'New' catalysts for the ester-interchange reaction: The role of alkali-metal alkoxide clusters in achieving unprecedented reaction rates

Article abstract of DOI:10.1021/ja980584t

The catalytic effect of alkali-metal tert-butoxide clusters on the rate of ester interchange for several pairs of esters has been determined in nonpolar and weakly polar solvents. Reactivities increase in the order (Li⁺ < Na⁺ < K⁺ < Rb⁺ < Cs⁺) with the fastest rates reaching 10⁷ catalytic turnovers per hour (TO/h). Ester interchange rates were sensitive to the size of both the transferring OR groups and the ester substituent. Phenyl esters did not exchange with aliphatic esters due to nonstatistical breakdown patterns in the tetrahedral intermediate. A first-order equilibration analysis on the interchange between tert-butyl acetate (tBuAc) and methyl benzoate (MeBz) (5 mol % NaOtBu) indicated enhanced reaction rates as the reaction proceeded. Isolation and quenching (DCl/D₂O) of precipitated catalyst points to a mechanism whereby sequential methoxy incorporation into the catalyst cluster increases activity, but eventually precipitates out of solution as a 3:1 OMe:OtBu cluster. The rate law was determined to be k(obs)[MeBz]¹[tBuAc]⁰[NaOtBu](x), where x = 1.2(1), 1.4(1), and 0.85(1) in hexane, ether, and THF, respectively, under conditions where tetrameric catalyst aggregates are expected. Reaction rates were generally observed to be higher in nonpolar solvents (hexane > toluene, ether > THF). Eyring analysis over a 40°C range yielded ΔH(≠) = 10.0(1) kcal mol^-1 and ΔS(≠) = -32(3) eu. A Hammett (σ) plot generated with para-substituted methyl benzoates gave ρ +2.35 (R 0.996). These results are interpreted in terms of a catalytic cycle composed of two coupled transesterification reactions with a turnover-limiting addition of a tert-butoxy-containing cluster (tetramer) to methyl benzoate. Catalyst relative reactivities (Cs⁺ > Rb⁺ > K⁺ > Na⁺ > Li⁺) are interpreted in terms of competitive electrostatic interactions between the alkali-metal and ground-state and transition-state anions. This analysis predicts the observed linear dependence between log(k(obs)) and l/r(ionic).

Full text of DOI:10.1021/ja980584t

J. Am. Chem. Soc. 1998, 120, 5981-5989
5981
“New” Catalysts for the Ester-Interchange Reaction: The Role of
Alkali-Metal Alkoxide Clusters in Achieving Unprecedented Reaction
Rates
Matthew G. Stanton, Cara B. Allen, Rebecca M. Kissling, Alice L. Lincoln, and
Michel R. Gagne´*
Contribution from the Department of Chemistry CB#3290, UniVersity of North Carolina,
Chapel Hill, North Carolina 27599-3290
ReceiVed February 20, 1998
Abstract: The catalytic effect of alkali-metal tert-butoxide clusters on the rate of ester interchange for several
pairs of esters has been determined in nonpolar and weakly polar solvents. Reactivities increase in the order
(Li+ < Na+ < K+ < Rb+ < Cs+) with the fastest rates reaching 107 catalytic turnovers per hour (TO/h).
Ester interchange rates were sensitive to the size of both the transferring OR groups and the ester substituent.
Phenyl esters did not exchange with aliphatic esters due to nonstatistical breakdown patterns in the tetrahedral
intermediate. A first-order equilibration analysis on the interchange between tert-butyl acetate (tBuAc) and
methyl benzoate (MeBz) (5 mol % NaOtBu) indicated enhanced reaction rates as the reaction proceeded.
Isolation and quenching (DCl/D2O) of precipitated catalyst points to a mechanism whereby sequential methoxy
incorporation into the catalyst cluster increases activity, but eventually precipitates out of solution as a 3:1
OMe:OtBu cluster. The rate law was determined to be kobs[MeBz]1[tBuAc]0[NaOtBu]x, where x ) 1.2(1),
1.4(1), and 0.85(1) in hexane, ether, and THF, respectively, under conditions where tetrameric catalyst aggregates
are expected. Reaction rates were generally observed to be higher in nonpolar solvents (hexane > toluene,
ether > THF). Eyring analysis over a 40 °C range yielded ∆H‡ ) 10.0(1) kcal mol-1 and ∆S‡ ) -32(3) eu.
