Hemi-labile ligands in organolithium chemistry: Rate studies of the LDA-mediated α- and β-metalations of epoxides

Article abstract of DOI:10.1021/ja992166+

Lithium diisopropylamide (LDA)-mediated α- and β-eliminations of epoxides are described. A comparison between LDA/n-BuOMe mixtures and LDA/MeOCH₂CH₂NMe₂ (LDA/3) mixtures reveals that the LDA dimer solvated by the "hemi-labile" amino ether 3 imparts dramatic rate accelerations, alters relative efficacies of monomer- and dimer-based metalations, and influences the partitioning between α- and β-metalation. The α-elimination of exo-norbornene oxide by LDA/n-BuOMe and LDA/3 proceeds exclusively via a dimer-based pathway with a transition structure, [(R₂NLi)₂(epoxide)(ligand)] ?. The β-elimination of tetramethylethylene oxide by LDA/n-BuOMe and LDA/3 proceeds exclusively via a monomer-based pathway with a transition structure, [(R₂NLi)(epoxide)(ligand)]?. cis-Cyclooctene epoxide reacts with LDA/3 to give two products: (1) a bicyclooctanol derived from an α-metalation and a dimer-based transition structure, [(R₂NLi)₂(epoxide)(ligand)]?, and (2) 2-cycloocten-l-ol derived from both a monomer-based transition structure, [(R₂NLi)(epoxide)(ligand)]?, and a dimer-based transition structure, [(R₂NLi)₂(epoxide)(ligand)]?. Possible geometries of the transition structures are discussed. Hemi-labile ligands that chelate lithium only at the rate-limiting transition structures maximize both the reaction rates and the mechanistic transparency.

Full text of DOI:10.1021/ja992166+

11114
J. Am. Chem. Soc. 1999, 121, 11114-11121
Hemi-Labile Ligands in Organolithium Chemistry: Rate Studies of
the LDA-Mediated R- and â-Metalations of Epoxides
Antonio Ram ´ı rez and David B. Collum*
Contribution from the Department of Chemistry and Chemical Biology, Baker Laboratory,
Cornell UniVersity, Ithaca, New York 14853-1301
ReceiVed June 24, 1999
Abstract: Lithium diisopropylamide (LDA)-mediated R- and â-eliminations of epoxides are described. A
comparison between LDA/n-BuOMe mixtures and LDA/MeOCH2CH2NMe2 (LDA/3) mixtures reveals that
the LDA dimer solvated by the “hemi-labile” amino ether 3 imparts dramatic rate accelerations, alters relative
efficacies of monomer- and dimer-based metalations, and influences the partitioning between R- and
â-metalation. The R-elimination of exo-norbornene oxide by LDA/n-BuOMe and LDA/3 proceeds exclusively
q
via a dimer-based pathway with a transition structure, [(R2NLi)2(epoxide)(ligand)] . The â-elimination of
tetramethylethylene oxide by LDA/n-BuOMe and LDA/3 proceeds exclusively via a monomer-based pathway
q
with a transition structure, [(R2NLi)(epoxide)(ligand)] . cis-Cyclooctene epoxide reacts with LDA/3 to give
two products: (1) a bicyclooctanol derived from an R-metalation and a dimer-based transition structure,
q
[
[
(R2NLi)2(epoxide)(ligand)] , and (2) 2-cycloocten-1-ol derived from both a monomer-based transition structure,
q
q
(R2NLi)(epoxide)(ligand)] , and a dimer-based transition structure, [(R2NLi)2(epoxide)(ligand)] . Possible
geometries of the transition structures are discussed. Hemi-labile ligands that chelate lithium only at the rate-
limiting transition structures maximize both the reaction rates and the mechanistic transparency.
Introduction
envisioned by simply reversing the logic: a difunctional ligand
1
that forms a weakly coordinated monodentate (η ) linkage in
Difunctional ligands containing both weakly and strongly
coordinating appendages, so-called hemi-labile ligands, have
received considerable attention in the transition metal literature.
Dissociating the weakly ligating group affords coordinatively
unsaturated metal centers at a relatively low energetic cost (eq
2
the reactant and a strongly coordinated bidentate (η ) linkage
in the rate-limiting transition structure should afford dramatic
rate accelerations. Indeed, lithium diisopropylamide (LDA)
dimer 1 ligated by hemi-labile amino ether 3 effects dehydro-
1
3
halogenations 10 times faster than does dimer 4 solvated by
1).
4
n-BuOMe (eq 2). The binding constants of n-BuOMe and
(1)
Although one might question the merits of using chelating
(2)
ligands in the first place, hemi-labile ligands appear to offer
advantages over their monodentate or other bidentate counter-
2
parts in some cases.
In contrast, hemi-labile ligands have received no explicit3
attention in organolithium chemistry. This may be due to the
generally limited development of the coordination chemistry
of lithium and to the belief that ligand dissociations are not
important. However, reactions in which the rate-limiting transi-
tion structures are more highly solvated than the reactants offer
ideal opportunities to exploit hemi-labile ligands. This is easily
amino ether 3 were shown to be identical, and rate studies
revealed that the reaction proceeds via monosolvated LDA
(1) For leading references to hemi-labile ligands, see: Slone, C. S.;
Weinberger, D. A.; Mirkin, C. A. Prog. Inorg. Chem. 1999, 48, 233. Bader,
A.; Lindner, E. Coord. Chem. ReV. 1991, 108, 27.
