Kinetic and computational studies of the composition and structure of activated complexes in the asymmetric deprotonation of cyclohexene oxide by a norephedrine-derived chiral lithium amide

Article abstract of DOI:10.1039/b101657l

Rational design of efficient chiral lithium amides for enantioselective deprotonations demands understanding of the origin of the selectivity. The mechanism of deprotonation of cyclohexene oxide 1 by lithium (1R,2S)-N-methyl-1-phenyl-2-pyrrolidinylpropanamide 3, which yields (S)-cyclohex-2-en-1-ol (S)-5 in 93% enantiomeric excess in tetrahydrofuran (THF), has been investigated. Kinetics have been used to show that the reaction is first order with respect to the reagents 1 and 3, respectively. NMR investigations of a ⁶Li and ¹⁵N labelled isotopologue of 3 have previously shown that 3 is mainly a dimer of the lithium amide monomer in THF in the initial state. On the basis of these results it is concluded that the rate-limiting activated complexes for the epoxide deprotonation are composed of two molecules of monomer of lithium amide 3 and one molecule of epoxide. Structures and energies of unsolvated and specific THF-solvated reagents and activated complexes have been calculated using PM3 and B3LYP/6-31+G(d). The results are currently being explored for the rational design of chiral lithium amides with improved stereoselectivities.

Full text of DOI:10.1039/b101657l

View Article Online / Journal Homepage / Table of Contents for this issue
Kinetic and computational studies of the composition and structure
of activated complexes in the asymmetric deprotonation of
cyclohexene oxide by a norephedrine-derived chiral lithium amide †
2
Daniel Pettersen, Mohamed Amedjkouh, Sten O. Nilsson Lill, Kristian Dahlén and
Per Ahlberg*
Organic Chemistry, Department of Chemistry, Göteborg University, SE-412 96, Göteborg,
Sweden. E-mail: Per.Ahlberg@oc.chalmers.se; Fax: +46 31 772 2899; Tel: ϩ46 31 772 2899
Received (in Cambridge, UK) 20th February 2001, Accepted 19th March 2001
First published as an Advance Article on the web 27th April 2001
Rational design of efficient chiral lithium amides for enantioselective deprotonations demands understanding of the
origin of the selectivity. The mechanism of deprotonation of cyclohexene oxide 1 by lithium (1R,2S)-N-methyl-1-
phenyl-2-pyrrolidinylpropanamide 3, which yields (S)-cyclohex-2-en-1-ol (S)-5 in 93% enantiomeric excess in
tetrahydrofuran (THF), has been investigated. Kinetics have been used to show that the reaction is first order with
respect to the reagents 1 and 3, respectively. NMR investigations of a 6Li and 15N labelled isotopologue of 3 have
previously shown that 3 is mainly a dimer of the lithium amide monomer in THF in the initial state. On the basis of
these results it is concluded that the rate-limiting activated complexes for the epoxide deprotonation are composed of
two molecules of monomer of lithium amide 3 and one molecule of epoxide. Structures and energies of unsolvated
and specific THF-solvated reagents and activated complexes have been calculated using PM3 and B3LYP/6-31+G(d).
The results are currently being explored for the rational design of chiral lithium amides with improved
stereoselectivities.
of the norephedrine-derived chiral lithium amide lithium
(1R,2S)-N-methyl-1-phenyl-2-pyrrolidinylpropanamide 3 and
Introduction
Enantiomers of chiral lithium amides are used in asymmetric
its enantiomer. An efficient synthesis of the corresponding
deprotonation of e.g. meso-epoxides to afford chiral allylic
precursor diamine (1R,2S)-N-methyl-1-phenyl-2-pyrrolidinyl-
alcohols in enantiomeric excess (ee).1–3 Such products are
propanamine 4 and its enantiomer was also developed.24 Using
important intermediates in the total synthesis of e.g. bio-
lithium amide 3 or its enantiomer ees ≥93% of (S)-cyclohex-2-
en-1-ol [(S)-5] or its enantiomer, respectively, were obtained
upon deprotonation of 1.
Independently O’Brien and co-workers have recently reported
logically active compounds.4–7 There is a need for invention
of new chiral amides with improved stereoselectivities and which
are easily available synthetically. The development of stereo-
selective lithium amides for deprotonations has mainly been
the synthesis of 4 and the application of the lithium amide 3
carried out in a trial-and-error fashion.8–17 To be able to ration-
in enantioselective deprotonation of different functionalised
ally develop new and highly selective chiral lithium amides
meso-epoxides yielding allylic alcohols with high ees.25,26
knowledge is needed about the detailed reaction mechanisms
In this paper we report a detailed study of the mechanism of
and, in particular, the composition and structures of the rate-
limiting activated complexes involved.
the reaction shown in Scheme 1 in which 3 deprotonates 1 in
By use of multinuclear NMR studies of isotopically labelled
lithium amides the molecular composition of the lithium
amide aggregates in the initial state has been determined. Such
knowledge and transition state theory, together with kinetic
reaction orders with respect to reactants in the deprotonation
reaction, give the composition of the rate-limiting activated
complexes.18–20 Based on this it is possible to computationally
model structures of initial state complexes and activated com-
plexes in solution.21–23 Such knowledge is a prerequisite for
rational design by e.g. computational methods of new and
highly stereoselective chiral lithium amides.