A Hammett (σ) plot generated with para-substituted methyl benzoates gave F ) +2.35 (R ) 0.996). These
results are interpreted in terms of a catalytic cycle composed of two coupled transesterification reactions with
a turnover-limiting addition of a tert-butoxy-containing cluster (tetramer) to methyl benzoate. Catalyst relative
reactivities (Cs+ > Rb+ > K+ > Na+ > Li+) are interpreted in terms of competitive electrostatic interactions
between the alkali-metal and ground-state and transition-state anions. This analysis predicts the observed
linear dependence between log(kobs) and 1/rionic
.
Introduction
ester interchange reaction (eq 1).3a,4 Reactivities were high
enough to realize a convenient method for the synthesis of tert-
butyl esters.3b At that time we speculated that the mechanism
involved a coupled transesterification mediated by a soluble
alkali-metal alkoxide (Scheme 1). Preliminary experiments
suggested that the catalyst was aggregated and that high activity
was intimately associated with the presence of clusters.
While the correlation between solution structure, aggregation
state, and reactivity has been extensively studied for lithium
alkyls,5 enolates,6 phenoxides,7 and amides,8 much less is known
The transesterification reaction is one of the most important
ways of manipulating the ester functionality as exemplified by
the many protocols that have been developed for carrying out
this reaction. These include basic,1 acidic,1a and metal-catalyzed
variants,2 with most methods being protic in nature.1a A process
capable of catalyzing the transesterification reaction with high
rates and under mild conditions is broadly applicable to fine
chemical and polymer synthesis.
(3) (a) Stanton, M. G.; Gagne´, M. R. J. Am. Chem. Soc. 1997, 119, 5075-
5076. (b) Stanton, M. G.; Gagne´, M. R. J. Org. Chem. 1997, 62, 8240-
8242.
(4) The first reported catalyst for this process was La(OiPr)3, see: Okano,
T.; Hayashizaki, Y.; Kiji, J. Bull. Chem. Soc. Jpn. 1993, 66, 1863-1865.
(5) (a) Power, P. P. Acc. Chem. Res. 1988, 21, 147-153. (b) McGarrity,
J. F.; Ogle, C. A.; Brich, Z.; Loosli, H.-R. J. Am. Chem. Soc. 1985, 107,
1810-1815. (c) McGarrity, J. F.; Ogle, C. A. J. Am. Chem. Soc. 1985,
107, 1805-1810. (d) Bates, R. B.; Ogle, C. A. Carbanion Chemistry;
Springer-Verlag: New York, 1983.
(6) (a) Jackman, L. M.; Bortiatynski, J. AdV. Carbanion Chem. 1992, 1,
45-87. (b) Seebach, S. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624-
1654. (c) Jackman, L. M.; Lange, B. C. Tetrahedron 1977, 33, 2737-
2769.
(7) See ref 6a and the following: (a) Jackman, L. M.; Cizmeciyan, D.;
Williard, P. G.; Nichols, M. A. J. Am. Chem. Soc. 1993, 115, 6262-6267.
(b) Jackman, L. M.; Smith, B. D. J. Am. Chem. Soc. 1988, 110, 3829-
3935. (c) Jackman, L. M.; DeBrosse, C. W. J. Am. Chem. Soc. 1983, 105,
4177-4184.
We recently reported that alkali-metal alkoxide clusters serve
as highly active catalysts (up to 106 turnovers (TO)/ h) for the
(1) (a) For a recent review, see: Otera, J. Chem. ReV. 1993, 93, 1449-
1470. (b) Rowan, S. J.; Sanders, J. K. M. Chem. Commun. 1997, 1407-
1408. (c) Okano, T.; Miyamoto, K.; Kiji, J. Chem. Lett. 1995, 246. (d)
Cacciapaglia, R.; Mandolini, L.; Castalli, V. V. A. J. Org. Chem. 1997,
62, 3089-3092 and references therein. (e) Zhang, B.; Breslow, R. J. Am.
Chem. Soc. 1997, 119, 1676-1681 and references therein. (f) Chin, J. Acc.
Chem. Res. 1991, 24, 145-152 and references therein.
(2) References 1a, 1f, and the following: (a) Parshall, G. W.; Ittel, S.
D. Homogeneous Catalysis, 2nd ed.; John Wiley & Sons: New York, 1992;
pp 269-280. (b) Mascaretti, O. A.; Furla´n, R. L. E. Aldrichim. Acta 1997,
30, 55-68.
S0002-7863(98)00584-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/05/1998

5982 J. Am. Chem. Soc., Vol. 120, No. 24, 1998
Stanton et al.
Scheme 1
tetrameric structures in solution (K) and in the solid state (A).16
Details relating aggregate structure to reactivity have been
provided by Jackman, who has documented conditions where
lithium 3,5-dimethylphenolate exists in solution as dimers and
tetramers (pyridine), tetramers (THF), and tetramers and hex-
amers (dioxolane).17 Careful kinetic studies for the degenerate
transesterification reaction in eq 2 revealed that the clusters were
about this correlation in heavier alkali-metal alkoxides.9,10 Since
heavy alkali-metal alkoxides are superior ester-interchange
catalysts (Cs+ > Rb+ > K+ > Na+), we have initiated
experiments to determine the fundamental relationship between
structure and reactivity in these compounds. These data will
ultimately lead to more reactive, selective, and stable catalysts.