(2) For example, see: Lindner, E.; Sickinger, A.; Wegner, Z. J.
Organomet. Chem. 1998, 349, 75. Abu-Gnim, C.; Amer, I. J. Mol. Catal.
993, 85, L275. Butgess, K.; Ohlmeyer, M. J.; Whitmire, K. H. Organo-
metallics 1992, 11, 3588. Tsuda, T.; Morikawa, S.; Saegusa, T. J. Org.
Chem. 1990, 55, 2978. Vineyard, B. D.; Knowles, W. S.; Sbacky, M. J.;
Bachman, G. L.; Weinkauff, D. J. J. Am. Chem. Soc. 1977, 99, 5946.
1
monomers in both cases. Consequently, the high rates observed
when dimer 1 is used stem from stabilization of the transition
structure by chelation without offsetting stabilization of the
(3) It is quite possible that some of the dramatic rate accelerations
observed when N,N,N′,N′-tetramethylethylenediamine (TMEDA) is present
stem from hemi-lability. Also, see: Collum, D. B. Acc. Chem. Res. 1992,
(4) Remenar, J. F.; Lucht, B. L.; Collum, D. B. J. Am. Chem. Soc. 1997,
119, 5567. Remenar, J. F.; Collum, D. B. J. Am. Chem. Soc. 1997, 119,
5573.
4
2
5, 448.
1
0.1021/ja992166+ CCC: $18.00 © 1999 American Chemical Society
Published on Web 11/17/1999

Hemi-Labile Ligands in Organolithium Chemistry
J. Am. Chem. Soc., Vol. 121, No. 48, 1999 11115
Scheme 1
We describe herein rate and mechanistic investigations of
three LDA-mediated epoxide rearrangements. A particularly
1
1,13
interesting
rearrangement of cis-cyclooctene oxide (5) is
shown in eq 3. Metalations of 5 using LDA solvated by
(3)
monodentate ethereal ligands such as Et2O11 and n-BuOMe
afford exclusively bicyclooctane 6 very slowly. In contrast,
treatment of 5 with LDA dimer 1 containing coordinated amino
ether 3 in hexane rapidly affords 6 along with minor quantities
of cyclooctenol 7. Interestingly, the [6]:[7] ratio depends on
the concentration of ligand 3.
To fully understand the bifurcated pathway in eq 3 we also
investigated two very clean, seemingly related eliminations: (1)
the LDA-mediated elimination of epoxide 8 (eq 4), which
reactant. Isolating the chelate effect exclusively to the rate-
limiting transition structure also eliminates the problem of
deconvoluting ground state and transition state chelation effects,
offering an unusually clear view of how chelation influences
reactivity.
To further explore the concept of hemi-lability in organo-
lithium chemistry, we have examined the LDA-mediated
reactions of epoxides. The lithium dialkylamide-mediated rear-
rangements of epoxides have been investigated extensively over
the past several years due to their synthetic and theoretical
interest.5 Both R- and â-metalations (Scheme 1) are known to
be sensitive to the choice of substrate, lithium amide, and
solvent. Deuterium-labeling experiments have confirmed the
(4)
,6
certainly proceeds via a vicinal elimination,14 and (2) the LDA-
mediated elimination of norbornene epoxide 10 (eq 5), which
7
stereo- and regiochemistry of metalation in many cases.
Whereas the â-metalation (vicinal elimination) is usually
(
5)
7a,c
implicated in the formation of allylic alcohols, R-metalations
and the resulting oxacarbenoids7 can lead to transannular C-H
,8
9
-11
bond insertions to give saturated alkoxides,
vicinal C-H
12
appears to proceed via an R-metalation/oxacarbenoid pathway.10
By using both LDA/3 and LDA/n-BuOMe mixtures for the
eliminations in eq 3-5, we will document the origins of the
regioselectivity as well as the merits of hemi-labile ligand 3.
insertions to give lithium enolates, or vicinal C-H insertions
to form allylic alkoxides.7
a,d
(
5) Reviews: Satoh, T. Chem. ReV. 1996, 96, 3303. Crandall, J. K.;
Apparu, M. Org. React. 1983, 29, 345.
6) For leading references to asymmetric eliminations of epoxides, see:
(
O’Brien, P. J. Chem. Soc., Perkin Trans. 1 1998, 1439. Hodgson, D. M.;
Gibbs, A. R.; Lee, G. P. Tetrahedron 1996, 52, 14361. Cox, P. J.; Simpkins,
N. S. Tetrahedron: Asymm. 1991, 2, 1.
(7) (a) Morgan, K. M.; Gajewski, J. J. J. Org. Chem. 1996, 61, 820. (b)
Crandall, J. K.; Crawley, L. C.; Banks, D. B.; Lin, L. J. Org. Chem. 1971,
3
6, 510. (c) Thummel, R. P.; Rickborn, B. J. Am. Chem. Soc. 1970, 92,
2
064. (d) Cope, A. C.; Berchtold, G. A.; Peterson, P. E.; Sharman, S. H. J.