Previous investigations of the structure of activated com-
plexes involved in the deprotonation of cyclohexene oxide 1 by
lithium (S)-2-(pyrrolidin-1-ylmethyl)pyrrolidide 2 have shown
that the rate-limiting deprotonation activated complexes are
built from one molecule of lithium amide monomer and one
molecule of 1.18,21 This was the starting point for the design
Scheme 1 The investigated enantioselective deprotonation reaction of
the epoxide 1 by the lithium amide 3 in THF.
THF to yield after isolation the allylic alcohol (S)-5 in 93% ee.
A kinetic study at 20 ЊC has been carried out and the reaction
orders with respect to 1 and 3 have been determined. A detailed
computational study of amide initial state structures and pos-
sible activated complexes is also presented.
Results and discussion
In this section the kinetic study of the molecular composition
of the rate-limiting deprotonation activated complexes is pre-
sented together with computationally modelled aggregates of
† Electronic supplementary information (ESI) available: structures
and energies of possible diastereoisomeric activated complexes leading
to (R)- and (S)-products. See http://www.rsc.org/suppdata/p2/b1/
b101657l/
1654
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661
This journal is © The Royal Society of Chemistry 2001
DOI: 10.1039/b101657l

View Article Online
Fig. 1 Monomer structure and dimeric structures of 3.
the lithium amide in the initial state and possible transition
The internal coordination bond between lithium and pyrrol-
state structures.
idine nitrogen could be broken giving a structure such as 8g
with one lithium dicoordinated and one tricoordinated. Similarly
8f yields 8h. The isomerisation of dimers could proceed via
structures with decreased internal coordination.
Structure of the lithium amide in the initial state
The composition and structure of the lithium amide reagent 3
The NMR results are not conclusive with regard to which
isomer with non-equivalent lithiums is dominant.
6
have been determined using its Li and 15N labelled isotopo-
logue, of which its monomer 6 is shown in Fig. 1, and NMR
coupling constants. The results have shown that in THF the
lithium amide 3 is present mainly as a dimer with magnetically
non-equivalent lithiums.27 Other types of amide structures
(monomer, trimer, tetramer, etc.) have not been detected.
Possible dimeric structures of 3, excluding THF solvation,
are shown in Fig. 1 together with a structure of monomer 7. In
isomers 8a–8c the two lithiums are non-equivalent; one is dico-
ordinated and one is tetracoordinated. In the isomer 8d the
lithiums are also non-equivalent but both are tricoordinated. In
the two structures 8e and 8f on the other hand the lithiums are
both equivalent and tricoordinated.
The enthalpies of formation and the structures shown in Fig.
1 with and without specific THF solvation have been calculated
using the semiempirical computational method PM3.28,29 This
method was chosen because it has been shown to yield useful
structures of lithium organic compounds and because the
aggregates under scrutiny are computationally demanding.
Several previous studies have shown the ability of PM3 to gen-
erate structures in good agreement with experimental results on
lithium organic structures.23,29–31 Optimisations have been per-
formed with preceding conformational search with respect to
the phenyl groups and solvent. The configurations of the nitro-
gens in the stereoisomeric dimers were also altered in the search
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661
1655

View Article Online
Scheme 2 Initial state equilibria between unsolvated and specific THF-solvated monomers and dimers and calculated enthalpies (kcal molϪ1
relative to 8bؒTHF ϩ 3 THF.
)
to find the most stable ones. Large variations in energy were
found. The equilibria between monomer and dimers have
also been studied including specific solvation by THF. In
Scheme 2 equilibria between some of the most stable unsol-
vated, monosolvated and disolvated species are shown together
with calculated relative enthalpies of formation.
than 5%. In Fig. 3 the concentration of 5 vs. time has been
plotted. The initial rates were measured as the slopes of fitted
straight lines. The runs were usually reproduced within <3% of
the average values, respectively.
The determined initial rates at different initial concentrations
of the reactants are collected in Table 1.
Unsolvated and monosolvated dimers are both strongly
favoured over unsolvated and monosolvated monomers. The
dimeric complex calculated to be most stable is the THF-
solvated dimer 8bؒTHF in which THF is bonded to the trico-
ordinated lithium while the other lithium is intramolecularly
tetracoordinated. However, 8gؒTHF is only 0.1 kcal molϪ1
higher in energy than 8bؒTHF. This indicates that steric strain
is released when the the pyrrolidine nitrogen coordination bond
is broken. Disolvation by THF is found to be destabilising
compared with their monosolvated counterparts.
