Chelation-based stabilization of the transition structure in a lithium diisopropylamide mediated dehydrobromination: Avoiding the 'universal ground state' assumption

Article abstract of DOI:10.1021/ja970030a

Dehydrobrominations of (±)-2-exo-bromonorbornane (RBr) by lithium diisopropylamide (LDA) were investigated to determine the roles of aggregation and solvation. Elimination with LDA/n-BuOMe occurs by deaggregation of disolvated dimers via a monosolvated monomer transition structure (e.g., [i-Pr₂NLi·n-BuOme·-RBr]. In contrast, elimination by LDA- THF displays THF concentration dependencies that are consistent with parallel reaction pathways involving both mono- and disolvated monomer transition structures. Elimination is markedly faster by LDA-DME than by LDA with monodentate ligands and follows a rate law consistent with a transition structure containing a chelated monomeric LDA fragment. A number of hemilabile amino ethers reveal the capacity of different coordinating functionalities to chelate. A protocol based upon kinetic methods affords the relative ligand binding energies in the LDA dimer reactants. Separating contributions of ground state from transition state stabilization allows us to attribute the stabilizing effects of chelation exclusively to the transition structure. The importance of chelating ligands in LDA-mediated dehydrobrominations, but not in previously studied reactions of LDA, sheds light on lithium ion chelation.

Full text of DOI:10.1021/ja970030a

J. Am. Chem. Soc. 1997, 119, 5573-5582
5573
Chelation-Based Stabilization of the Transition Structure in a
Lithium Diisopropylamide Mediated Dehydrobromination:
Avoiding the “Universal Ground State” Assumption
Julius F. Remenar and David B. Collum*
Contribution from the Department of Chemistry, Baker Laboratory, Cornell UniVersity,
Ithaca, New York 14853-1301
ReceiVed January 3, 1997X
Abstract: Dehydrobrominations of (()-2-exo-bromonorbornane (RBr) by lithium diisopropylamide (LDA) were
investigated to determine the roles of aggregation and solvation. Elimination with LDA/n-BuOMe occurs by
deaggregation of disolvated dimers via a monosolvated monomer transition structure (e.g., [i-Pr2NLi‚n-BuOMe‚-
RBr]‡). In contrast, elimination by LDA-THF displays THF concentration dependencies that are consistent with
parallel reaction pathways involving both mono- and disolvated monomer transition structures. Elimination is markedly
faster by LDA-DME than by LDA with monodentate ligands and follows a rate law consistent with a transition
structure containing a chelated monomeric LDA fragment. A number of hemilabile amino ethers reveal the capacity
of different coordinating functionalities to chelate. A protocol based upon kinetic methods affords the relative ligand
binding energies in the LDA dimer reactants. Separating contributions of ground state from transition state stabilization
allows us to attribute the stabilizing effects of chelation exclusively to the transition structure. The importance of
chelating ligands in LDA-mediated dehydrobrominations, but not in previously studied reactions of LDA, sheds
light on lithium ion chelation.
The consequences of aggregation in organolithium chemistry
have been the focus of considerable discussion for over three
decades.1 While most organolithium reactants are aggregated
under the conditions employed by synthetic chemists, it is not
known in many cases whether the observable aggregates react
directly or by first deaggregating to transient lower aggregates
or monomers. Moreover, the critical role of solvation is very
poorly understood.2 Reaction rates typically display marked
dependencies on the choice of solvent or additive; we offer the
relative rate constants shown in parentheses in Chart 1 as
representative. To understand organolithium reactivity at even
a rudimentary level, one must determine the dependence of the
solvation numbers, aggregation states, and relative stabilities
of the reactants and the rate limiting transition structure(s) on
the solvent structure. In a disturbing number of instances, the
discussion focuses upon the effects that each solvent and additive
has on the rate limiting transition structures, paying no attention
to the ground state. This implicit normalization of all ground
states to a single energysa misconception we have facetiously
come to call the “universal ground state”sis completely
unjustifiable under normal circumstances. Even the notion that
solvent effects are predominantly ascribable to differential
transition structure solvation would be difficult to justify.3,4
We initiated rate and mechanistic studies of the dehydrobro-
mination of (()-2-exo-bromonorbornane (1, eq 1) mediated by
lithium diisopropylamide (LDA) solvated by different mono-
dentate ethereal ligands (disolvated dimers 2-7).5-7 The
dehydrobromination of 1 proceeds without complicating side
reactions, allowing the study of the solvent-dependent relative
rate constants in terms of reactant and transition structure
solvation numbers, aggregation numbers, and relative stabili-
(3) A particularly clear discussion of the relationship between ground
state and transition state solvation in organolithium chemistry has been
presented: Klumpp, G. W. Recl. TraV. Chim. Pays-Bas 1986, 105, 1. See
also ref 4.
(4) Collum, D. B. Acc. Chem. Res. 1993, 26, 227.
(5) Lithium dialkylamides are important reagents in synthetic organic
chemistry: d’Angelo, J. Tetrahedron 1976, 32, 2979. Heathcock, C. H. In
ComprehensiVe Carbanion Chemistry; Buncel, E., Durst, T., Eds.; Elsevi-
er: New York, 1980; Vol. B, Chapter 4. Cox, P. J.; Simpkins, N. S.
Tetrahedron: Asymmetry 1991, 2, 1. Asymmetric Synthesis; Morrison, J.
D., Ed.; Academic Press: New York, 1983; Vols. 2 and 3. Evans, D. A. In
Asymmetric Synthesis; Morrison, J. D., Ed.; Academic Press: New York,
1983; Vol. 3, Chapter 1. Snieckus, V. Chem. ReV. 1990, 90, 879. Juaristi,
E.; Beck, A. K.; Hansen, J.; Matt, T.; Mukhopadhyay, M.; Simson, M.;
Seebach, D. Synthesis 1993, 1271. Caube`re, P. Chem. ReV. 1993, 93, 2317.
Majewski, M.; Gleave, D. M. J. Orgnanomet. Chem. 1994, 470, 1.
(6) Gregory, K.; Schleyer, P. v. R.; Snaith, R. AdV. Inorg. Chem. 1991,
37, 47. Mulvey, R. E. Chem. Soc. ReV. 1991, 20, 167.
(7) For selected examples of applications and mechanistic investigations
of lithium amide mediated 1,2-eliminations, see: (a) Mass, R. J.; Rickborn,
B. J. Org. Chem. 1986, 51, 1992. (b) Kopka, I. E.; Nowak, M. A.; Rathke,
M. W. Synth. Commun. 1986, 16, 27. (c) Evans, K. L.; Prince, P.; Huang,
E. T.; Boss, K. R.; Gandour, R. D. Tetrahedron Lett. 1990, 31, 6753. (d)
Matsuda, H.; Hamatani, T.; Matsubara, S.; Schlosser, M. Tetrahedron 1988,
44, 2865. (e) Reisdorf, D.; Normant, H. Organomet. Chem. Synth. 1972, 1,
375.
X Abstract published in AdVance ACS Abstracts, May 15, 1997.
(1) Szwarc, M. Carbanions, LiVing Polymers, and Electron Transfer
Processes; Interscience: New York, 1968. Ions and Ion Pairs in Organic
Reactions; Szwarc, M., Ed.; Wiley-Interscience: New York, 1972; Vols. 1
and 2. Anionic Polymerization: Kinetics, Mechanism, and Synthesis;
McGrath, J. E., Ed.; American Chemical Society: Washington, 1981,
Chapters 1, 2, and 29. Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27,
1624. Seebach, D. In Proceedings of the Robert A. Welch Foundation
Conferences on Chemistry and Biochemistry; Wiley: New York, 1984.