Am. Chem. Soc. 1960, 82, 6370.
8) Leading references to oxacarbenoids: Boche, G.; Bosold, F.; Lohrenz,
J. C. W.; Opel, A.; Zulauf, P. Chem. Ber. 1993, 126, 1873. Baumgartner,
T.; Gudat, D.; Nieger, M.; Niecke, E.; Schiffer, T. J. J. Am. Chem. Soc.
(
Results
General. LDA was prepared and recrystallized as described
1
999, 121, 5953.
9) McDonald, R. N.; Steppel, R. N.; Cousins, R. C. J. Org. Chem. 1975,
0, 1694. Crandall, J. K.; Chang, L.-H. J. Org. Chem. 1967, 32, 532. Cope,
1
5
previously and exists as a disolvated dimer (1) under all
(
4
4
conditions studied herein. Pseudo-first-order conditions were
A. C.; Lee, H.-H.; Petree, J. E. J. Am. Chem. Soc. 1958, 80, 2849. Cope,
A. C.; Brown, M.; Lee, H.-H. J. Am. Chem. Soc. 1958, 80, 2855.
established at normal LDA concentrations (0.04-0.40 M) by
restricting the epoxide concentrations to e0.004 M. LDA
concentrations refer to the concentration of the monomer subunit
(10) Hodgson, D. M.; Lee, G. P.; Marriott, R. E.; Thompson, A. J.;
Wisedale, R.; Witherington, J. J. Chem. Soc., Perkin Trans. 1 1998, 2151.
Hodgson, D. M.; Marriott, R. E. Tetrahedron Lett. 1997, 38, 887. Crandall,
J. K. J. Org. Chem. 1964, 29, 2830.
(
normality). The ligand concentrations refer to the concentration
of free ligand by excluding lithium-bound ligand on LDA dimers
1 and 4. The reaction rates were monitored by following loss
of the epoxides and formation of products of quenched samples.
(11) Apparu, M.; Barrelle, M. Tetrahedron 1978, 34, 1541. Boeckman,
R. K. Tetrahedron Lett. 1977, 49, 4281. Whitesell, J. K.; White, P. D.
Synthesis 1975, 602. Cope, A. C.; Lee, H. H.; Petree, H. E. J. Am. Chem.
Soc. 1958, 80, 0, 2849.
(
12) Yanagisawa, A.; Yasue, K.; Yamamoto, Y. Chem. Commun. 1994,
(13) Tierney, J. P.; Alexakis, A.; Mangeney, P. Tetrahedron: Asymm.
1997, 8, 1019. Asami, M.; Suga, T.; Honda, K.; Inoue, S. Tetrahedron Lett.
1997, 38, 6425. Also, see ref 10.
(14) Price, C. C.; Carmelite, D. D. J. Am. Chem. Soc. 1966, 88, 4039.
(15) Bernstein, M. P.; Romesberg, F. E.; Fuller, D. J.; Harrison, A. T.;
Williard, P. G.; Liu, Q. Y.; Collum, D. B. J. Am. Chem. Soc. 1992, 114,
5100.
2
103. Thies, R. W.; Chiarello, R. H. J. Org. Chem. 1979, 44, 1342. Bond,
F. T.; Ho, C.-Y. J. Org. Chem. 1976, 41, 1421. Thummel, R. P.; Rickborn,
B. J. Org. Chem. 1972, 37, 3919. Kissel, C. L.; Rickborn, B. J. Org. Chem.
972, 37, 2060. Thummel, R. P.; Rickborn, B. J. Org. Chem. 1972, 37,
250. Crandall, J. K.; Chang, L.-H. J. Org. Chem. 1967, 32, 435. Cope, A.
1
4
C.; Trumbull, P. A.; Trumbull, E. J. Am. Chem. Soc. 1958, 80, 2844.

11116 J. Am. Chem. Soc., Vol. 121, No. 48, 1999
Ram ´ı rez and Collum
Table 1. Summary of Rate Studies for the LDA-Mediated Elimination of 5, 8, and 10 (Eq 1)
entry
epoxide
temp, °C
ligand
LDA order
solvent order
H D
k /k
d
b
1
2
5
5
20
20
n-BuOMe
3
1.06 ( 0.02
-1.06 ( 0.06
-1.10 ( 0.01
2.7 ( 0.2
3.6 ( 0.4
a,b
0.92 ( 0.06
(
k
R(obsd) + kâ(obsd)
)
a,c
2.8 ( 0.3c
3
4
5
6
7
5
5
5
5
5
20
20
20
20
20
3
3
3
3
3
0.85 ( 0.06
(k
R(obsd) + kâ(obsd)
)
c
c
0.94 ( 0.06
-1.11 ( 0.01
(kR(obsd))
(k
R(obsd)
)
b
c
b
0.95 ( 0.06
(k )
R(obsd)
c,e
0.61 ( 0.05
-0.97 ( 0.05, 0
(kâ(obsd))
(k
â(obsd)
)
0.78 ( 0.06
(k )
â(obsd)
0.51 ( 0.04
0.47 ( 0.02
0.99 ( 0.06
1.03 ( 0.03
8
9
0
1
8
8
10
10
0
0
0
n-BuOMe
3
n-BuOMe
3
0
0
11 ( 1d
12.3 ( 0.6
d
1
1
-0.96 ( 0.05
-1.03 ( 0.05
4.3 ( 0.4
-70
6.4 ( 0.4
a
Determined by fitting the data for loss of 5 to eq 15. b [3] ) 0.5 M. c [3] ) 3.0 M. d [n-BuOMe] ) 0.5 M. e Orders correspond to dimer- and
monomer-based eliminations, respectively. (See Figure 5.)