The solid-state, solution, and gas-phase structures of several
soluble alkali-metal alkoxides have been reported. In the solid-
state, lithium neophyloxide (LiOCMe2Ph) adopts a hexameric
structure of type B, which persists in nonpolar solvents (up to
80 °C)11 but is broken down to a tetra-solvated tetramer in donor
solvents such as THF.12 Mass spectrometry experiments also
establish that hexamers are the dominant form in the gas phase.13
Sodium alkoxides are also highly aggregated as the solid-state
structure of NaOtBu contains a one-to-one ratio of a B-type
hexamer and a C-type nonomer.14 Cryoscopy in both benzene
and cyclohexane report an average solution aggregation number
of 8.3 over a wide concentration range (0.06-0.3 M), consistent
with a mixture of hexamers and nonomers.11 FT-IR experiments
confirmed these observations and noted that, in strong donor
solvents such as THF, high-symmetry tetrameric clusters
dominate, A.15 The heavier alkali-metal alkoxides KOtBu,
RbOtBu, and CsOtBu are tetramers in the solid state16 and at
least for potassium are known to be tetramers in toluene, ether,
and THF solutions.15 In general, the lighter, smaller, alkali-
metal tert-alkoxides (Li and Na) form higher aggregates (6-9)
that in the presence of good donor solvents reorganize to
tetramers, while heavier alkali metals (K, Rb, and Cs) prefer
the primary reactants and that each aggregate had a unique
reactivity (tetramer shown). For example, in pyridine, the dimer
is 13 times more reactive than the tetramer (Keq ) 330 M-1 at
35 °C), and in dioxolane, the hexamer is 5.5 times more reactive
than the tetramer (Keq ) 0.57 M-1/3 at 35 °C).
Despite the high-level understanding of the Jackman system,
analogous data on the reactivity of heavy alkali-metal alkoxides
and aryl oxides with aliphatic and aromatic esters is not
available. This paper reports our studies addressing the
underlying mechanistic features of the alkali-metal alkoxide-
cluster-catalyzed ester interchange reaction. The primary focus
is on the kinetics of the equilibration of a mixture of esters as
a function of the nature of the metal cation, solvent, electronic
and steric perturbations, concentration, and temperature. Future
submissions will address the structural aspects of catalyst
intermediates.
Results
Effect of Ester Structure on Reactivity. Using tert-butyl
acetate (tBuAc) as the test substrate the rate of tert-butoxide
incorporation into various methyl esters was determined using
5 mol % NaOtBu in THF (eq 3). The relative rates were found
(8) (a) Remenar, J. F.; Collum, D. B. J. Am. Chem. Soc. 1997, 119,
5573-5582 and references therein. (b) Hilmersson, G.; Davidsson, O¨ . J.
Org. Chem. 1995, 60, 7660-7669. (c) Collum, D. B. Acc. Chem. Res. 1993,
26, 227-234. (d) Collum, D. B. Acc. Chem. Res. 1992, 25, 448-454.
(9) Msayib, K. J.; Watt, C. I. F. Chem. Soc. ReV. 1992, 237-243.
(10) For a series of MOPh (M ) Li, Na, K, Rb, Cs) high-resolution
powder structures, see: Dinnebier, R. E.; Pink, M. Sieler, J.; Stephens, P.
W. Inorg. Chem. 1997, 36, 3398-3401.
(11) For a compilation of aggregation states for lithium, sodium, and
potassium tert-butoxides in various solvents, see: Halaska, W.; Lochmann,
L.; Lim, D. Collect. Czech. Chem. Commun. 1968, 33, 3245-3253.
(12) Chisholm, M. H.; Drake, S. R.; Naiini, A. A.; Streib, W. E.
Polyhedron 1991, 10, 805-810.
to generally parallel the steric bulk of the R group as expressed
by their respective A values (Table 1).18 The exception to this
trend was methyl benzoate (MeBz), the enhanced rate likely
reflecting an electronic effect.
The effect on rate of the transferring alkoxide group was also
examined (eq 4, Table 2). In the case of MeBz and ethyl
(13) Kahn, J. D.; Haag, A.; Schleyer, P. v. R. J. Phys. Chem. 1988, 92,
212-220 (b) Hartwell, G. E.; Brown, T. L. Inorg. Chem. 1966, 5, 1257-
1259.