Williard, P. G. In ComprehensiVe Organic Synthesis; Pergamon: New York,
1991; Vol. 1, p 1.
(2) For leading references to rate studies affording insights into transition
structure solvation numbers see: (a) Bernstein, M. P.; Collum, D. B. J.
Am. Chem. Soc. 1993, 115, 8008. (b) Schmitz, R. F.; Dekanter, F. J. J.;
Schakel, M.; Klumpp, G. W. Tetrahedron 1994, 50, 5933.
S0002-7863(97)00030-9 CCC: $14.00 © 1997 American Chemical Society

5574 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Remenar and Collum
Chart 1
ties.8,9 The results expose a pivotal role for transition structures
bearing monomeric LDA fragments, provide insights into the
“chelate effect”,10 and reveal details of transition structure
solvation separated from reactant solvation.3,4
byproduct.15,16 The ethereal ligand concentrations ([ligand] )
1.0 M-neat) were adjusted with use of toluene as an inert co-
solvent.17 The decay of 1 displayed first-order behavior over
>3 half-lives for all LDA-solvent combinations. Substantial
kinetic isotope effects (kobsd(H)/kobsd(D) ) 1.8-4.0) determined
by comparing 1 with the 3,3-dideuterio analog (1-d2)13 con-
firmed a rate limiting proton abstraction in all instances. Table
1 compiles reaction orders, activation parameters, isotope effects,
and relative solvent binding constants (KA; Vide infra) for a
variety of LDA/solvent combinations.
Rate Equation: LDA/n-BuOMe (2). The pseudo-first-order
rate constants for the dehydrobrominations of 1 by LDA in
n-BuOMe-toluene are independent of [n-BuOMe] (Figure 1)
and display an approximate half-order dependence on [LDA]
(Figure 2, Table 1). The data are consistent with the idealized
rate law shown in eq 2. While the LDA order deviates below
-d[1]/dt ) kobsd[1]
(2)
Results
such that kobsd ) k′[LDA]1/2[S]0
The structures of the LDA dimers 2-7 were determined by
6Li and 15N NMR spectroscopic analyses of [6Li,15N]LDA as
described in the preceding paper.11,12 Relative ligand binding
constants determined by independent methods in the previous
manuscript and as described below are in full accord with the
assignment of 4-7 as η1, oxygen-bound solvates.
the theoretical value of 0.5, such deviations have been noted in
previous rate studies2a,18 and seem mechanistically inconse-
quential. The isotope effect (Table 1) is independent of the
n-BuOMe concentration. Overall, the data are consistent with
The rate of LDA-mediated dehydrobrominations of (()-2-
exo-bromonorbornane13 (1, eq 1) was monitored via GC analysis
of quenched aliquots following the decrease of 1 relative to an
internal undecane standard. The LDA was prepared and
recrystallized as described previously.14 The LDA concentra-
tions ([LDA] ) 0.025-0.40 M) were maintained high relative
to 1 (0.004 M) to ensure pseudo-first-order conditions. These
conditions also precluded formation of substantial concentrations
of mixed aggregates arising from incorporation of the LiBr
(15) (a) For an example of autocatalysis of an N-alkylation by LiBr,
see: Depue, J. S.; Collum, D. B. J. Am. Chem. Soc. 1988, 110, 5524. (b)
For leading references to apparent lithium amide mixed aggregation effects,
see: Hall, P.; Gilchrist, J. H.; Collum, D. B. J. Am. Chem. Soc. 1991, 113,
9571. Romesberg, F. E.; Collum, D. B. J. Am. Chem. Soc. 1994, 116, 9198.
(c) For additional lithium amide mixed aggregation effects, see: Seebach,
D. Synthesis 1993, 1271. Cox, P. J.; Simpkins, N. S. Tetrahedron:
Asymmetry 1991, 2, 1. Hilmersson, G.; Davidsson, O. Organometallics 1995,
14, 912. Ruck, K. Angew. Chem., Int. Ed. Engl. 1995, 34, 433. Nichols, M.
A.; Waldmuller, D.; Williard, P. G. J. Am. Chem. Soc. 1994, 116, 1153.
Goralski, P.; Legoff, D.; Chabanel, M. J. Organomet. Chem. 1993, 456, 1.
Bunn, B. J.; Simpkins, N. S.; Spavold, Z.; Crimmin, M. J. J. Chem. Soc.,
Perkin Trans. 1 1993, 3113. Strzalko, T.; Seyden-Penne, J.; Wartski, L.;
Froment, F.; Corset, J. Tetrahedron Lett. 1994, 35, 3935. Sorger, K.;
Schleyer, P. v. R.; Fleischer, R.; Stalke, E. J. Am. Chem. Soc. 1996, 118,
6924. Bunn, B. J.; Simpkins, N. S. J. Org. Chem. 1993, 58, 533. Mair, F.
S.; Clegg, W.; Oneil, P. A. J. Am. Chem. Soc. 1993, 115, 3388. Williard,
P. G.; Liu, Q.-Y. J. Am. Chem. Soc. 1993, 115, 3380. Sugasawa, K.; Shindo,
M.; Noguchi, H.; Koga, K. Tetrahedron Lett. 1996, 37, 7377.
(16) Elimination of 0.1 M 1 by 0.1 M LDA (1.0 equiv) in THF shows
no sign of autocatalysis despite the buildup of substantial concentrations
of LiBr.
(8) Bartsch, R. A.; Lee, J. G. J. Org. Chem. 1991, 56, 212.
(9) For reviews describing base-mediated eliminations, see: Bartsch, R.
A.; Zavada, J. Chem. ReV. 1980, 80, 453. Caube`re, P. Chem. ReV. 1993,
93, 2317.
(10) Cotton, F. A.; Wilkinson, G. AdVanced Inorganic Chemistry, 4th
ed.; Wiley: New York, 1980; p 71.
(11) Remenar, J. F.; Lucht, B. L.; Collum, D. B. J. Am. Chem. Soc.
1997, 119, 5567.
(12) Collum, D. B. Acc. Chem. Res. 1993, 26, 227.
(13) Brown, H. C.; Krishnamurthy, S. J. Am. Chem. Soc. 1972, 94, 7159.
Creary, X.; Hatoum, H. N.; Barton, A.; Aldridge, T. E. J. Org. Chem. 1992,
57, 1887. See ref 8.
(14) Kim, Y.-J.; Bernstein, M. P.; Galiano-Roth, A. S.; Romesberg, F.
E.; Williard, P. G.; Fuller, D. J.; Harrison, A. T.; Collum, D. B. J. Org.
Chem. 1991, 56, 4435.
(17) Toluene appears to participate in both the primary and higher
solvation shells of lithium ions.23,41
(18) Galiano-Roth, A. S.; Collum, D. B. J. Am. Chem. Soc. 1989, 111,
6772.