Table 2. Relative Rate Constants for the LDA-Mediated Reaction of Epoxides 5, 8, and 10a
a
[LDA] ) 0.1 M and [ligand] ) 0.5 M in hexane.
The gas chromatographic analyses used protocols described
previously.4 The time-dependent loss of all three epoxides
followed clean first-order kinetics, providing rate constants that
are independent of the initial concentrations. Isotope effects were
determined by comparing the eliminations of epoxides 5, 8, and
,16
0 to the deuterated analogues 5-d2,17 8-d12,
14,18
and 10-d2,
19
1
respectively, and are all consistent with rate-limiting steps
involving C-H(D) bond cleavages. The rate data and isotope
effects are summarized in Table 1. Representative data are
depicted in Figures 1-5; additional data are included in
Supporting Information.
Relative Rate Constants. The eliminations of 5, 8, and 10
are highly solvent dependent as shown in Table 2. (The relative
rate constants are normalized to n-BuOMe for each substrate.)
No measurable reaction is observed in the absence of an ethereal
solvent. We have also included the results from the dehydro-
4
halogenation of exo-norbornyl bromide for comparison.
LDA/8. LDA-mediated eliminations of 8 in hexane solutions
containing ligand 3 revealed a half-order dependence on the
LDA concentration (Figure 1; Table 1, entry 9) and a zeroth-
order dependence on the concentration of ligand 3 (Figure 2;
Table 1), providing the idealized rate law in eq 6 and generalized
Figure 1. Plot of kobsd vs [LDA] in MeOCH
hexane cosolvent for the â-deprotonation of 2,3-dimethyl-2-butene oxide
8, 0.004 M) at 0 °C. The curve depicts the result of an unweighted
2 2 2
CH NMe (0.5 M) and
(
n
-3
least-squares fit to kobsd ) k[LDA] (k ) 4.6 ( 0.1 × 10 , n ) 0.47
(
0.02).
(
(
16) Remenar, J. F.; Collum, D. B. J. Am. Chem. Soc. 1998, 120, 4081.
17) Hayward, R. C.; Whitham, G. H. J. Chem. Soc., Perkin Trans. 1
mechanism in eqs 7 and 8. Studies using mixtures of LDA in
n-BuOMe/hexane solutions revealed substantially reduced rates
(Table 2), yet afforded an analogous rate law (eq 6, Table 1,
1
975, 2267.
18) Mazzocchi, P. H.; Klingler, L. J. Am. Chem. Soc. 1984, 106, 7567
and references therein.
(

Hemi-Labile Ligands in Organolithium Chemistry
J. Am. Chem. Soc., Vol. 121, No. 48, 1999 11117
Figure 4. Plot of kobsd vs [n-BuOMe] in hexane cosolvent for the
R-deprotonation of exo-norbornene oxide (10, 0.004 M) by LDA (0.10
M) at 0 °C. The curve depicts the result of an unweighted least-squares
fit to kobsd ) k[n-BuOMe] (k ) 8.6 ( 0.3 × 10 , n ) -0.96 ( 0.05).
2 2 2
Figure 2. Plot of kobsd vs [MeOCH CH NMe ] in hexane cosolvent
for the â-deprotonation of 2,3-dimethyl-2-butene oxide (8, 0.004 M)
by LDA (0.10 M) at 0 °C. The curve depicts the result of an unweighted
n
-3
least-squares fit to kobsd ) k[MeOCH
2
CH
2
NMe
2
] + k′ (k ) 1 ( 7 ×
-
6
-3
1
0
, k′ ) 1.55 ( 0.02 × 10 ).
entry 10) and affiliated mechanism (eqs 10 and 11).
LDA/5. Eliminations of 5 with LDA/n-BuOMe mixtures
provided exclusively bicyclooctanol 6 resulting from an R-
elimination. Plots of kobsd vs [n-BuOMe] and kobsd vs [LDA]
afford the idealized rate law (eq 12; Table 1, entry 1) consistent
with the dimer-based mechanism described by eqs 13 and 14.
This is analogous to the results for the R-metalation of substrate
1
0.
Figure 3. Plot of kobsd vs [LDA] in MeOCH
hexane cosolvent for the R-deprotonation of exo-norbornene oxide (10,
.004 M) at -70 °C. The curve depicts the result of an unweighted
2 2 2
CH NMe (3.0 M) and
0
n
-2
least-squares fit to kobsd ) k[LDA] (k ) 2.9 ( 0.2 × 10 , n ) 1.03
(
0.05).