The calculated structures of the stable dimers 8bؒTHF,
8gؒTHF and 8hؒTHF, respectively, are shown in Fig. 2. Thus the
computational results are consistent with our NMR findings
that dimers with non-equivalent lithiums are more abundant
than ones with equivalent lithiums. The isomerisation of
dimeric structures with non-equivalent lithiums to structures
with equivalent lithiums is probably rapid.
Using eqn. (2) reaction orders were determined from the
slopes of fitted straight lines of plots of log (d[5]/dt) vs. log [1]
and log [3] shown in Fig. 4 and 5, respectively. The determined
slopes were 0.97 and 1.05 for 1 and 3, respectively.
Thus, the deprotonation reaction is concluded, within
experimental error, to be first order with respect to both the
epoxide 1 and 3, i.e., first order in lithium amide dimer.
To be able to interpret the experimentally observed reaction
orders these have been compared with predicted reaction orders
for monomer- and dimer-based pathways. Reactions through
the β-elimination activated complexes of types 9 and 10 (which
are kinetic representatives for monomer- and dimer-based tran-
sition states, respectively) have been considered.
Activated complexes 9 are built from the monomer 7 of 3 and
one molecule of 1. The transition states 10 on the other hand
are composed of two monomers (a dimer) of 3 and one epoxide
molecule. Transition states 10a to 10f are built from one
molecule of 1 and dimers 8a to 8f, respectively. According to
transition state theory the activated complexes are assumed to
be in equilibrium with the reagents 1 and 3. The reaction via
activated complex 9 is shown in eqn. (3) and reactions involving
dimers in the activated complexes are described by eqn. (4).
In the next section the kinetic determination of reaction
orders with respect to reactants of the epoxide deprotonation is
presented.
Kinetics
Knowledge of the lithium amide reagent 3 as being dimeric in
the initial state in THF makes it possible to use the kinetics of
the epoxide deprotonation to determine the composition of the
rate-limiting activated complexes involved. To achieve this, the
reaction orders of the reactants have to be experimentally
determined. A general rate equation for product formation is
shown in eqn. (1) with the reaction orders x and y with respect
to the reagents 1 and 3, respectively.
0.5 3 + 1
3 + 1
9 → 5 + 4
(3)
(4)
10 → 5 + 0.5 3 + 4
From eqn. (3) and eqn. (4) and transition state theory the rate
equations in eqn. (5) and eqn. (6) have been derived, respectively.
d[5]/dt = k1[3]0.5[1]
d[5]/dt = k2[3][1]
(5)
(6)
Taking the logarithm of eqn. (1) yields eqn. (2).
d[5]/dt = kobs[1]x [3]y
(1)
(2)
Thus, if the deprotonation proceeds via rate-limiting acti-
vated complexes 9, the reaction order will be 0.5 with respect to
3 [eqn. (5)] since the lithium amide has been shown to be
dimeric in the initial state. On the other hand, if the reaction
takes place via activated complexes 10, which are all built from
lithium amide dimers and one molecule of 1, respectively, the
reaction order is predicted to be 1 with respect to 3 [eqn. (6)].
Thus, in both cases the reaction order with respect to 1 will be 1.
Clearly only the latter alternative [eqn. (6)] is consistent with
the experimentally determined reaction orders. Thus, it is con-
cluded that the rate-limiting activated complexes are composed
of a lithium amide dimer and one molecule of the epoxide 1. In
the next section the computational modelling of structures and
energies of possible transition states is presented.
log (d[5]/dt) = log kobs + xlog [1] + ylog [3]
A calibrated quench-extraction-gas chromatography pro-
cedure described in the Experimental has been developed to
determine initial rates (d[5]/dt). The time dependence of the
concentration of the product 5 has been measured using differ-
ent initial concentrations of the epoxide 1 and lithium amide 3,
respectively. Initial rates were used to determine reaction orders
in order to avoid possible interference of reaction products. The
asymmetric deprotonation of 1 by 3 was carried out in THF at
20.00 0.05 ЊC. The reaction was started by addition of 1 to a
solution of 3 and the reaction was monitored by measuring the
appearance of extracted 5 at conversions of 1 usually lower
1656
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661

View Article Online
Table 1 Initial rates obtained from runs with different initial concen-
trations of 1 and 3 in THF at 20.00 ЊC
Run
[1]/mM
[3]/mM
(d[5]/dt)/106 M sϪ1
1
2
3
4
5
6
7
8
9
200
200
200
200
200
200
20
50
60
60
100
100
50
50
65
8.97
9.19
13.4
19.5
20.1
24.6
2.08
5.13
6.30
6.67
10.3
10.2
100
100
130
100
100
100
100
100
100
10
11
12
Fig. 3 The concentration of 5 plotted vs. time for the initial part of the
deprotonation reaction of is shown for two different initial
1
concentrations of 1: lower line 100 mM; upper line 200 mM.