Chelation-Based Stabilization of Transition Structure
Table 1. Summary of Rate Studies for the LDA-mediAted Dehydrobromination of (()-2-exo-Bromonorbornane (eq 1)a
entry ligand (Chart 1)
LDA order solvent orderb ∆G°acta (kcal/mol) ∆H°act (kcal/mol) ∆S°act (cal/mol‚deg) KA(Sâ/SR)c
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5575
kH/kD
1
2
3
4
5
6
7
A (n-BuOMe)
B (THF)
0.40 ( 0.04
0.39 ( 0.02
0
21.1 ( 2.0
20.3 ( 0.5
21.1 ( 1.5f
19.1 ( 0.1
17.6 ( 0.3
18.0 ( 0.6
18.6 ( 0.3
13.0 ( 1.1
11.5 ( 0.3
13.1 ( 0.8f
g
9.1 ( 0.2
9.5 ( 0.3
8.9 ( 0.3
-31 ( 4
-35 ( 1
-32 (3f
g
-33 ( 1
-34 ( 1
-39 ( 1
1.00
1.9 ( 0.1
0,1d
0, 1d
0
2.5 ( 0.1
B (THF/toluene) 0.48 ( 0.01e
2.0 ( 0.1e
E (DME)
0.55 ( 0.02
0.55 ( 0.02
0.59 ( 0.02
0.54 ( 0.01
0.93(0.14h 2.8 ( 0.3
G
K
H
0
0
0
1.02(0.09
4.3 ( 0.5
0.81 ( 0.11 4.0 ( 0.5
0.95 ( 0.15 3.2 ( 0.5
a All measurements were made in neat donor solvent at -20 °C unless noted otherwise. b Zeroth order is assigned due to the <10% rise in rate
over >10-fold changes in donor solvent concentrations. c The equilibrium constants (KA(Sâ/SR)) for the ligand (Sâ) were determined relative to
n-BuOMe (SR) at -20 °C according to Scheme 1 and eq 10. d See Figure 3. e Values measured in 2.0 M THF/toluene solution. f Derived by
extrapolation to zero free THF concentration (see text). The value of ∆G°act is obtained from ∆H°act and ∆S°act. g Decomposition at high temperatures
and low reactivity at low temperatures precluded determination. h Fit to eq 9.
Figure 1. Plot of kobsd versus [n-BuOMe] in toluene co-solvent for
Figure 3. Plot of kobsd versus [THF] in toluene co-solvent for the
the elimination of 1 (0.004 M) by LDA (0.10 M) at 20 ( 0.1 °C. The
elimination of 1 (0.004 M) by LDA (0.10 M) at 20 ( 0.1 °C. The
curve depicts the result of an unweighted linear least-squares fit to f(x)
curve depicts the result of an unweighted linear least-squares fit to f(x)
) a.
) a + bx. (See eq 5.) The values of the parameters were calculated to
be a ) 0.24 ( 0.01 and b ) 1.17 ( 0.08.
cies,19 or (2) a single pathway that does not formally require a
second coordinated THF to lithium at the transition structure,
but is sensitive to the surrounding medium. Replacing the
toluene co-solvent with either 2,2-dimethyltetrahydrofuran (2,2-
Me2THF) or hexane revealed no measurable co-solvent depen-
dence,20 arguing against long-range medium effects as the source
of the [THF] dependence.21 The fractional LDA orders in 2.0
M THF and 12.3 M (neat) THF implicate dimer-monomer pre-
equilibria at all THF concentrations. The kinetic isotope effects
(six measurements in total) increase monotonically over the THF
concentration interval from 2.0 M to neat THF (Table 1). The
rate data are consistent with the idealized rate law in eq 5 and
implicate competing mechanisms based upon monosolvated
monomers (eqs 3 and 4) and disolvated monomers (eqs 6 and
7).
Figure 2. Plot of kobsd versus [LDA] for the elimination of 1 (0.004
M) by LDA in n-BuOMe (2.0 M) with toluene co-solvent at 20 ( 0.1
-d[1]/dt ) kobsd[1]
(5)
°C. The curve depicts the result of an unweighted nonlinear least-squares
fit to f(x) ) axb (b ) 0.40 ( 0.04; see Table 1).
such that kobsd ) k′[LDA]1/2 + k′′[THF][LDA]1/2
a single reaction pathway involving monosolvated monomers
1/2(i-Pr2NLi‚S)2 + S U i-Pr2NLiS2
i-Pr2NLiS2 + 1 f product
(6)
(7)
(eqs 3 and 4) at all n-BuOMe concentrations.
1/2(i-Pr2NLi‚S)2 U i-Pr2NLi‚S
i-Pr2NLi‚S + 1 f norbornene
(3)
(4)
Rate Equation: LDA/DME (4). The dehydrobromination
of 1 is 50-times faster by LDA dimer 4 bearing non-chelated
(η1) DME ligands than by LDA/n-BuOMe. Following the loss
of 1 at -20 °C, we obtained a zeroth-order DME dependence
(19) A fit of kobsd vs [THF] to the expression kobsd ) k′ + k′′[THF]n
affords n ) 1.02 ( 0.15.
(20) At low (<4.0 M) THF concentrations, deviations ascribable to a
ligand substitution by 2,2-Me2THF begin to appear.
Rate Equation: LDA/THF (3). A potentially different
mechanistic picture emerges from rate studies in THF solutions.
A plot of kobsd vs [THF] in toluene (Figure 3) shows a linear
dependence with a significant non-zero intercept. We consid-
ered two explanations for this behavior: (1) competing parallel
pathways that manifest zeroth- and first-order THF dependen-

5576 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Remenar and Collum
Scheme 1
and a fractional LDA order consistent with the idealized rate
law in eq 2 and the mechanism generally described by eqs 3
and 4. The substantially higher rate compared with that
observed with monodentate ethers strongly implicates the “i-
Pr2NLi.S” monomer (eq 4) to be 8 bearing a chelating DME.
Relative binding constant determinations for n-BuOMe and
DME (Vide infra) confirm that the rate accelerations stem
entirely from differential transition state energies.
Figure 4.
respectively. The rate constants measured in the neat solvents
provide the free energies of activation (∆G°(A) and ∆G°(B)).
Equation 8 describes kobsd in terms of mechanistic constants and
Rate Equation: LDA/Amino Ethers (5-7). Systematic
variations of the dialkylamine moiety of the amino ethers of
general structure R2N(CH2)2OMe revealed highly structure-
dependent elimination rates (Chart 1) discussed below.22 Rate
laws for several representative amino ethers confirmed that the
mechanism described by eqs 2-4 predominates (Table 1).
Relative Ligand Binding Constants. The NMR spectro-
scopic studies together with results from previous
investigations2a,23 support the assignment of the disolvated
dimers 3-7 bearing η1 (methoxy-coordinated) rather than
chelated ligands. The interpretation of the relative rates hinges
upon the assumption that the dimers 3-7 are related by
thermoneutral ligand substitution. This was confirmed as
described below.24
[Sâ]
1/2
kobsd ) k1KB1/2 + k2KD 2KA
×
{
(
)}
[SR]
1/2
AT
(8)
2
[Sâ]
A[SR]
[Sâ]
[SR]
1 + K
+ 2KA
{
}
(
)
solvent concentrations. (See Supporting Information.) Substitut-
ing for [Sâ] and [SR] in terms of their mole fractions, X2 and 1
- X2, respectively, affords eq 9. Equation 9 includes provisions
for the mixed solvate ii and allows the binding constant KA to
be found by simple numerical methods.26 In the event that the
rate in neat SR is not appreciable, eq 9 simplifies to give eq 10.