(14) Davies, J. E.; Kopf, J.; Weiss, E. Acta Crystallogr. 1982, B38, 2251-
2253.
(15) Schmidt, P.; Lochmann, L.; Schneider, B. J. Mol. Struct. 1971, 9,
403-411.
(16) (a) Chisholm, M. H.; Drake, S. R.; Naiini, A. A.; Streib, W. E.
Polyhedron 1991, 10, 337-345. (b) Weiss, E.; Alsdorf, H.; Ku¨hr, H.;
Gru¨tzmacher, H.-F. Chem. Ber. 1968, 101, 3777-3786.
(17) Jackman, L. M.; Rakiewicz, E. F.; Benesi, A. J. J. Am. Chem. Soc.
1991, 113, 3451-3458.
(18) Carey, F. A.; Sundberg, R. J. AdVanced Organic Chemistry, 3rd
ed.; Plenum Press: New York, 1990; p 135.

“New” Catalysts for the Ester-Interchange Reaction
J. Am. Chem. Soc., Vol. 120, No. 24, 1998 5983
Table 1. First-Order Rate Constants for the Equilibration of
Various Methyl Esters (Eq 3)a
a 0.8 M in each ester, 40 mM NaOtBu in THF. See the Experimental
Section for full details. b In units of kcal mol-1 (ref 25).
Table 2. Effect of the Transferring OR Group on the Rate of
Equilibration (Eq 4)a
Figure 1. [tBuBz] vs time showing initial zero-order kinetics followed
by a rapid deviation from linearity (eq 7). The half-lives for conversion
to equilibrium are 8.0, 6.5, 3.8, 3.5, and 3.0 min. The line represents
a linear fit to the first 10 min (∼1.5 half-lives) (0.8 M in each ester, 40
mM (5 mol %) NaOtBu, [tBuBz]eq) 0.426 M).
a 0.8 M in each ester, 40 mM NaOtBu in THF. See the Experimental
Section for full details. b Based on a tetrameric catalyst.
suitable reacting partners to be identified a priori. No attempt
was made to optimize this process.
acetate, 5 mol % of NaOtBu was found to affect the equilibration
in <5 s. By assuming that six half-lives are required to reach
near-equilibrium and that a minimum of 40 turnovers are
necessary to progress one half-live (2 TO per productive event,
tetrameric catalyst), we estimate that 240 productive TO/mol
of catalyst tetramer are required to reach equilibrium. The
observed reactivity for NaOtBu therefore reflects a minimum
reactivity of 173 000 TO/h. The isopropyl acetate was similarly
at equilibrium after a 5-s quench. Not surprisingly, the rates
for establishing equilibrium were sensitive to the size of the R
substituent with R ) Et, iPr . tBu . Ph. Although tert-butyl
pivalate was not readily accessible using NaOtBu, the more
reactive KOtBu catalyst (vide infra) was able to achieve a ∼23%
conversion to this hindered ester over 4 h. We do not yet know
if this value reflects a true equilibrium, as steric effects will
certainly disfavor tert-butyl pivalate.19
Worthy of note is the complete lack of reactivity using phenyl
acetate. We attribute this to a nonstatistical breakdown of the
tetrahedral intermediate; the pKa’s of the conjugate acids (MeOH
) 29, PhOH ) 18)20 ensure that phenoxide ejection dominates,
and no productive turnovers occur (eq 5). The corollary to these
observations is that the pKa of the conjugate acids of the two
competing leaving groups be sufficiently close that the relative
breakdown rates of the tetrahedral intermediate be competitive.
Rate of Approach to Equilibrium of a Test Reaction.
Reactions utilizing tBuAc and MeBz led to convenient rates
that were amenable to aliquot quenching techniques coupled
with GC analysis. Thus, the reaction depicted in eq 7 was
chosen for an in-depth analysis. The conversion of MeBz to
tBuBz with 5 mol % NaOtBu in THF was analyzed to ∼4 half-
lives. A plot of the increasing concentration of tBuBz vs time
is shown in Figure 1 and highlights the gross features of an
initially zero-order reaction that rapidly deviates from linearity
after ∼1.5 half-lives. From the first half-life the initial rate could
be measured. Alternatively, treating the data with a first-order
relaxation approach (i.e., MeBz relaxing to an equilibrium
mixture of MeBz and tBuBz) generates a plot wherein the data
deviates from linearity after 1 half-life and reaches equilibrium
at a rate faster than expected by extrapolating the first half-life.