The elimination of cyclooctene epoxide 5 by LDA/3 mixtures
provides two products via parallel pathways. Following loss of
[5] vs time affords (kobsd(R) + kobsd(â)) according to eq 15. kobsd(R)
entry 8) and affiliated mechanism (eqs 7 and 8).
corresponds to the pseudo-first-order rate constant for the
formation of bicyclooctanol 6, while kobsd(â) corresponds to the
pseudo-first-order rate constant for the formation of cyclooctenol
7
. The [6]:[7] ratio provides kobsd(R)/kobsd(â), in turn, affording
values for kobsd(R) and kobsd(â).
[
5] ) [5] exp{-(k
+ kobsd(â))t}
obsd(R)
(15)
0
A plot of kobsd(R) vs [LDA] ([3] ) 3.0 M) furnishes an LDA
order of 0.94 ( 0.06. A plot of kobsd(R) vs [3] affords an order
in ligand 3 of -1.11 ( 0.01. Taken together, the reaction orders
are in reasonable accord with the idealized rate law for the
formation of bicyclooctane 6 in eq 12 and the dimer-based
mechanism depicted in eqs 13 and 14.
LDA/10. LDA-mediated eliminations of 10 in hexane solu-
tions containing 3 revealed a first-order dependence on the LDA
concentration (Figure 3; Table 1, entry 11) and an inVerse-first-
order dependence on the ligand concentration (Figure 4; Table
1
), providing the idealized rate law in eq 9 and generalized
From a qualitative inspection of the data and by drawing
analogy with the elimination of 8, we anticipated that plotting
kobsd(â) vs LDA and ligand concentrations would implicate a
simple monomer-based mechanism for the formation of allylic
mechanism in eq 10 and 11. Analogous studies using mixtures
of LDA in n-BuOMe/hexane solutions revealed reduced rates
(Table 2), yet afforded an analogous rate law (eq 9; Table 1,

11118 J. Am. Chem. Soc., Vol. 121, No. 48, 1999
Ram ´ı rez and Collum
(22)
mixture at high ligand concentration (Table 2) derives from the
ligand]-dependent inhibition of the dimer-based pathways. As
[
expected, reactions of 5-d2 at high [3] revealed a significant
isotope effect accompanied by a substantial (approximately
3
-fold) decrease in the [6]:[7] product ratio due to the selective
inhibition of the dimer-based R-metalation(s). Metalation of 5-d2
at slightly sub-stoichiometric ligand concentrations (0.9 equiv
per lithium) provides dramatic rate increases due to the emerging
dominance of the dimer-based pathways; however, the substan-
tial kinetic isotope effect (kH/kD ) 3.8 ( 0.5) accompanied by
virtually no change in the 6/7 ratio strongly suggests that both
2 2 2
Figure 5. Plot of kobsd(â) vs [MeOCH CH NMe ] in hexane cosolvent
for the R- and â-deprotonation of cis-cyclooctene oxide (5, 0.004 M)
by LDA (0.10 M) at 20 °C. The curve depicts the result of an
n
unweighted least-squares fit to kobsd(â) ) k[MeOCH
2
2
CH NMe
2
] + k′
-5
-5
(
k ) 8.6 ( 0.6 × 10 , n ) -0.97 ( 0.05, k′ ) 2.1 ( 0.4 × 10 ).
6
and 7 derive from a dimer-based R-metalation and a common
intermediate.
alcohol 7. However, the results proved to be more complex than
expected. A plot of kobsd(â) vs [3] (Figure 5) revealed evidence
of a zeroth-order dependence at high ligand concentration and
an inverse dependence at low ligand concentration. A plot of
kobsd(â) vs [LDA] at 3.0 M concentration of 3 provides an order
of 0.61 ( 0.05, a value that is slightly outside the normal range
Discussion
Lithium dialkylamide-mediated eliminations of epoxides
afford a remarkably diverse range of products including allylic
alcohols, homoallylic alcohols, enolates, and a variety of
15,20
for previously measured half-order dependencies.
A plot of
kobsd(â) vs [LDA] at 0.5 M concentration of 3 affords a higher
LDA order (0.78 ( 0.06), indicating the emergence of a dimer-
based vicinal elimination at lower ligand concentrations. We
conclude that the dominant â-metalation at high ligand con-
centration is a monomer-based pathway as described by eqs
5
7a
carbenoid-derived products. Even with labeling studies and
an enormous body of empirical evidence documenting the
competing R- and â-metalations illustrated in Scheme 1, the
solvation and aggregation events underlying these reactions are
21
unclear. The reaction of cyclooctene epoxide 5 (eq 3) offers
1
6-18. The inverse ligand dependence implicates a significant
a particularly interesting opportunity to investigate two mecha-
nistically and synthetically interesting issues: (1) the mechanistic
basis of the observed regioselectivity and (2) the accelerating
influence of hemi-labile ligands (eq 2). More specifically,
treatment of cyclooctene epoxide 5 with LDA solvated by amino
ether 3 provides both 6 and 7, products subsequently shown to
derive from competitive R- and â-metalations (Scheme 1), and
does so very rapidly when compared with the reaction of 5 with
n-BuOMe-solvated LDA dimer 4. For further clarification, we
also investigated the isolated examples of R- and â-metalations
illustrated in eqs 5 and 4, respectively. For all three cases, we
investigated the consequences of competing monomer- and
dimer-based pathways as well as the rate accelerations imparted
by hemi-labile ligands.
contribution from a dimer-based â-metalation pathway as is
described by eqs 19-21. This conclusion is corroborated by
Competitive Monomer- and Dimer-Based Metalations.