Consider, for example, the equilibria in Scheme 1 and the
simple thermochemical picture in Figure 4. (To simplify visual
retrieval, the mixed solvate ii found in Scheme 1 has been
omitted from Figure 4.25a) In the limit of either neat SR or neat
Sâ, the two limiting observable species are dimers i and iii,
[X2]
1/2
kobsd ) k1KB1/2 + k2KD 2KA
×
(21) During the course of hydrazone metalation rate studies, a seemingly
similar first-order dependence on the THF concentration was attributed to
medium effects rather than primary shell solvation due to a co-solvent
dependence.18
(22) Endo, K.; Otsu, T. Macromol. Rapid Commun. 1994, 15, 233.
Poshyachinda, S.; Edwards, H. G. M.; Johnson, A. F. Polymer 1991, 32,
338. Bywater, S.; Black, P.; Worsfold, D. J.; Schue, F. Macromolecules
1985, 18, 335. Fraenkel, G.; Geckle, M. J.; Kaylo, A.; Estes, D. W. J.
Organomet. Chem. 1980, 197, 249. Sugahara, K.; Fujita, T.; Watanabe, S.;
Hashimoto, H. J. Chem. Technol. Biotechnol. 1987, 37, 95. Fujita, T.;
Watanabe, S.; Suga, K.; Sugahara, K.; Tsuchimoto, K. Chem. Ind. 1983,
167.
(23) Lucht, B. L.; Collum, D. J. Am. Chem. Soc. 1996, 118, 10707.
(24) A similar binding constant analysis was described previously.2a
However, several simplifying algebraic assumptions, while not seriously
flawed, seem less appropriate in hindsight.
(25) (a) If solvation is non-cooperative, then the ground and transition
state corresponding to the mixed solvated LDA will correspond to the mean
value of the homosolvated ground states and transition states (respectively)
in Figure 4. (b) Benson, S. W. J. Am. Chem. Soc. 1958, 80, 5151.
{
(
)}
[1 - X2]
1/2
AT
(9)
2
[X2]
[X2]
1 + K
+ 2KA
{
}
A[1 - X2]
[1 - X2]
(
)
[X2]
1/2
kobsd ) k2KD 2KA
×
(
)
[1 - X2]
1/2
AT
(10)
2
[X2]
[X2]
1 + K
+ 2KA
{
}
A[1 - X2]
[1 - X2]
(
)
(26) Application of eq 10 causes division by zero at X2 ) 1. This problem
is adequately alleviated by substituting [1.01 - X2] for [1 - X2].

Chelation-Based Stabilization of Transition Structure
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5577
highly curved portion of the function cannot be adequately
sampled.
The results are as follows. Figure 6 shows the measured
pseudo-first-order rate constants for the elimination of 1 plotted
as a function of the Me2NCH2CH2OMe mole fraction in
n-BuOMe. Fitting the data to eq 11 (SR ) n-BuOMe and Sâ )
Me2NCH2CH2OMe) affords KA(Me2NCH2CH2OMe/n-BuOMe)
) 1.02 ( 0.09, corresponding to a ground state free energy
difference, ∆G°(C), of <0.05 kcal/mol. Similar investigations
of representative amino ethers revealed equivalent binding
constants (Table 1).28 The 25% reduction in the binding
constant for pyrrolidine-containing amino ether K compared to
the other methyl ethers is too small to warrant further discus-
sion.29
Figure 5. Predicted rate constants (kobsd) vs mole fraction (X2) of
solvent Sâ in co-solvent SR according to eq 9. Assumptions: (kobsd in
neat Sâ)/(kobsd in neat SR) ) 10; KA ) 10 (curve A), 1.0 (curve B), and
0.1 (curve C).
The calculated value of KA provides a measure of the relative
free energies of the two ground states corresponding to dimers
i and iii (∆G°(C), Figure 4). The estimated free energies of
activation and relative ground state free energies, in turn, provide
the difference in the transition state energies (i.e., ∆G°(D)) to
complete the thermochemical picture in Figure 4.
Temperature-Dependent Rates and Activation Param-
eters. The solvent-dependent elimination rates were monitored
over 30-50 °C temperature ranges to obtain activation enthal-
pies (∆H°act) and entropies (∆S°act) for several cases (Table 1)
by using the linear form of the Eyring equation (eq 12). For
This strategy for investigating LDA solvation offers both
qualitative and quantitative insight. Figure 5 shows theoretical
curves assuming a 10-fold relative rate difference in neat donor
solvents (kobsd(Sâ)/kobsd(SR) ) 10) for three relative binding
affinities: KA ) 10 (curve A), 1.0 (curve B), and 0.1 (curve
C). Curve A will result if the superior donor solvent (Sâ) affords
kinetically reactive species (as commonly suggested).25 On the
other hand, curve C will result if the stronger solvent (SR in
this case) affords kinetically less reactive species. When the
two ligands bind equally, the inflection point will be nearly
centered on the x-axis as in curve B.
ln(kobsdh/kT) ) -∆H°act/RT + ∆S°act/R
(12)
the reactions described by eqs 3 and 4 involving monosolvated
monomer fragments, kobsd is a composite rate constant corre-
sponding to k′[LDA]1/2[S]0 (eq 2). We will exploit the data in
the forthcoming discussion to better understand transition
structure stabilization by chelation. The data are of reasonable
quality as exemplified by Figure 7. We hasten to add that,
without a more rigorous treatment,30 such activation parameters
(especially ∆S°act) are susceptible to systematic error and should
be viewed with caution.
The activation parameters for the elimination of 1 by LDA
in THF and THF-toluene solutions were investigated with three
protocols to separate contributions from pathways involving
monosolvated monomers (eqs 3 and 4) and disolvated monomers
(eqs 6 and 7).
(1) Plotting kobsd vs [THF] (as in Figure 3) with extrapolation
to the y-intercepts affords kobsd at zero free THF concentration.
Plotting the extrapolated values of kobsd vs 1/T at four temper-
atures afforded the activation parameters corresponding to the
zeroth-order pathway (Table 1, entry 3).
(2) Determination of kobsd as a function of temperature in neat
THF (Figure 8; Table 1, entry 2) provides values for ∆S°act
and ∆H°act which, based upon the rate studies described above,
represent the total contributions of the pathways involving
mono- and disolvated LDA monomers. The substantial decrease
in ∆H°act compared to the value obtained by the protocol
Before considering the experimental results, it is important
to understand the underlying assumptions and approximations.
(1) If the magnitudes of the y-intercepts in Figure 5 are
fortuitously equivalent, the consequent failure to observe a
change in the rate constant renders the method inoperative. In
this instance, one might erroneously infer that systematic
increases in [Sâ] do not affect the reactant structure or
mechanism. (2) If the metalation mechanisms that dominate
in the neat donor solvents have non-zero solvent reaction orders,
the mathematical description would become more complex than
described by eq 9. Fortunately, the eliminations described
herein are zeroth order in donor solvent (except THF). (3) The
precise shape of the function and the accuracy of the value of
KA depend on the assumed solvation states of i-iii. If one or
more of the solvation state assignments are incorrect, the
mathematical form of eq 9 will be incorrect. For hindered
lithium amide dimers, the assignments are based upon compel-
ling evidence.12,27 (4) The model assumes non-cooperative
solvation on mixed solvated dimer ii. In other words, the dimer
subunit exchanges in eq 11 are presumed to be thermoneutral.