A simple first-order equilibration scenario is inadequate for
describing the time course of the reaction, but it is useful for
determining initial rate constants.21
The approach to equilibrium of the reaction in eq 7 is
characterized by a gradual clouding of the solution, roughly
coinciding with the deviations from zero-order kinetics (Fig-
ure 1). Isolation of the catalyst which precipitates from eq 7
(using 10 mol % NaOtBu in hexane) followed by quenching
with DCl/D2O establishes that the ratio of methoxide to tert-
butoxide in the precipitate is 3:1 as judged by 1H NMR. Since
(NaOtBu)x (x ) 6, 9) clusters are soluble in hexane (the
aggregation state is unknown in 1.6 M ester) and NaOMe is an
insoluble sheetlike material,22 it follows that clusters containing
mixed ratios of -OtBu and -OMe groups must exist in solution.
It is reasonable to suggest that the observed 3:1 OMe:OtBu ratio
To test this hypothesis, the cross interchange of two active
esters containing leaving groups of similar ability (pKa: PhOH
) 18, pentanethiol ) 17)20 was attempted with 5 mol % KOtBu
(eq 6). This reaction proceeds to ∼30% conversion and then
halts, proving, at least in principle, that pKa matching allows
(19) In cases involving primary and secondary alkoxy substituents,
equilibrium is usually represented by a ∼1:1:1:1 ratio of esters. Conversion
to a single product has been achieved by selectively removing a volatile
product ester, see ref 3b.
(20) Bordwell F. G. Acc. Chem. Res. 1988, 21, 456-463.

5984 J. Am. Chem. Soc., Vol. 120, No. 24, 1998
Stanton et al.
Scheme 2
Table 3. Effect of Alkali-Metal Cation on the Approach to
represents a tetrameric cluster such as F (Scheme 2) that is no
longer soluble in the reaction media. Conversely, ratios of
1:3 (D) or 2:2 (E) must represent viable alkoxide cluster
catalysts that are soluble in the mixture of ester and sol-
vent. The observed rate accelerations suggest that as methyl
groups are substituted for tert-butyl groups via the ester
interchange the decreased steric hindrance in the cluster results
in increased catalytic activity. Precedence for heterosubstituted
alkoxide clusters is found in the work of Jackman, where LiI
and LiClO4 were found to incorporate into an aryl oxide tetramer
to form Li4(OAr)3(X), where X ) I-, ClO4-, and Li4(OAr)2-
(ClO4)2.23,24
Effect of Alkali Metal Cation on Reaction Rates. The
ability of various alkali-metal tert-butoxides to affect the equi-
libration of the reaction in eq 7 was determined and the results
summarized in Table 3. A dramatic trend is observed where
the larger/heavier alkali metals give rise to faster rates (Figure
2),25 the highest rates being near the limit of reproducible
kinetics.
Equilibrium of Eq 7a
a Reaction conditions: 0.8 M in each ester, 40 mM in MOtBu (5
mol %) in THF under argon atmosphere. b See footnote 25. c Ill-be-
haved kineticssreaction never reached equilibrium. d Lower limits
reaction rates were too fast to accurately measure (equilibrium in ∼15-
30 s). e Single measurement. f Turnover frequency through first half-
life.
Assuming that 40 productive turnovers of a tetrameric catalyst
are required to complete a single half-life (vide supra), turnover
frequencies have been calculated for each metal. That Nt for
n-alkyl ester interchange is ∼100 times faster than tert-butyl
esters (Table 2) suggests that initial Nt’s for n-alkyl esters using
the fastest Cs catalysts should approach 1 × 107 h-1
. In
contrast, the lanthanide alkoxide-based ester interchange catalyst
previously reported by Kiji has a maximum reported Nt of ∼20
h-1 (10 °C) for an analogous process.4
Effect of Solvent on the Rate of Equilibration. Again using
eq 7 as a test reaction, the rate of equilibration of MeBz and
tBuBz was determined in various solvents (Table 4). Although
CH2Cl2 and DME are exceptional the general trend is that poor
(21) Such a treatment is in principle justified on the basis of the
observation that the reaction is first order in MeBz and zero order in tBuAc.
This is consistent with Scheme 1, wherein the rate-determining step is tert-
butoxide addition to MeBz and is followed by a fast second step. Since the
concentrations of the two products are necessarily equal, the reverse process
must also have a rate-determining step that is first order in one of the esters
(we do not yet know which).
(22) (a) Weiss, von E.; Alsdorf, H. Z. Anorg. Allg. Chem. 1970, 372,
206-213. (b) Weiss, Von E. Z. Anorg. Allg. Chem. 1964, 332, 197-203.
(23) (a) Jackman, L. M.; Rakiewicz, E. F.; Benesi, A. J. J. Am. Chem.
Soc. 1991, 113, 4101-4109. (b) Jackman, L. M.; Rakiewicz, E. F. J. Am.
Chem. Soc. 1991, 113, 1202-1210.
