copies of
1H and 13C NMR spectra for compounds 1-D (S3), (R*,S*,S*)-
7a (S4), (R*,R*,S*)-7b (S5), (S*,S*,S*)-7c (S6), (S*,R*,S*)-
Third minor diastereomer, presumably, (S*,R*,S*)-7d: white
solid, mp 121-122 °C (Et2O), 14%, Rf 0.15 (petroleum ether/AcOEt
1
9/1). H NMR (400 MHz): δ 0.98 (s, 3 H), 1.11 (s, 3 H), 1.37 (s,
1
7d (S7); copy of H NMR spectrum of oxiranyllithium Li-1
3 H), 1.66 (s, 3 H), 2.38-2.44 (m, 1 H), 2.86 (dd, J ) 13.5, 1.9
Hz, 1 H), 3.01 (d, J ) 8.2 Hz, 1 H), 3.03-3.12 (m, 2 H), 3.64 (d,
J ) 15.1 Hz, 2 H), 3.71 (d, J ) 8.2 Hz, 1 H), 4.11 (d, J ) 15.1,
2 H), 4.73 (br s, exchanges with D2O, 1 H), 7.00-7.05 (m, 4 H),
7.10-7.16 (m, 5 H), 7.27-7.33 (m, 6 H). 13C NMR (100 MHz,
DEPT): δ 19.1 (CH3), 22.3 (CH3), 27.5 (CH3), 28.3 (CH3), 31.8
(CH2), 53.7 (2 × CH2), 61.6 (CH), 62.9 (Cq), 64.5 (Cq), 67.2 (Cq),
71.9 (CH), 78.9 (CH2), 125.9 (CH), 126.3 (2 × CH), 127.6 (2 ×
CH), 128.0 (4 × CH), 128.1 (4 × CH), 130.4 (2 × CH), 140.0 (2
× Cq), 140.1 (Cq), 161.5 (Cq). ESI-MS: 521 [M+ + 23 (Na)]; FT-
(S8); copy of phase-sensitive gHSQC-DEPT spectrum of
oxiranyllithium Li-1 (Figure S1, S9); copy of the HPLC
chromatogram relative to the analytical separation of the two
enantiomers (+)-1 and (-)-1 (Figure S2, S10); xyz coordinates
(B3LYP/6-31+G*) relative to the two equilibrium geometries
found for oxazolinyloxirane 1 (S11). This material is available
free of charge via the Internet at http://pubs.acs.org.
JO801646E
9564 J. Org. Chem. Vol. 73, No. 24, 2008