Fig. 4 log ([5]/dt) is plotted vs. log [1] and the slope of the fitted
straight line is the measured order of the reaction with respect to the
reactant 1.
Fig. 5 log ([5]/dt) is plotted vs. log [3] and the slope of the fitted
straight line is the measured order of the reaction with respect to
reactant 3.
modelling on dimer-based transition state structures.
Monomer-based activated complexes have also been modelled.
However, only syn-β-elimination activated complexes have been
modelled in the absence of any evidence for α-elimination.
Structures and energies of possible diastereoisomeric activated
complexes leading to (R)- and (S)-products, respectively, have
been calculated by PM3 and B3LYP/6-31+G(d). Altogether
about 82 unsolvated, mono- and di-THF-solvated activated
complexes have been calculated (see electronic supplementary
information†). In Table 2 enthalpies of activated complexes
having the lowest enthalpy are reported.
Fig. 2 The calculated most stable structures by PM3 of the lithium
amide. Relative stabilities are shown in Scheme 2. Hydrogens are
omitted for clarity.
Computational study of activated complexes
The acquired knowledge about the composition of the rate-
limiting activated complex focused the efforts on computational
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661
1657

View Article Online
Table 2 Calculated enthalpies (kcal molϪ1) of formation (PM3) for different types of unsolvated and THF-solvated diastereomeric transition states
Solvent
9
10a
10f
10g
10h
Unsolvated
1 THF
(R) 20.630
(S) 20.831
(R) Ϫ36.629
(S) Ϫ37.056
(R) 25.772
(R) 20.282
(R) 21.326
(R) 28.974
(S) 23.449
(R) Ϫ30.593
(S) Ϫ28.337
(S) 20.303
(R) Ϫ32.920a
(S) Ϫ34.810b
(S) 20.052
(R) Ϫ32.987c
(S) Ϫ32.476
(S) —d
(R) Ϫ25.097
(S) Ϫ31.778
a Absolute energy: Ϫ1868.68586 Eh. (R)-10fؒTHF is +1.36 kcal molϪ1 less stable than (S)-10fؒTHF calculated using B3LYP/6-31+G(d)//PM3.
b Absolute energy: Ϫ1868.68802 Eh calculated using B3LYP/6-31+G(d)//PM3. c Absolute energy: Ϫ1868.68418 Eh. (R)-10gؒTHF is +2.41 kcal molϪ1
less stable than (S)-10fؒTHF calculated using B3LYP/6-31+G(d)//PM3. d Not found at PM3 level.
Table 3 Calculated enantioselectivities for different types of pairs of transition states using the data in Table 2. The ees have been estimated
assuming that the difference of free energies of activation is equal to the difference of the enthalpies of formation for the activated complexes yielding
the (R)- and (S)-products, respectively
Solvent
9
10a
10f
10g
10h
Exp.
Unsolvated
1 THF
1 THF
17% (R)
35% (S)
96% (S)
96% (R)
2% (R)
79% (S)
41% (R)
—
100% (S)
92% (S)
93% (S)
82% (S)a
a Calculated using B3LYP/6-31+G(d)//PM3.
Of the dimer-based, unsolvated activated complexes the
calculated diastereoisomeric complexes (R)-10f and (S)-10g
(Scheme 3) were found to have the lowest enthalpy.
the stereoisomer (R)-5 is the strongly preferred enantiomer
by the reactions via 10aؒTHF and 10gؒTHF, and (S)-5 is the
preferred one via 9ؒTHF, 10hؒTHF and 10fؒTHF—the acti-
vated complex with the lowest activation enthalpy. The calc-
ulated ee using the transition states with lowest enthalpy is 91%.
The agreement of this latter predicted value with the experi-
mental 93% ee may be fortuitous. However, good agreement
between PM3-calculated and measured enantioselectivities has
previously been reported.21,22 The predictive power of PM3 in
the present context could be due to major cancellation of errors
in energies because of close structural similarity between the
diastereoisomeric activated complexes. Single-point calcula-
tions using B3LYP/6-31+G(d) on PM3-optimised transition
states were also performed. (R)-10gؒTHF was found to be 1.06
kcal molϪ1 higher in energy than (R)-10fؒTHF at DFT-level in
contrast to the result at PM3 level. The energy difference
between (R)-10fؒTHF and (S)-10fؒTHF was lowered from 1.89
kcal molϪ1 to 1.36 kcal molϪ1. This corresponds to an ee of 82%
in favour of (S)-5 also in agreement with the experimental value
of 93%.
Their calculated enthalpies relative to the reference (Scheme
3) were almost equal (31.6 and 31.4 kcal molϪ1, respectively, see
Scheme 3) and 1.1 and 0.2 kcal molϪ1 more stable than the
second most stable couple of diastereoisomeric complexes (R)-
10g and (S)-10f, respectively. Upon monosolvation by THF the
transition states become stabilised and (R)-10gؒTHF and (S)-
10fؒTHF are now the complexes with lowest enthalpy. These
have a 2.1 and 3.7 kcal molϪ1 lower barrier, respectively, than
the unsolvated counterparts. (R)-10fؒTHF is of almost equal
energy as (R)-10gؒTHF. The calculated enthalpies of solvation
of (R)-10g and (S)-10f are 3.2 and 3.9 kcal molϪ1, respectively.