Figure 2. Observed initial rates (one half-life) for eq 7 vs six-coordinate
ionic radius of the metal counterion25 (0.8 M in each ester, 40 mM in
MOtBu, THF). The line represents a smooth curve through the data
points.
donor solvents give faster rates (hexane > toluene g ether >
THF). This effect has previously been observed by Jackman
et al.17 for the degenerate transesterification reaction in eq 2
under conditions where the tetramer is the dominant species.
This is most reasonably interpreted as a result of competitive
inhibition by donor solvents for the preequilibrium binding site
on the cluster (vide infra).26 Consistent with this proposal are
the relative rates of THF, 2-Me-THF, and 2,5-Me2-THF, which
have similar dielectric constants but progressively poorer donor
(24) For studies relating to Li-enolate/LiCl(Br) aggregate structure and
reactivities, see also: (a) Abu-Hasanayn F.; Streitwieser, A. J. Am. Chem.
Soc. 1996, 118, 8136-8137. (b) Imai, M.; Hagihara, A.; Kawasaki, H.;
Manabe, K.; Koga, K. J. Am. Chem. Soc. 1994, 116, 8829-8830. (c)
Jackman, L. M.; Dunne, T. S. J. Am. Chem. Soc. 1985, 107, 2805-2806.
(d) Jackman, L. M.; Lange, B. C. J. Am. Chem. Soc. 1981, 103, 4494-
4499. (e) Jackman, L. M.; Haddon, R. C. J. Am. Chem. Soc. 1973, 95,
3687-3692. For an example of a lithium enolate/lithium amide aggregate
structure, see: (f) Williard, P. G.; Hintze, M. J. J. Am. Chem. Soc. 1990,
112, 8602-8604.
(25) Shannon, R. D. Acta Crystallogr. 1976, A32, 751-767.

“New” Catalysts for the Ester-Interchange Reaction
J. Am. Chem. Soc., Vol. 120, No. 24, 1998 5985
Table 4. Effect of Solvent on the Rate of Approach to
Equilibrium for Eq 7a
a Reaction conditions: 0.8 M in each ester, 40 mM NaOtBu.
Reactions were monitored to 25% conversion by GC and analyzed using
standard approach to equilibrium kinetic expressions. b Single measure-
ment.
Figure 3. Log of observed first-order relaxation constants for eq 7
versus log [NaOtBu]. The line through the toluene data is a smooth
curve.
Table 5. Effect of Ester Concentration on the Initial Rate of
Reaction for Eq 7a
Table 6. Initial Rates of Reaction of Eq 7 as a Function of
Temperaturea
a Reaction conditions: 0.8 M in second ester, total volume ) 6.3
mL, [NaOtBu] ) 40 mM, in THF at room temperature.
a Reaction conditions: 0.8 M in each ester, 40 mM in NaOtBu, in
THF.
Scheme 3
the bulky tBuAc would be rate-limiting. The reaction orders
in esters, however, clearly implicate the involvement of MeBz
in the turnover-limiting step (Scheme 3, b).
The rate of equilibration of the process in eq 7 was also
determined as a function of catalyst concentration in various
solvents. In THF a fractional order of 0.85(2) was obtained
from the log kobs vs log [NaOtBu] plot (Figure 3). In hexane
and ether, fractional orders in NaOtBu are also obtained but
are greater than unity (1.2(1) and 1.4(1), respectively). In
toluene a nonlinear relationship was observed.
properties.27 The steric encumbrances of R-methyl substituents
on the solvent result in slight rate increases but are still slower
than reactions carried out in ether, toluene, or hexane. The
overall rate effect spans a factor of ∼5.5. This relatively small
difference is likely due to the dielectric/donor buffering of the
solvent by the 1.6 M concentration of esters.28
The rate law for the reaction in eq 7 can thus be expressed
as
Rate Law Determination. The initial rate of tBuBz forma-
tion (υinitial, M s-1) was determined by independently varying
the concentrations of tBuAc and MeBz over a 10-fold range.
The concentration of the catalyst (40 mM) and nonvarying ester
(0.80 M) were kept constant by varying the volume of THF
added to the reaction solutions. A plot of log(υinitial) vs log-
[tBuAc] and log(υinitial) vs log[MeBz] yielded slopes of -0.1-
(1) and 1.1(1), indicating zero- and first-order dependencies on
tBuAc and MeBz, respectively (Table 5). A simplified mecha-
nistic scenario for the forward direction of this reaction is shown
in Scheme 3. A priori, it was not clear whether addition of a
bulky tert-butoxide to MeBz or the addition of methoxide to
Determination of Activation Energies. The rate of equili-
bration of the process in eq 7 was measured over a 40 °C
temperature range (Table 6), and Eyring analysis yielded
apparent activation energies for the addition of NaOtBu to MeBz
of ∆H‡ ) 10.0(1) kcal mol-1 and and ∆S‡ ) -32(3) eu.