Recent investigations of LiHMDS support this assumption.27
The factor of 4 in the second ligand substitution (e.g., 4KA)
stems from statistical contributions in a multi-site ligand
substitution.25b (5) When measuring pseudo-first-order rate
constants with use of solvent mixtures, both solvents must be
in excess (g10 equiv per Li) to avoid substantial errors in the
estimates of free (as opposed to total) solvent concentration.
The error becomes large when KA . 10 or KA , 0.1 since the
(28) The binding constant quoted for DME (KA, Table 1) has been
adjusted to account for the statistical bias imparted by two equivalent
methoxy groups.
(29) The measured binding constant can be quite sensitive to the
measured end points (rate constants in neat donor solvent). Neglecting the
endpoint corresponding to neat MeOCH2CH2N(CH2)4 (in a 9 point plot)
affords KA(MeOCH2CH2N(CH2)4/n-BuOMe) ) 1.02 ( 0.10.
(30) Lyons, B. A.; Pfeifer, J.; Peterson, T.; Carpenter, B. K. J. Am. Chem.
Soc. 1993, 115, 2427.
(27) Lucht, B. L.; Collum, D. B. J. Am. Chem. Soc. 1995, 117, 9863.

5578 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Remenar and Collum
Figure 6. Observed rate constants for dehydrobromination of 1 by
LDA vs mole fraction of Me2NCH2CH2OMe (X2) in n-BuOMe under
the following conditions: -20 °C; [LDA] ) 0.10 M; [1] )0.004 M.
The curve corresponds to a nonlinear least-squares fit to eq 10, affording
Figure 8. Observed pseudo-first-order rate constants (kobsd) measured
as a function of temperature over a 30 °C temperature range (0 to 30
°C) for the elimination of 1 (0.004 M) by LDA (0.10 M) solvated by
neat THF. The curve corresponds to a linear least-squares fit to the
following equation: ln{kobsdh/kT} ) -∆H°act/RT + ∆S°act/R. The
activation parameters (Table 1) were derived from a nonlinear least-
squares fit to the following expression: kobsd ) kT/h[exp(-∆H°act/RT)-
exp(∆S°act/R)].
KA ) 1.02 ( 0.09 and k2KD ) 1.05 ( 0.02 × 10-2
.
Figure 7. Observed pseudo-first-order rate constants (kobsd) measured
as a function of temperature over a 30 °C temperature range (-20 to
-50 °C) for the elimination of 1 (0.004 M) by LDA (0.10 M) solvated
by Me2NCH2CH2OMe (2.0 M) in toluene. The curve corresponds to a
linear least-squares fit to the following equation: ln{kobsdh/kT} )
-∆H°act/RT + ∆S°act/R. The activation parameters (Table 1) were
derived from a non-linear least-squares fit to the following expres-
sion: kobsd ) kT/h[exp(-∆H°act/RT)exp(∆S°act/R)].
Figure 9. Observed pseudo-first-order rate constants (kobsd) measured
as a function of temperature over a 30 °C temperature range (0 to 30
°C) for the elimination of 1 (0.004 M) by LDA (0.10 M) in neat THF.
The curve corresponds to a non-linear least-squares fit to the following
expression: kobsd ) kT/h{[exp(-∆H°act1/RT) exp(∆S°act1/R)] + [exp-
(-∆H°act2/RT) exp(∆S°act2/R)]}. The values of the activation parameters
∆H°act2 and ∆S°act2 were determined by fitting the data for the reactions
in neat THF to the above equation, where the values of ∆H°act1 and
described in (1) is consistent with the emergence of a second,
mechanistically distinct pathway.
∆So
are the activation parameters determined from the data
act1
extrapolated to free THF concentration.
(3) To dissect the activation parameters into contributions
from the parallel reaction pathways, we substitute into the two-
term rate equation (eq 5) to obtain eq 13, where ∆S°act(1) and
studied extensively by other investigators.8,31 Cursory examina-
tion of the exo selectivity with monodeuterated bromides endo-
1-d1 and exo-1-d1 reveals >90% syn-exo proton abstraction by
LDA/n-BuOMe, LDA/THF, LDA/DME, and LDA/MeOCH2-
CH2NMe2 as shown by GC-MS analysis of the product.31
kobsd ) kT/h{exp(-∆H°act(1)/RT) exp(∆S°act(1)/R)} +
kT/h{exp(∆H°act(2)/RT) exp(∆S°act(2)/R)} (13)
∆H°act(1) correspond to the k′[LDA]1/2 term, and ∆S°act(2) and
∆H°act(2) correspond to the k′′[THF][LDA]1/2 term. Since
∆S°act(1) and ∆H°act(1) are available by the protocol described
in (1), ∆S°act(2) and ∆H°act(2) are the remaining unknowns.
Plotting kobsd vs T with a nonlinear least-squares fit to eq 13
(Figure 9) affords ∆H°act(2) ) 11.2 ( 0.2 and ∆S°act(2) ) -36
( 1.
(31) The endo-exo selectivities were determined by GC-MS via com-
parison with authentic samples of norbornene and norbornene-d1. For related
studies see ref 8.
Syn-Exo Selectivity. The endo-exo selectivity of the base-
mediated (()-2-exo-bromonorbornane eliminations has been

Chelation-Based Stabilization of Transition Structure
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5579
Scheme 2
parallel pathways involving mono- and disolvated transition
structures such as 9 and 10, respectively. The ∆H°act values
for eliminations associated with 9 are indistinguishable for
n-BuOMe and THF. We infer that any additional stabilization
offered by THF compared to n-BuOMe35 in the ground state
and transition state cancel. In contrast, ∆H°act for reaction via
10 where ROR corresponds to THF is ≈2.0 kcal/mol lower than
∆H°act for reaction via 9. It seems logical that the superior
ligating properties of THF compared to n-BuOMe35 would
stabilize such a highly solvated transition structure. The low
stabilization when compared with estimated lithium-THF bond
enthalpies27 is consistent with severe steric interactions. The
surprising insensitivity of ∆S°act to the number of ligands in
the transition structures is more difficult to explain, but may
make sense in the context of open and closed transition
structures (Vide infra).
We do not intend to engage in a discussion of the many
classifications of eliminations.36,37 Nevertheless, some com-
ments about the events occurring at the substrate are warranted.
We depict the rate limiting transition structures 9 and 10 as
syn, exo, and cyclic. Several observations confirm the syn and
exo deprotonation, but do not necessarily support cyclic transi-
tion structures: (1) Eliminations of endo-1-d1 and exo-1-d1 occur
with >90% syn-exo proton abstraction. (2) (()-2-exo-iodo-
norbornane13 eliminates at nearly the same rate as the bromide
1 (kobsd(I)/kobsd(Br) ) 2.1 in THF at 20 °C). Orbital alignment
and steric factors may be more important than the Li-Br
interaction. The low coordination number of putative transition
structure 9 seems to demand the existence of a distinct Br--Li
interaction, although this may be dogmatic logic. Nevertheless,
a change from a cyclic transition structure such as 9 to an open
transition structure such as 11 could account for the marginally
more negative ∆S°act associated with a pathway involving an
additional ligand. A related issue arises for chelating ligands
as described below.