The solvation energy for (R)-10f is 2.1 kcal molϪ1. Optimised
structures of the transition states are given in Fig. 6.
Solvation of any of the mono-THF-solvated dimers by
another THF molecule resulted in destabilisation. Such transi-
tion states are therefore not further considered in the present
context.
Using the data in Scheme 3 product enantiomeric excesses in
the deprotonations via the unsolvated and mono-THF-solvated
activated complexes 9, 10a, 10f, 10g, and 10h have been pre-
dicted and the results are shown in Table 3. Reactions through
both 10a and 10g favour the product enantiomer (S)-5 but the
reaction through 9 and 10f slightly prefers the other enantiomer
(R)-5. Mono-solvation by THF of the transition states has
dramatic effects with reversal of the enantioselectivities. Now
The solvation is expected to lower both the entropy and the
enthalpy of activation. Depending on the magnitude of these
changes and the temperature, the free energy of activation may
either decrease or increase upon solvation. If the former is the
case for activated complex 10fؒTHF, it will have lower free
energy than 10f + THF and thus be the transition state mainly
used in the enantioselective deprotonation together with
10gؒTHF. In the latter situation 10f and 10g may be the pre-
Scheme 3 Equilibria between initial states and possible unsolvated and specific THF-solvated activated complexes. Enthalpies are given in kcal
molϪ1
.
1658
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661

View Article Online
The THF-monosolvated dimer-based transition states
(10aؒTHF and 10fؒTHF) with lowest energy have been con-
structed using the dimers 8a and 8f, respectively (Fig. 1 and 6).
These transition states may be seen as built from open dimers.
10g was optimised by opening one lithium–pyrrolidine nitrogen
bond in 10a.
Along the reaction coordinate following coordination of the
epoxide 1 to the dimer, one of the amide N–Li bonds opens.
The lithium coordinates to the epoxide oxygen and the nitrogen
that becomes basic abstracts a β-proton. As shown also in an
earlier study of epoxide deprotonation using 2 the proton in
flight in the transition state is found to be somewhat more than
half-transferred towards the nitrogen from the carbon.21 The
C–O bond has opened up somewhat (1.53–1.54 Å) and the C–C
double bond has started to develop (1.44–1.45 Å). Using 2
where the activated complexes were composed of monomers
these bonds were found to be 1.52–1.55 Å and 1.45–1.46 Å,
respectively, using PM3. These bond lengths are comparable to
those of a recent high-level ab initio study (MP2/6-31+G(d)) of
hydroxide deprotonation (no coordination of lithium to the
epoxide oxygen) of 1 by Morgan and Gronert who found the
corresponding distances to be 1.49 and 1.48 Å, respectively.33
The calculated barriers are somewhat higher (28 kcal molϪ1)
than observed, which is a known error for PM3.29,34
Conclusion
The enantioselective deprotonation of the epoxide 1 by the
lithium amide 3 in THF has been shown to proceed via rate-
limiting transition states built from two monomers of 3 and one
molecule of 1. PM3 and B3LYP/6-31+G(d) calculations of
structures and energies of possible activated complexes show
the decisive role of THF solvation of the activated complexes
for the enantioselectivity. The results suggest that proper
structural modification of the activated complexes through
structural modification of 3 may yield deprotonations with
increased enantioselectivities.
Experimental
The synthesis of the lithium amide 3 precursor diamine
(1R,2S)-N-methyl-1-phenyl-2-pyrrolidinylpropanamine 4 and
its enantiomer (1S,2R)-N-methyl-1-phenyl-2-pyrrolidinylprop-
anamine from the commercially available enantiomers (1R,2S)-
norephedrine and (1S,2R)-norephedrine, respectively, were
carried out in three-step, one-pot preparations. In the first
step the amino group in the norephedrine was transformed to a
pyrrolidino group by reaction with 1,4-dibromobutane using
a modification of a procedure reported by Zhao et al.35 The
second and third steps were carried out similarly to a recipe by
Miao and Rossiter.36 Upon mesylation of the hydroxy group
an aziridinium ion intermediate formed. Reaction of the
aziridinium ion with methylamine gave the amine 4. The first
step in the above sequence is a modification of a previously
reported procedure by O’Brien and co-workers.25
Synthesis of (1R,2S)-N-methyl-1-phenyl-2-pyrrolidinyl-
propanamine 4
Fig. 6 The calculated most stable activated complexes by PM3.
Relative stabilities are shown in Scheme 3. Some hydrogens are omitted
for clarity.