Striking is the large and negative entropy of activation for this
process. Since the preceding rate-law implicated a rate-
determining addition of tert-butoxide to MeBz, the activation
parameters naturally reflect this addition process. What is less
clear, however, is the number of elementary processes that
combine to make up the pathway to the transition state (vide
infra). Therefore, although the activation entropy reflects a good
deal of order being built up in the transition state, this must be
interpreted cautiously.
(26) tert-Butyl isobutyrate has been observed to tetrasolvate (NaOtBu)4,
see ref 11. Complexes wherein tert-butyl isobutyrate and pivalate coordinate
to each lithium atom of lithium hexamethyldisilazide dimers (LiHMDS)2
have been crystallographically and spectroscopically characterized, see:
Williard, P. G.; Liu, Q.-Y.; Lochmann, L. J. Am. Chem. Soc. 1992, 114,
348-350.
(27) Pomarenko, S. M.; Mushtakova, S. P.; Demakhin, A. G.; Faifel, B.
L.; Kal’manovich, D. G. Russ. J. Gen. Chem. 1995, 65, 160-167.
(28) For a tabulation of ∆Hrxn of various nonprotic solvents with BF3,
see: Maria, P.-C.; Gal, J.-F. J. Phys. Chem. 1985, 89, 1296-1304. Relevent
enthalpies in this context are as follows: dichloromethane, 10(3); methyl
benzoate, 59(1); ethyl acetate, 75.5(3); diethyl ether, 78.8(4); THF, 90.4(3)
kcal mol-1.
Linear Free Energy Relationships. Measuring the initial
rate of approach to equilibrium for various para-substituted

5986 J. Am. Chem. Soc., Vol. 120, No. 24, 1998
Stanton et al.
Table 7. Initial Rate of Reaction Eq 8 as a Function of Para
Scheme 4
Substituenta
a Reaction conditions: 0.8 M in each ester, 40 mM in NaOtBu, THF.
methyl benzoates (eq 8) allowed a Hammett plot to be
While interpretation of these data is relatively straightforward
using the simple model in Scheme 4, the analysis becomes more
complex with cluster catalysts. A reasonable chain of events
for a cluster-mediated transesterification has previously been
proposed by Jackman.17 A slightly modified version of this
process is shown in Scheme 5 and highlights (1) the degree of
reorganization that accompanies the addition/elimination se-
quence and (2) the number of potential intermediates that could
be present along the reaction coordinate.
A typical transesterification reaction as shown in Scheme 5
requires three different charged oxygen atoms through the
sequence. The fact that the rates of these reactions are not
accelerated by polar solvents (the opposite is true) suggests that,
despite the shuttling of negative charge, little charge buildup
occurs during the transformation. It is therefore tempting to
speculate that one role of the cluster in accelerating the rate of
this reaction is to provide multiple points of attachment for
neutralizing (i.e., stabilizing) developing negative charge in the
transition state. Such a scheme would not be accessible in a
mononuclear ion pair and thus provides a reasonable rationale
for why clusters are exceptionally reactive. A second role for
the cluster may be that the architecture of the aggregate ensures
that alkoxide delivery to the carbonyl π*-orbital occurs through
a six-membered transition state that delivers the nucleophile
along the stereoelectronically preferred approach trajectory (H,
Scheme 5).31
The rate law, Eyring, and LFE relationships confirm a rate-
determining addition of the metal alkoxide to MeBz. However,
the exact role of the alkoxide nucleophile’s structure and
nuclearity on the reaction energetics is less clear. It has been
established that NaOtBu exists as a tetramer in ether, THF, and
tert-butyl isobutyrate, and a mixture of hexamers and nonamers
in cyclohexane and benzene. The aggregation state of the
clusters in solutions that are 1.6 M in ester is unknown; however,
the rather basic nature of esters28 should tend to destabilize the
higher (>4) aggregates found in nonpolar solvents. Our
working model thus assumes that the major catalyst structure
in solution is a tetrameric cube.