Discussion
In the most general sense, we are interested in understanding
the factors that influence lithium amide reactivity. We were
drawn to dehydrobrominations by a suggestion of Schlosser and
co-workers that LDA-mediated dehydrofluorinations proceed
by transition structures based upon LDA open dimers.7d Lithium
amides and other organolithium open dimers have become a
topic of significant attention.32-34 While an open dimer-based
mechanism was not detected in this particular study, the
eliminations of (()-2-exo-bromonorbornane (eq 1) provide
highly solvent-dependent rates and a mechanistic probe uncom-
plicated by side reactions. LDA offers substantial advantages
over other commonly employed lithium amides and other
organolithium reagents in that LDA exists exclusively as a
disolvated dimer for a large number of donor solvent structures
and concentrations.11 Our efforts to understand the mechanistic
origins of the solvent effects are summarized below.
Mechanism of Dehydrobrominations: The Role of Mono-
dentate Ethereal Ligands. The elimination of 1 by LDA/n-
BuOMe (2) appears to proceed by a single pathway involving
a transition structure such as 9 bearing a monosolvated LDA
monomer fragment (Scheme 2). The corresponding elimination
in THF is more complex. A first-order THF dependence with
an appreciable non-zero y-intercept (Figure 3), [THF]-dependent
isotope effects, and a lack of co-solvent dependence implicate
(32) Romesberg, F. E.; Collum, D. B. J. Am. Chem. Soc. 1992, 114,
2112. Williard, P. G. Unpublished.
(33) Romesberg, F. E.; Collum, D. B. J. Am. Chem. Soc. 1994, 116,
2166.
Transition Structure Solvation: Resolving the “Universal
Ground State” Problem. The possible importance of transition
structures containing disolvated LDA monomer fragments led
us to suppose that chelating ligands might offer substantial
transition structure stabilization. The typical protocol to test
such a hypothesis is to monitor the reaction rates as a function
of the chelating ligand. Barring some unlikely coincidences,
the reaction rates will vary. However, serious problems arise
in the interpretation. The most common practice is to ascribe
the rate changes to variable transition structure solvation, paying
no attention to solvent-dependent changes in the reactant
structures and energies. Inspection of the relative elimination
rates in n-BuOMe, THF, and 2-MeTHF (in parentheses in Chart
(34) Hamatani, T.; Matsubara, S.; Schlosser, M. Tetrahedron 1988, 44,
2865. Williard, P. G.; Liu, Q.-Y. J. Am. Chem. Soc. 1993, 115, 3380.
Henderson, K. W.; Dorigo, A. E.; Liu, Q-Y.; Williard, P. G.; Schleyer, P.
v. R.; Bernstein, P. R. J. Am. Chem. Soc. 1996, 118, 1339. Bernardi, A.;
Capelli, A. M.; Cassinari, A.; Comotti, A.; Gennari, C.; Scolastico, C. J.
Org. Chem. 1992, 57, 7029. Bernardi, F.; Bongini, A.; Cainelli, G.; Robb,
M.; Valli, G. S. J. Org. Chem. 1993, 58, 750. Petasis, N. A.; Teets, K. A.
J. Am. Chem. Soc. 1992, 114, 10328. Gallagher, D. J.; Beak, P. J. Org.
Chem. 1995, 60, 7092. Romesberg, F. E.; Collum, D. B. J. Am. Chem.
Soc. 1992, 114, 2112. Romesberg, F. E.; Collum, D. B. J. Am. Chem. Soc.
1994, 116, 2166. Romesberg, F. E.; Gilchrist, J. H.; Harrison, A. T.; Fuller,
D. J.; Collum, D. B. J. Am. Chem. Soc. 1991, 113, 5751. Romesberg, F.
E.; Collum, D. B. J. Am. Chem. Soc. 1994, 116, 9187. Bernstein, M. P.;
Collum, D. B. J. Am. Chem. Soc. 1993, 115, 789. Bernstein, M. P.; Collum,
D. B. J. Am. Chem. Soc. 1993, 115, 8008. Bernstein, M. P. Ph.D.
Dissertation, Cornell University, Ithaca, NY, 1993. Koch, R.; Wiedel,
Anders, E. J. Org. Chem. 1996, 61, 2523. Pratt, L. M.; Khan, I. M.
Tetrahedron: Asymmetry 1995, 6, 2165. Pratt, L. M.; Hogen-Esch, T. E.;
Khan, I. M. Tetrahedron 1995, 51, 5955. Harder, S.; van Lenthe, J. H.;
van Eikema Hommes, N. J. R.; Schleyer, P. v. R. J. Am. Chem. Soc. 1994,
116, 2508. Ashby, E. C.; Noding, S. A. J. Org. Chem. 1979, 44, 4371.
Shimano, M.; Meyers, A. I. J. Org. Chem. 1996, 61, 5714. Alvaro, G.;
Savoia, D.; Valentini, M. R. Tetrahedron 1996, 38, 12571. Williard, P. G.
Unpublished.
(35) Binding affinities to LiHMDS dimers follow the order THF >
MeTHF > n-BuOMe.27
(36) March, J. AdVanced Organic Chemistry, 4th ed.; Wiley: New York,
1980; Chapter 17.
(37) For leading references to theoretical investigations of dehydroha-
logenations, see: Glad, S. S.; Jensen, F. J. Org. Chem. 1997, 62, 253.
Gronert, S. J. Org. Chem. 1994, 59, 7046. Dewar, M. J. S.; Yuan, Y.-C. J.
Am. Chem. Soc. 1990, 112, 2088.

5580 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Remenar and Collum
Scheme 3
Of course, a more direct approach to avoid the “universal
ground state” assumption is to determine the relative stabilities
of the reactants. We described a method by which the kinetics
of elimination provide the thermodynamics of solvation. Be-
cause of the different inherent reactivities of the dimeric
reactants i-iii (Scheme 1) coordinated by the solvents SR and
Sâ, a plot of kobsd vs solvent mole fraction (X) affords a sigmoidal
function (Figure 6). The two y-intercepts provide the rate
constants for metalations by i and iii with the corresponding
activation energies ∆G°(A) and ∆G°(B) (Figure 4). A nonlinear
least-squares fit to eq 9 or eq 10 affords the relative solvent
binding constants and the relative stabilities of the ground states
based upon reactants i and iii (∆G°(C) in Figure 4). Overall,
the single plot in Figure 6 provides the relative ground state
stabilities and transition state stabilities. Results from selected
cases confirmed that the hemilabile ligands bearing methyl ether
moieties bind equivalently to the LDA dimer. Consequently,
this confirms the assumption that the dramatic and variable rate
increases observed for DME and a variety of bidentate ligands
(Chart 1) stem entirely from differential transition structure
solvation.
1) suggests the lack of such a simple correlation.38 By ignoring
the solvent effects on the reactants, one implicitly assumes that
all ground states are of the same energy. We facetiously call
this the “universal ground state”. It is difficult, yet essential,
to dissect relatiVe reaction rates into the rate retarding influence
of ground state stabilization and the rate accelerating influence
of transition state stabilization. Failure to consider both is
logically flawed and will lead to inValid conclusions. More
detailed discussions of this problem have been published.3,4
We avoided the “universal ground state” assumption in two
ways. The simple solution was to choose potentially chelating
ligands in which the binding affinities to the LDA dimer can
be assumed to be equal. The non-chelated (η1) DME ligands
on LDA dimer 4 should have the same binding constant as
n-BuOMe. In other words, the substitution of DME for
n-BuOMe should be thermoneutral. If so, the 50-fold rate
acceleration for LDA/DME compared to LDA/n-BuOMe can
be ascribed to the stabilization of a chelated transition structure
such as 12 (Scheme 3). We extended this study to include a
variety of methyl ethers bearing potentially chelating pendant
groupssso-called “hemilabile ligands” (Chart 1). We have
previously shown that even the least hindered trialkylamines
are poor ligands for lithium amide dimers.2a,11,20-23,39-42 This
ensures that only the methoxy groups are coordinated in the
ground state and that ligand substitution is thermoneutral.