The rate studies reveal a bifurcation between R- and â-meta-
lation of epoxides. The â-metalations of epoxide 8 using
n-BuOMe-solvated dimer 4 or amino ether-solvated dimer 1
follow equivalent idealized rate laws (eq 6; Table 1, entries 8
and 9), consistent with transition structures 12 and 13 based
upon monosolvated LDA monomers. The importance of che-
lation in transition structure 13 is confirmed by the 22-fold
acceleration (vide infra). In contrast, rate studies of the R-meta-
lation of epoxide 10 by n-BuOMe-solvated dimer 4 or amino
ether-solvated dimer 1 furnish the same idealized rate laws (eq
the fact that eliminating the excess ligand causes a dramatic
rate acceleration (consistent with the dominance of the dimer-
based pathway) while the [6]:[7] ratio approaches a limiting
ratio of only 6:1.
The overall rate law for the conversion of epoxide 5 to 6 and
7
(eq 3) is described by eq 22. The preference for 6 at low
ligand concentrations and the formation of an equimolar 6/7
(
19) Stille, J. K.; Sonnenberg, F. M.; Kinstle, T. H. J. Am. Chem. Soc.
966, 88, 4915.
20) (a) Galiano-Roth, A. S.; Collum, D. B. J. Am. Chem. Soc. 1989,
9
) consistent with transition structures 14 and 15 based upon
1
monosolvated LDA dimers. The importance of chelation in
(
1
1
11, 6772. (b) Bernstein, M. P.; Collum, D. B. J. Am. Chem. Soc. 1993,
15, 8008.
(21) Tobias, D.; Rickborn, B. J. Org. Chem. 1989, 54, 777.

Hemi-Labile Ligands in Organolithium Chemistry
J. Am. Chem. Soc., Vol. 121, No. 48, 1999 11119
transition structure 15 is confirmed by the 80-fold acceleration
concentrations. We also found that the measured LDA order
approaches that expected for a dimer-based pathway with
decreasing ligand concentration.
(vide infra).
We conclude that the bicyclooctanol 6 derives from a dimer-
q
based transition structure, [(R2NLi)2(3)(5)] . Allylic alcohol 7
derives from primarily a dimer-based transition structure, [(R2-
q
NLi)2(3)(5)] , at low ligand concentration and a monomer-based
q
transition structure, [(R2NLi)(3)(5)] , at high ligand concentra-
tions. The dimer-based formation of allylic alcohol 7 could
proceed by either an R-metalation (via 15) followed by
oxacarbenoid insertion or a direct â-metalation (17). Indistin-
guishable [6]:[7] ratios observed for metalations of 5 and 5-d2
using sub-stoichiometric concentrations of ligand argue strongly
in favor of 15 rather than 17 as the source of allylic alcohol 7.
When taken in conjunction with the spectroscopic studies of
LDA dimers 1 and 4, kinetics only provide the stoichiometries
of the transition structures, not the geometries. It is appropriate
to comment upon some of the liberties we have taken by
depicting a range of transition structures 12-16. A discrete
interaction of the epoxide oxygen with lithium and resulting
cyclic transition structures seems intuitively logical and reason-
7
a,22,23
ably well-supported.
As noted above and discussed more
thoroughly below, the depiction of amino ether 3 as chelated
in transition structures 13 and 15 is firmly supported by
comparisons with n-BuOMe. The open dimer motifs of 14 and
1
5 with coordination only at the terminal lithium are well-
founded by kinetic, crystallographic, spectroscopic, and com-
2
4
putational data. We will continue to depict the dimer-based
reactions of lithium amides as open dimers until something
shakes our confidence.
In light of the rate studies, why do the R-metalations proceed
via open dimer-based transition structures such as 14 and 15
rather than a monomer-based transition structure such as 16?
The answer may be simple: angle strain. Whereas transition
structures 14 and 15 may comfortably attain an approximate
The reaction of LDA with epoxide 5 is by far the most
complex of the three reactions studied, yet can be understood
when supported by results from the other two case studies.
Quantitative rate studies using LDA/n-BuOMe mixtures (LDA
dimer 4) reveal that the exclusive formation of bicyclooctanol
arises from an R-metalation involving a mono-solvated LDA
dimer; we invoke transition structure 14 based upon the open
dimer-like LDA fragment.
1
80° N-H-C bond angle believed to be optimal for proton
6
2
5
transfer, transition structure 16 containing a five-membered
ring certainly cannot.