(1R,2S)-Norephedrine (6.0 g, 40 mmol), 1,4-dibromobutane
(9.5 g, 44 mmol) and NaHCO3 (7.4 g, 88 mmol) were refluxed in
100 ml toluene for 32 h. The solid was removed by filtration and
the filtrate was concentrated in vacuo to yield the product ter-
tiary amino alcohol as a colourless oil (97%). The oil was dis-
solved in dry THF (distilled from sodium–benzophenone ketyl)
in a nitrogen atmosphere and the reaction flask was equipped
with septa and cooled to 0 ЊC in an ice bath. Triethylamine (12.4
g, 120 mmol) was added to the stirred solution and methane-
sulfonyl chloride (9.35 g, 80 mmol) was added dropwise with a
syringe over 20 min while a yellow precipitate formed. The reac-
dominating transition states depending on the concentration
of THF. PM3 has been shown to underestimate solvation
enthalpies by 3–4 kcal molϪ1.21,31,32 Assuming an entropy of
activation of about 5 kcal molϪ1 a net stabilisation by solvation
of (S)-10f is expected.31,32 It is thus predicted that (S)-5 will be
the major enantiomer resulting from the chiral lithium amide
deprotonation in line with what has been experimentally
observed.
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661
1659

View Article Online
tion flask was removed from the ice bath and stirred for another
1.5 h. Triethylamine (8.2 g, 80 mmol) and methylamine (30 ml,
240 mmol, 33% in ethanol) were added to the reaction mixture
and the main part of the yellow precipitate dissolved. Water (10
ml) was then added and the remaining precipitate dissolved
yielding a clear solution which was stirred at rt for 20 h. The
aqueous phase was extracted with 3 × 40 ml diethyl ether. The
collected organic phases were dried over Na2SO4 and filtered
through 1.5 cm of silica to remove traces of amino alcohol.
Evaporation in vacuo gave a brownish oil which was distilled at
reduced pressure using a Vigreux to yield the amine 4 (6.6 g,
73%) as a colourless oil (>99% NMR). Bp 65–67 ЊC/3 × 10Ϫ2
mbar; δH (400 MHz; CDCl3) 0.80 (3H, d, Me, J = 6.4 Hz), 1.75–
1.83 (4H, m, CH2), 1.90 (1H, br s, NH), 2.26–2.28 (1H, m,
CHN, J = 3.2 Hz), 2.34 (3H, s, NMe), 2.54–2.57 (4H, m,
NCH2), 3.85 (1H, d, PhCHN, J = 3.2 Hz), 7.15–7.30 (5H, m,
Ph). δC (125 MHz; THF-d8) 13.4 (Me), 25.3 (CH2), 35.8
(NHMe), 53.0 (NCH2), 67.8 (CHN), 67.9 (CHN), 127.2, 128.7,
128.8, 143.3 (Ph). HRMS FAB (M + H) calculated C14H22N2
219.1861, found 219.1787.
cyclohexene oxide 1 and 0.85 and 4.9 min for cyclohex-2-en-1-
ol 5, respectively. The retention time for the reference com-
pound hexan-1-ol was 3.8 min.
The enantiomeric excesses of 5 were measured using a
Chrompack Chirasil-CB Dex (30 m, 0.25) at 90 ЊC. tR(S)-
5 = 7.45 min, tR(R)-5 = 7.90 min.
Typical kinetic procedure. Amine 4 (43.7 µl, 0.2 mmol) was
dissolved in THF (775 µl) in a reaction vessel inside the glove-
box and then transferred out of the glovebox, and n-
butyllithium (81.5 µl, 2.45 M in hexanes) was added under a
nitrogen atmosphere. The yellow reaction solution was allowed
to equilibrate at 20.00 0.05 ЊC for 10 minutes in a thermostat
(Heto Birkerød). The reaction was started by addition of
cyclohexene oxide 1 (100 µl, 2.0 M) and samples (50 µl) were
withdrawn from the reaction vessel at approximately 2 minute
intervals and quenched in hydrochloric acid solution (100 µl,
0.6 M saturated with sodium chloride). Compounds 1 and 5
were extracted with carbon tetrachloride (500 µl) containing
the standard hexan-1-ol (3.16 mM). The liquid phases were
separated by centrifugation and 250 µl of the organic phase
were transferred to a vial and analysed by capillary gas
chromatography.
Kinetics
General. Reaction vessels and syringes were dried in a vac-
uum oven (50 ЊC) overnight. Transfers of reagents were per-
formed with gas-tight syringes in a nitrogen atmosphere. THF
was distilled from sodium–benzophenone ketyl in a nitrogen
atmosphere and stored over 4 Å molecular sieves in septum-
sealed vials in a glovebox (Mecaplex GB 80 equipped with a gas
purification system that removes oxygen and water). A stock
solution (2.0 M) of cyclohexene oxide 1 (distilled from calcium
hydride) in THF was prepared inside the glovebox. Hexan-1-ol
was used as a standard in the GC-analysis; a stock solution
(3.16 mM) of hexan-1-ol (distilled from calcium hydride) in
carbon tetrachloride (distilled from calcium chloride) was
prepared.