Determination of selected rate laws (Table 1) secured the
mechanistic homology with LDA/DME (Scheme 3). Thus, we
can attribute the marked ligand-dependent rate variationssup
to 103-fold in optimal casessto differential transition structure
stabilization. It is both interesting and important to note that,
had we used potentially chelating diamines, we would have
seriously undermined mechanistic interpretations by forfeiting
our understanding of relative reactant stabilities.
Transition Structure Solvation: The Chelate Effect. The
rate accelerations imparted by DME and related amino ethers
(Chart 1) do not derive from a general promotion of high
solvation numbers, but from selectiVe stabilization of the
transition structure relatiVe to the LDA dimer reactant. If
chelation stabilizes the reactants and the transition structures,
the sign and magnitude of a DME-induced rate change is not
readily predicted. In contrast, the view of differential transition
state solvation unobscured by differential ground state solvation
effects offers insights into the chelate effect and allows
interesting comparisons with quantitative studies of chelate
formation on LiHMDS monomers. These are as follows:
(1) Chelate ring size is critical. Hemilabile diethers and
amino ethers capable of forming five-membered chelates (12,
n ) 1) offer substantial (up to 103-fold) rate advantages when
compared to n-BuOMe. The corresponding six-membered-ring
chelates (12, n ) 2) are much less consequential. This is
consistent with their low binding affinities to monomeric
LiHMDS.23
(2) Transition structure stabilization correlates inversely with
an increasing bulk of the pendant ligand. For example, the
elimination rates for LDA solvated by MeOCH2CH2NR2 follow
the order NR2 ) NMe2 > NEt2 > N(i-Pr)2. This is consistent
with LiHMDS solvation studies. However, we observe some
unexpected substituent effects. We predicted that the transition
structure stabilization would correlate with the ground state
stabilization of chelated lithium amide monomers; the best
comparison currently available is with chelated LiHMDS
monomers.23 The relative transition structure stabilizations
follow the order RN(CH2)5 (J) < RN(CH2)4 (K) < RN(CH2)3
(L) < RNMe2 (M). Binding affinities of related ethylenedi-
amines (R2NCH2CH2NR2) to LiHMDS monomers follow the
order RNMe2 ≈ RN(CH2)5 < RN(CH2)4 < RN(CH2)3.23 This
dramatic change in the relative ligating strength of the NMe2
moiety may stem from congestion in the transition structure.
This can be seen by comparing ligating strength with ligand
substitution rates on LiHMDS. While the dipyrrolidinoethane
ligand in LiHMDS monomer 13 is more strongly bound than
the TMEDA in 14, the rates of associative ligand substitutions
via monomer 15 are much slower than the corresponding
substitutions via monomer 16; the pyrrolidine groups appear to
be relatively intolerant of the additional steric congestion
occurring upon ligand association. The dehydrobrominations
of 1 by LDA are also susceptible to steric congestion associated
(38) There are a number of reports where ostensibly weaker solvent-
lithium interactions lead to increased overall reaction rates: Apparu, M.;
Barrelle, M. Tetrahedron 1978, 34, 1541. Loupy, A.; Seyden-Penne, J.
Tetrahedron 1980, 36, 1937. Reich, H. J.; Green, D. P.; Phillips, N. H. J.
Am. Chem. Soc. 1989, 111, 3444. Loupy, A.; Seyden-Penne, J.; Tchoubar,
B. Tetrahedron Lett. 1976, 1677. Bywater, S.; Worsfold, D. J. Can. J. Chem.
1962, 40, 1564. Ku¨ndig, E. P.; Desobry, V.; Simmons, D. P.; Wenger, E.
J. Am. Chem. Soc. 1989, 111, 1804. Reich, H. J.; Phillips, N. H.; Reich, I.
L. J. Am. Chem. Soc. 1985, 107, 4101. Kim, Y. H.; Choi, J. Y. Tetrahedron
Lett. 1996, 37, 5543. Reich, H. J.; Dykstra, R. R. Angew. Chem., Int. Ed.
Engl. 1993, 32, 1469. See refs 2a, 3, and 18.
(39) For early studies revealing poor coordination of trialkylamines to
organolithium aggregates, see: Settle, F. A.; Haggerty, M.; Eastham, J. F.
J. Am. Chem. Soc. 1964, 86, 2076. Lewis, H. L.; Brown, T. L. J. Am. Chem.
Soc. 1970, 92, 4664. Brown, T. L.; Gerteis, R. L.; Rafus, D. A.; Ladd, J.
A. J. Am. Chem. Soc. 1964, 86, 2135. Quirk, R. P.; Kester, D. E. J.
Organomet. Chem. 1977, 127, 111.
(40) For a discussion of steric effects of amines in the context of transition
metal ligation see: Seligson, A. L.; Trogler, W. C. J. Am. Chem. Soc. 1991,
113, 2520. Choi, M.-G.; Brown, T. L. Inorg. Chem. 1993, 32, 1548.
Widenhoefer, R. A.; Buchwald, S. L. Organometallics 1996, 15, 3534.
(41) Lucht, B. L.; Collum, D. B. J. Am. Chem. Soc. 1996, 118, 2217.
(42) Bernstein, M. P.; Romesberg, F. E.; Fuller, D. J.; Harrison, A. T.;
Collum, D. B.; Liu, Q.-Y.; Williard, P. G. J. Am. Chem. Soc. 1992, 114,
5100.

Chelation-Based Stabilization of Transition Structure
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5581
with the substrate-amide interactions. These additional steric
effects would not be expected to correlate with chelate stability.
renders a detailed consideration of ∆H°act even less appropriate.
Lastly, we must confess that all of these inferences are further
weakened by the lack of relative ground state enthalpies and
entropies; the “universal ground state” emerges again.
(5) Additional ligating groups only offer rate advantages if
the mechanism includes provisions for the additional coordina-
tion. The equivalent rates for LDA/diglyme and LDA/DME
attest to the lack of η3 coordination of the diglyme at the
transition structure. Diglyme binds as a tridentate ligand on
the LiHMDS monomer, but is not unusually stabilizing com-
pared with bidentate diamines.23 Along a similar vein, LDA-
mediated dehydrobrominations and LDA-mediated metalations
of N,N-dimethylhydrazones both proceed via rate limiting
transition structures bearing monosolvated LDA monomer
fragments. However, the hydrazones are not metalated via more
highly solvated transition structures,2a,18,42 possibly due to a high
steric demand33 of the hydrazones. Since the potentially
polydentate ligands do not selectively stabilize the hydrazone
metalation transition structures,13 such ligands offer no rate
advantages.2a,42
(3) The approximate 20-fold rate enhancement offered by
Me2NCH2CH2OMe compared to DME implicates a higher
azaphilicity rather than oxaphilicity at the transition structure.