On the Role of Hemi-Labile Ligands. One of the primary
motivations for investigating the reactions of the epoxides with
LDA was to further explore the accelerating influence of hemi-
labile ligands first developed in the context of LDA-mediated
Analogous rate studies using both LDA/3 mixtures (LDA
dimer 1) implicated a dimer-based R-metalation to give 6 and
monomer-based â-metalation to give 7. Increasing the concen-
trations of ligand 3 decreased the rate of consumption of 5,
decreased the [6]:[7] ratio, and increased the LDA reaction order.
We observed a 3-fold decrease in the [6]:[7] ratio for 5-d2,
further implicating competing R- and â-metalations. However,
two predictions stemming from such a simple two-pathway
model proved to be incorrect:
4
dehydrohalogenations. The principle of hemi-lability applied
in this context is very simple: A difunctional ligand that chelates
exclusively at the rate-limiting transition state allows the full
benefits of chelation to be exploited without offsetting ground-
state stabilization. To clarify the role of the chelate effect, we
compared the efficacy of LDA solvated by n-BuOMe, amino
ether 3, and a variety of other difunctional ligands for mediating
the eliminations in eqs 3-5. The relative rate constants are listed
in Table 2. A critical feature of the comparisons is that the
(1) In the absence of excess ligand, the reaction rate should
increase dramatically due to the acceleration of the dimer-based
pathway. The rate increases should be accompanied by the
formation of 6 as the dominant (if not exclusive) product.
(
2) Deconvolution of the rate constants for R-metalation
(
22) The cyclic model has received widespread attention in the depro-
(
kobsd(R)) and â-metalation (kobsd(â)) should reveal inverse-first-
order and zeroth-order dependencies on the ligand concentra-
tions, respectively.
tonation by chiral lithium amides: Whitesell, J. K.; Felman, S. W. J. Org.
Chem. 1980, 45, 755. Asami, M. Chem. Lett. 1984, 829. Leonard, J.; Hewit,
J. D.; Ouali, D.; Simpson, S. J. Tetrahedron Lett. 1990, 31, 6703. Bhuniya,
D.; Singh, V. K. Synth. Commun. 1994, 24, 375. Bhuniya, D.; DattaGupta,
A.; Singh, V. K. Tetrahedron Lett. 1995, 36, 2847. Bhuniya, D.; DattaGupta,
A.; Singh, V. K. J. Org. Chem. 1996, 61, 6108. Nilsson Lill, S. O.;
Arvidsson, P. I.; Ahlberg, P. Tetrahedron: Asymm. 1999, 10, 265.
While the overall rate at which 5 is consumed does indeed
increase dramatically at sub-stoichiometric ligand concentrations,
the [6]:[7] ratio approaches a limiting value of only 6:1. This
suggests that there is a dimer-based pathway leading to allylic
alcohol 7. Second, a plot of kobsd(R) vs [3] displays a zeroth-
order dependence at high ligand concentrations, but shows an
inverse dependence at low ligand concentrations. Once again,
this is consistent with a measurable contribution from dimer-
based formation of allylic alcohol 7 appearing at low ligand
(
23) Hilmersson, G.; Arvidsson, P. I.; Davidsson, O¨ . J. Am. Chem. Soc.
1
996, 118, 3539. Khan, A. Z.-Q.; de Groot, R. W.; Arvidsson, P. I.;
Davidsson, O¨ . Tetrahedron: Asymm. 1998, 9, 1223.
(24) For a bibliography of lithium amide open dimers, see: Remenar, J.
F.; Lucht, B. L.; Kruglyak, D.; Collum, D. B. J. Org. Chem. 1997, 62,
5
748. Also, see ref 16.
25) Narula, A. S. Tetrahedron Lett. 1981, 27, 4119. Bell, R. P. The
Tunnel Effect in Chemistry; Chapman and Hall: New York, 1980.
(

11120 J. Am. Chem. Soc., Vol. 121, No. 48, 1999
Ram ´ı rez and Collum
various solvated LDA dimers are isostructural and related by
thermoneutral ligand substitutions (eq 23); significant differ-
promotes the monomer-based elimination to give 7 more so than
the dimer-based elimination to give 6.
4
(
4) The dehydrohalogenation appears to benefit more from
4
chelation at the rate-limiting transition structure than do the
epoxide eliminations. We suspect that a weaker RBr-Li
interaction than epoxide-Li interaction places a higher premium
on the ancillary ligands.
Conclusion
The rate studies described herein show that R- and â-meta-
lations of epoxides stem from competing dimer- and monomer-
based pathways. This, in turn, offers potential strategies for the
control of reaction rates and selectivities through control of these
distinct pathways. As we have found on several occasions27 and
contrary to conventional wisdom, the overall reaction rates are
optimal under conditions promoting the dimer-based rather than
monomer-based pathways. The results also underscore the
advantages offered by hemi-labile ligands in organic chemistry.
Such “designer” ligands hold tremendous promise for controlling
reaction rates and selectivities within organolithium chemistry.
(
23)
ences in the rates or the mathematical forms of the rate laws
are ascribable to differences exclusiVely at the rate-limiting
transition structures. This allows us to extract relative chelate
stabilities at the transition structures directly from the relative
rate constants without having to factor in differential ground-
state stabilities.