The quantitative transfer of 1 and 5 from the aqueous phase
to the carbon tetrachloride phase during the work-up was
determined as follows: Solutions of hexan-1-ol and 5 in THF
with compositions similar to those in the kinetic experiments
were prepared. Samples (50 µl) were withdrawn and added to
solutions of carbon tetrachloride (500 µl) containing hexan-1-
ol (3.16 mM). Other samples (50 µl) of solutions of 1 and 5
were added to solutions of hydrochloric acid (100 µl, 0.6 M
saturated with sodium chloride). The latter mixtures were
extracted with solutions of carbon tetrachloride (500 µl) con-
taining hexan-1-ol (3.16 mM). After separation by centri-
fugation 250 µl of the organic layers were transferred to vials.
The samples from the two types of preparations were analysed
by capillary gas chromatography. The concentration ratios of
epoxide 1 and allylic alcohol 5 to hexan-1-ol determined for the
two types of preparations were found to be within 0.5% of the
average value, respectively.
Determination of [BuLi]. A double Gilman titration follow-
ing the (ASTM) standard E233-90 was used with some modifi-
cations. n-Butyllithium (1.0 ml, 2.5 M in hexanes) was added to
hexane (10 ml) and the solution was quenched by addition of
water (10 ml). Phenolphthalein (3 drops; 0.5 g lϪ1, in ethanol–
water 1 : 1) was added and the pink solution was titrated with
hydrochloric acid (0.0902 M calibrated with NaOH) to colour-
lessness. A second sample of n-butyllithium (1.00 ml) was
slowly added dropwise to a mixture of diethyl ether (2.5 ml,
distilled from sodium–benzophenone ketyl) and benzyl chloride
(1.00 ml, distilled from P2O5) under a nitrogen atmosphere.
Water (10 ml) was added and the pink solution was titrated with
hydrochloric acid (0.0902 M calibrated with NaOH) to colour-
lessness. The concentration of the n-butyllithium solution was
calculated from the difference between the two titrations and
was found to be 2.45 M.
The concentration of 5 was measured for about the first 5%
of conversion of 1. Initial rates were determined as the slopes
of fitted straight lines to plots of concentrations of 5 vs. time
(see Fig. 3). Initial rates were usually reproduced within 2% of
the average values, respectively.
Computational details. Geometries were optimised at PM3
level of theory.28 In Spartan37 the option HHON38 was used to
correct for hydrogens in close contact.39,40 All geometries were
characterised as minima or transition states on the potential
energy surface (PES) by use of the sign of the eigenvalues of
the force constant matrix obtained from a frequency calcula-
tion. Transition states with exactly one imaginary frequency
were confirmed to describe the correct movement on the PES
by a mode analysis. Reaction energies and activation barriers
were calculated at PM3 level of theory. Single point calcula-
tions using B3LYP/6-31+G(d)41–45 were performed on selected
transition states.
Gas chromatography analysis. Gas chromatography analyses
were performed on a Varian 3400 chromatograph equipped
with an 8200 Cx autosampler and a flame ionisation detector
(FID). For the separation an achiral DBWX-30W column (30
m, 0.25 µm) from J & W Scientific was used with hydrogen as
carrier gas (2 ml minϪ1). Reaction samples (1.0 µl) were intro-
duced on to the column via a split injector (split flow 15 ml
minϪ1) and the components were separated using a temperature
program. Initially the temperature was held at 80 ЊC for 2 min-
utes and then during 2 minutes it was increased to 120 ЊC. The
injector temperature was 225 ЊC and the detector was held at
250 ЊC.
Gas chromatography response factors for 1 and 5 were
determined using hexan-1-ol as a reference. Carbon tetra-
chloride samples of known compositions similar to those used
in the kinetic experiments were analysed. The response factors
and retention times measured were 1.01 and 2.3 min for
Acknowledgements
We thank the Swedish Natural Science Research Council for
financial support and Mr Roine I. Olsson for discussions.
References
1 P. J. Cox and N. S. Simpkins, Tetrahedron: Asymmetry, 1991, 2, 1.
2 D. M. Hodgson, A. R. Gibbs and G. P. Lee, Tetrahedron, 1996, 52,
14361.
3 P. O’Brien, J. Chem. Soc., Perkin Trans. 1, 1998, 1439.
4 J. S. Sabol and R. J. Cregge, Tetrahedron Lett., 1989, 30, 3377.
1660
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661

View Article Online
25 B. Colman, S. E. de Sousa, P. O’Brien, T. D. Towers and W. Watson,
5 T. Kasai, H. Watanabe and K. Mori, Bioorg. Med. Chem., 1993,
1, 67.
Tetrahedron: Asymmetry, 1999, 10, 4175.