While this seems surprising considering the assignment of amino
ether solvated LDA as η1-oxygen-bound dimers (e.g., 5-7), the
high azaphilicity is fully compatible with LiHMDS solvation
studies. Lithium amide dimers are far too hindered to comfort-
ably support trialkylamine ligands,2a,11,20-23,39-42 whereas mono-
mers offer a less congested environment.23 In addition, we
found that dialkyl ethers and dialkylamines show indistinguish-
able affinities for binding to the LiHMDS dimers,41 while the
dialkylamines show a greater tendency than the ethers to bind
to the monomers. We concluded that the LiHMDS monomers
are more “azaphilic” than the dimers, possibly equating with a
stronger Lewis acidity. Interestingly, a similar preference for
amine- vs ether-based chelation noted by Reich was ac-
companied by a higher propensity to simultaneously bind an
equivalent of HMPA.43 An enhanced azaphilicity of organo-
lithium monomers and monomer-based transition structures,
when compared to higher oligomers, may account for a myriad
of di- and polyamine solvation effects on lithium salt structures
and reactivities.4,44
(4) Although we are skeptical of activation parameters due
to their susceptibility to systematic and random error, we
attempted to dissect ∆G°act into ∆S°act and ∆H°act to gain a more
detailed view of chelation in the transition structure (Table 1).
The rate increases observed for LDA solvated by DME and
related amino ethers are clearly ascribable to the pendant ligands
and reveal factors that affect the chelate stabilities. However,
problems arise when we attempt to dissect the chelate effect
into separate contributions. For example, the ≈2.0 kcal/mol
reduction in ∆H°act for LDA/DME compared to LDA/n-BuOMe
stemming from the additional metal-ligand interaction in a
chelated transition structure such as 12 is attenuated by the
torsional enthalpy loss upon forming the five-membered ring.
Similarly, enthalpic gains offered by the second solvent-lithium
interaction in di(n-BuOMe)-solvated transition structure 10 will
be offset by destabilizing solvent-solvent and related steric
interactions. It is not obvious that attempting to make such
energetic distinctions is a productive exercise. The situation
becomes acutely complex when entropy is considered. Com-
parisons with LDA/n-BuOMe reveal that chelation at the
transition structure occurs with very little entropic cost despite
the anticipated restricted degrees of freedom associated with
ring formation. We suspect that open transition structures such
as 17 may be intervening, offsetting entropic losses associated
with chelate formation. While this model might neatly account
for the insensitivity of the ∆S°act to chelate formation, it also
Conclusions
Solvent effects on reaction rates can only be understood with
a knowledge of the reaction mechanism(s) as well as the
influence of solvation on both the reactant and transition
structure stabilities. While this statement seems self-evident,
the tendency to ignore solvent-dependent ground state effectssthe
so called “universal ground state” assumptionsis so widespread
as to defy explanation. We are unaware of any fundamental
principle dictating that transition states are more susceptible to
solvation than ground states. For example, the solution phase
SN2 reaction is retarded relative to its gas-phase counterpart by
disproportionate ground state solvation.45 The work described
herein is not free of simplifying assumptions. However, the
protocol employed for separating ground state and transition
state solvation to understand the chelate effect has promise, and
such distinctions are essential if we are to understand the
relationship between solvation, aggregation, and reactivity within
organolithium chemistry.
Experimental Section
Reagents and Solvents. n-BuOMe, THF, 2-MeTHF, DME, and
diglyme were obtained from Aldrich. Vicinal amino ethers G, H, I,
J, and K (see Chart 1) were prepared by N-alkylation as described in
the previous paper.46 F47 , L,48 and M49 were prepared by (modified)
standard procedures. All solvents were distilled by vacuum transfer
(45) Craig, S. L.; Brauman, J. I. J. Am. Chem. Soc. 1996, 118, 6786.
(46) Gibson, M. S. In The Chemistry of the Amino Group; Patai, S.,
Ed.; Wiley: New York, 1968; p 440. Bitsi, G.; Schleiffer, E.; Antoni, F.;
Jenner, G. J. Organomet. Chem. 1989, 373, 343. Spialter, L.; Pappalardo,
J. A. The Acyclic Aliphatic Tertiary Amines; McMillan; New York, 1965;
pp 14-29.
(47) Noyes, A. A. Am. Chem. J. 1897, 19, 766.
(48) Sammes, P. G.; Smith, S. J. Chem. Soc., Perkin Trans. 1 1984,
2415.
(43) Reich, H. J.; Kulicke, K. J. J. Am. Chem. Soc. 1996, 118, 273. For
additional studies of O vs N chelation, see ref 3.
(44) Polyamine-Chelated Alkali Metal Compounds; Langer, A. W., Jr.,
Ed.; American Chemical Society: Washington, DC, 1974.
(49) Pine, S. H.; Sanchez, B. L. J. Org. Chem. 1971, 36, 829 and
references cited therein.
(50) McEwen, W. E.; Janes, A. B.; Knapczyk, J. W.; Kyllingstad, V.
L.; Siau, W.-I.; Shore, S.; Smith, J. H. J. Am. Chem. Soc. 1978, 100, 7304.

5582 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Remenar and Collum
from sodium benzophenone ketyl. The hydrocarbon stills contained
1% tetraglyme to dissolve the ketyl. (()-2-exo-Bromonorbornane and
the deuterated derivatives were prepared according to literature
methods.13 The LDA was prepared as a solid with commercial n-BuLi
(Aldrich) and purified by the standard literature procedure.14 The
diphenylacetic acid used to check solution titers51 was recrystallized
from methanol and sublimed at 120 °C under full vacuum. Air- and
moisture-sensitive materials were manipulated under argon or nitrogen
following standard glovebox, vacuum line, and syringe techniques.
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-toluene 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 (5-10 per rate
constant) fitted with stir bars were charged with the LDA stock solution
and brought to the desired temperature ((0.2 °C) with use of a
constant-temperature bath fitted with a National Bureau of Standards
thermometer. The (()-2-exo-bromonorbornane was added as a 0.047
M stock solution in the appropriate ligand-toluene mixture containing
undecane (0.047 M) as a GC standard. The vessels were periodically
quenched with 1:1 H2O-THF at intervals chosen to ensure an adequate
sampling of each of the first three half-lives. The quenched aliquots
were extracted into Et2O, and the extracts were analyzed via capillary
GC. Use of a Hewlett Packard GC fitted with an autoinjector and a
60-m DB-5 column (J & W Scientific) provided excellent reproduc-
ibility. The eliminations were monitored by following the decrease of
the (()-2-exo-bromonorbornane (1) relative to the internal undecane
standard. Following the formation of the norbornene afforded equiva-
lent rate constants ((10%). Rate constants were determined by
numerical integration with a convergence protocol using the Scientist
distributed by MicroMath. Nonlinear least-squares fits using the integral
form of the rate laws afforded numerically indistinguishable results.
The reported errors correspond to one standard deviation. The observed
rate constants were shown to be reproducible within (5%.
Acknowledgment. We wish to express thanks to Dmitriy
Kruglyak for developing the synthesis of ligand L. We also
thank the National Institutes of Health for direct support of this
work and acknowledge the National Science Foundation In-
strumentation Program (CHE 7904825 and PCM 8018643), the
National Institutes of Health (RR02002), and IBM for support
of the Cornell Nuclear Magnetic Resonance Facility.
Supporting Information Available: Plots and tables of rate
data (19 pages). See any current masthead page for ordering
and Internet access instructions.
(51) Kofron, W. G.; Baclawski, L. M. J. Org. Chem. 1976, 41, 1879.
JA970030A