The relative rate constants listed in Table 2 offer revealing
comparisons as follows:
Experimental Section
(
1) As expected, the portion of difunctional ligand, “L”, that
is not coordinated to the lithium of dimer 1 strongly influences
the rate for all eliminations. Of special note is that the three
epoxide metalations as well as the dehydrohalogenation show
qualitatively the same relative ordering of reactivities. This
suggests that the principles governing chelate stability at the
transition structures may be general.
Reagents and Solvents. cis-Cyclooctene oxide (5), exo-norbornene
oxide (10), n-BuOMe, MeO(CH
2 2
) O-t-Bu, and DME were obtained
14
from Aldrich. 2,3-Dimethyl-2-butene oxide (8), deuterated epoxides
17
19
5
-d
MeO(CH
to literature procedures. 2,3-Dimethyl-2-butene oxide-d (8-d ) was
2
,
and 10-d
2
,
and vicinal amino ethers MeO(CH
)
2 2
NMe
2
(3),
4
2 2
) N(CH
) , and MeO(CH ) N(CH ) were prepared according
2 5 2 2 2 4
1
2
12
18
(2) While dimethoxyethane (DME) facilitates the dehydro-
synthesized from 2,3-dimethyl-2-butene-d12 following the procedure
for the nondeuterated analogue.14 All solvents were distilled by vacuum
transfer from sodium benzophenone ketyl. The hexane still contained
halogenation (albeit to a lesser extent than amino ether 3), there
is little evidence that DME promotes the epoxide eliminations
through chelation at the transition structure. The higher aza-
philicity than oxaphilicity of monomeric lithium amides (and
possibly open dimers) has been noted previously.26
1
% tetraglyme to dissolve the ketyl. The LDA was prepared from
n-BuLi (Acros) and purified by recrystallization from hexane as
15
described previously. The diphenylacetic acid used to check solution
titers28 was recrystallized from methanol and sublimed at 120 °C under
full vacuum. Air- and moisture-sensitive materials were manipulated
under argon or nitrogen using standard glovebox, vacuum line, and
syringe techniques.
(3) The relative influences of chelation on the monomer- and
dimer-based metalations (eq 24) are somewhat enigmatic. On
Kinetics. For a kinetic run corresponding to a single rate constant,
a relatively concentrated (0.5-0.8 M) stock solution of LDA in a
ligand-hexane solution was prepared and titrated to determine the
precise concentration. The solution was diluted to a concentration
appropriate for the particular series and titrated a second time. A series
of oven-dried, nitrogen-flushed 5 mL serum vials (10 per rate constant)
fitted with stir bars were charged with the LDA stock solution and
brought to the desired temperature (( 0.2 °C) using a constant-
temperature bath fitted with a thermometer. The epoxides were added
as a 0.16 M stock solution in hexane containing dodecane (0.16 M) as
(24)
a GC standard. The vessels were periodically quenched with 1:1 H
THF at intervals chosen to ensure an adequate sampling of each of the
first three half-lives. The quenched aliquots were extracted into Et
and washed with a saturated aqueous solution of NH Cl, and the extracts
2
O-
2
O
4
were analyzed using a Hewlett-Packard GC fitted with an autoinjector
and a 60 m DB-5 column (J & W Scientific). The metalations were
monitored by following the decrease of epoxides 5, 8, and 10 relative
to the internal dodecane standard. Following the formation of the
corresponding alcohols afforded equivalent rate constants within (10%.
Rate constants were determined by a nonlinear least-squares fit using
the Scientist distributed by MicroMath. The reported errors correspond
to one standard deviation. The observed rate constants were shown to
be reproducible within (10%.
one hand, the acceleration of the putative monomer-based
eliminations of 8 and that of the dimer-based eliminations of
1
0 are nearly equal (if the different reaction temperatures are
considered). However, the ligand-dependent regioselectivities
observed in the elimination of 5 (Table 2) suggest that chelation
(27) Sun, X.; Collum, D. B. J. Am. Chem. Soc. 1999, submitted for
(
26) Lucht, B. L.; Bernstein, M. P.; Remenar, J. F.; Collum, D. B. J.
publication. Bernstein, M. P.; Collum, D. B. J. Am. Chem. Soc. 1993, 115,
789. Also, see refs 16 and 20b.
(28) Kofron, W. G.; Baclawski, L. M. J. Org. Chem. 1976, 41, 1879.
Am. Chem. Soc. 1996, 118, 10707. Lucht, B. L.; Collum, D. B. J. Am.
Chem. Soc. 1996, 118, 2217.

Hemi-Labile Ligands in Organolithium Chemistry
J. Am. Chem. Soc., Vol. 121, No. 48, 1999 11121
Acknowledgment. We acknowledge the National Science
Supporting Information Available: Tabular and graphical
Foundation Instrumentation Program (CHE 7904825 and PCM
presentation of rate data as well as general experimental methods
8018643), the National Institutes of Health (RR02002), and IBM
(PDF). This material is available free of charge via the Internet
for support of the Cornell Nuclear Magnetic Resonance Facility.
We thank the National Institutes of Health for direct support of
this work and the Spanish Ministry of Education and Culture
for a postdoctoral fellowship supporting A. Ram ´ı rez.
at http://pubs.acs.org.
JA992166+