6 M. Asami, J. Takahashi and S. Inoue, Tetrahedron: Asymmetry,
1994, 5, 1649.
7 S. Barrett, P. O’Brien, H. C. Steffens, T. D. Towers and M. Voith,
Tetrahedron, 2000, 56, 9633.
8 M. Kirihara, Chem. Lett., 1987, 389.
9 M. Asami, Bull. Chem. Soc. Jpn., 1990, 63, 721.
10 D. Bhuniya and V. K. Singh, Synth. Commun., 1994, 24, 375.
11 D. Bhuniya and V. K. Singh, Synth. Commun., 1994, 24, 1475.
12 D. Bhuniya, A. DattaGupta and V. K. Singh, Tetrahedron Lett.,
1995, 36, 2847.
13 D. Bhuniya, A. DattaGupta and V. K. Singh, J. Org. Chem., 1996,
61, 6108.
14 M. Asami, T. Suga, K. Honda and S. Inoue, Tetrahedron Lett., 1997,
38, 6425.
15 J. P. Tierney, A. Alexakis and P. Mangeney, Tetrahedron:
Asymmetry, 1997, 8, 1019.
26 S. E. de Sousa, P. O’Brien and H. C. Steffens, Tetrahedron Lett.,
1999, 40, 8423.
27 M. Amedjkouh, D. Pettersen, S. O. Nilsson Lill, Ö. Davidsson and P.
Ahlberg, submitted for publication.
28 J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209.
29 E. Anders, R. Koch and P. Freunscht, J. Comput. Chem., 1993, 14,
1301.
30 G. Hilmersson, P. I. Arvidsson, Ö. Davidsson and M. Håkansson,
J. Am. Chem. Soc., 1998, 120, 8143.
31 A. Abbotto, A. Streitwieser and P. v. R. Schleyer, J. Am. Chem. Soc.,
1997, 119, 11255.
32 E. Kaufmann, J. Gose and P. v. R. Schleyer, Organometallics, 1989,
8, 2577.
33 K. M. Morgan and S. Gronert, J. Org. Chem., 2000, 65, 1461.
34 S. Morpurgo, M. Bossa and G. O. Morpurgo, THEOCHEM, 1998,
429, 71.
16 M. J. Södergren and P. G. Andersson, J. Am. Chem. Soc., 1998, 120,
10760.
17 M. J. Södergren, S. K. Bertilsson and P. G. Andersson, J. Am. Chem.
Soc., 2000, 112, 6610.
18 R. I. Olsson and P. Ahlberg, Tetrahedron: Asymmetry, 1999, 10,
3991.
35 D. Zhao, C.-y. Chen, F. Xu, L. Tan, R. Tillyer, M. E. Pierce and
J. R. Moore, Org. Synth., 2000, 77, 12.
36 G. Miao and B. E. Rossiter, J. Org. Chem., 1995, 60, 8424.
37 W. J. Hehre, B. J. Deppmeier, A. J. Driessen, J. A. Johnson, P. E.
Klunzinger, L. Lou, J. Yu, J. Baker, J. E. Carpenter, R. W. Dixon,
S. S. Fielder, H. C. Johnson, S. D. Kahn, J. M. Leonard and W. J.
Pietro, Spartan v. 5.0.1, Irvine, CA, 1997.
19 A. Ramírez and D. B. Collum, J. Am. Chem. Soc., 1999, 121,
11114.
38 W. Huang, personal communication.
20 M. Majewski and P. Nowak, Tetrahedron Lett., 1998, 39, 1661.
21 S. O. Nilsson Lill, P. I. Arvidsson and P. Ahlberg, Tetrahedron:
Asymmetry, 1999, 10, 265.
22 E.-U. Würthwein, K. Behrens and D. Hoppe, Chem. Eur. J., 1999,
5, 3459.
23 P. I. Arvidsson, G. Hilmersson and P. Ahlberg, J. Am. Chem. Soc.,
1999, 121, 1883.
24 M. Amedjkouh and P. Ahlberg, Poster presented at the 11th
European Symposium on Organic Chemistry (ESOC 11), Göteborg,
Sweden, July 23–28, 1999.
39 G. I. Csonka, J. Comput. Chem., 1993, 14, 895.
40 G. I. Csonka and J. G. Angyan, J. Mol. Struct., 1997, 393, 31.
41 A. D. Becke, J. Chem. Phys., 1993, 98, 5648.
42 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B., 1988, 37, 785.
43 P. C. Hariharan and J. A. Pople, Theor. Chim. Acta, 1973, 28,
213.
44 M. J. Frisch, J. A. Pople and J. S. Binkley, J. Chem. Phys., 1984, 80,
3265.
45 T. Clark, J. Chandrasekhar, G. W. Spitznagel and P. v. R. Schleyer,
J. Comput. Chem., 1983, 4, 294.
J. Chem. Soc., Perkin Trans. 2, 2001, 1654–1661
1661