Since the experimentally determined reaction orders in
NaOtBu are close to unity (Figure 3) and structures with an
aggregation state of four are the major solution species, they
point to tetrameric primary reactants. A reasonable explanation
for the deviations from unity in catalyst order focuses on the
equilibrium between the major tetrameric cube and a smaller
or larger cluster reacting at a different rate (on a per alkoxide
basis). When two active species are in equilibrium, the order
in that catalyst is not, strictly speaking, defined. However, if
the rates of catalysis are similar or one is present in a small
concentration, then a perturbation in the log-log plot is
constructed (Table 7). Consistent with the first-order depen-
dence on MeBz in the rate law, an excellent correlation was
observed with the Hammett σ-scale (R ) 0.996) and a F of
+2.35 was obtained. The magnitude and sign of F compares
favorably with the measured F of +2.3 for the saponification
of para-substituted methyl benzoates in 60% aqueous acetone29
and a F of +1.8-2.2 for the transesterification of aryl benzoates
by metal ethoxides in ethanol.30 In both cases, hydroxide and
ethoxide addition are turnover limiting. These data point to a
rate-determining tert-butoxide addition step that is highly
sensitive to the electrophilicity of the carbonyl carbon. In
comparison, the acid-catalyzed hydrolysis of ethyl benzoates
occurs in 60% aqueous acetone solution with a F of +0.11
indicating few parallels to this mechanistic pathway.29
Discussion
A series of experiments probing the mechanistic details of
the alkali-metal alkoxide-cluster-catalyzed ester interchange
reaction have been performed. Kinetic evidence points to rates
that are highly sensitive to the identity of the metal counterions
in the catalyst and to both steric and electronic effects in the
substrates. A rate-determining addition of tert-butoxide to
MeBz is implicated by the orders in ester and the simplistic
scenario in Scheme 3. The involvement of MeBz in the
turnover-limiting step is also consistent with the linear free
energy (LFE) relationship for para-substituted methyl benzoates
(F ) +2.35).
Although kinetic and LFE relationships point to the notion
that MeBz is involved in the rate-limiting step, the putative
involvement of tetrahedral intermediate G suggests two mecha-
nistic possibilities, where the slow step is (a) addition of tert-
butoxide to form G (a, Scheme 4) or (b) its breakdown (b,
Scheme 4). We favor the former scenario for two reasons: (1)
the magnitude and sign of F (+2.35) is most consistent with
the addition of a nucleophile to the electrophilic carbonyl carbon
and (2) the large and negative entropy of activation (-32(3)
eu) argues for substantial entropic reorganization, a scenario
most easily reconciled by a rate determining formation of the
tetrahedral intermediate G.
(29) Jaffe, H. H. Chem. ReV. 1953, 53, 191.
(30) Pregel, M. J.; Dunn, E. J.; Nagelkerke, R.; Thatcher, G. R. J.; Buncel,
E. Chem. Soc. ReV. 1995, 449-455.
(31) Bu¨rgi, H. B.; Dunitz, J. D. Acc. Chem. Res. 1983, 16, 153-161.

“New” Catalysts for the Ester-Interchange Reaction
J. Am. Chem. Soc., Vol. 120, No. 24, 1998 5987
Scheme 5
Scheme 6
Using equilibration rates (eq 7), apparent second-order rate
constants of ∼0.03 (Li+), 0.13 (Na+), 3.2 (K+), 9 (K+), and 14
(Cs+) M-1 s-1 at 25 °C are obtained assuming a tetrameric
alkoxide catalyst. Clearly a stronger alkali-metal effect exists
in the nonpolar to weakly polar aprotic solvents studied herein
as compared to polar protic solvents.
The observation that donor solvents slow the rates of reaction
is consistent with a competitive inhibition for the ester preequi-
librium binding site on the cluster (Scheme 5).26 The central
tenet to this analysis further suggests that prebinding of the
carbonyl oxygen of the ester to the alkali metal is necessary to
ensuring intermediates (charged) of reasonable stability.32
Conventional wisdom would also add that the metal acts as an
electrophilic activator of the carbonyl group, thereby enhancing
the rate of alkoxide addition.26,33 However, the putative role
of the alkali metal as a Lewis acid is less clear in the present
case as rates run opposite the cations’ inherent Lewis acidity,
i.e. lithium, the strongest Lewis acid (size/charge), is the poorest
catalyst while cesium is the best.34
An explanation for the ordering in metal reactivities evolves
from considering the relative interaction of two metal cations
with a single negative charge. From the ion-exchange and ion-
selective electrode literature, a theory formulated by Eisenman
describes the interaction of two ions in solution as an inter-
play between electrostatics and solvation. In the case of low
electric field strength anions (charge diffuse), the electrostatic
terms diminish and cation solvation strength becomes more
important. This leads to weakly solvated cations (Cs+) being
more strongly attracted. In the case of high electric field
strength anions (concentrated charge), the electrostatic terms
dominate and the cation with the smallest size is most strongly
attracted.35
observed. For example, the reaction in hexane gives a catalyst
order of 1.2(1) (Figure 3), consistent with a tetramer in
equilibrium with a small concentration of a more reactive
hexamer. As the catalyst concentration increases, the equilib-
rium shifts toward the more reactive aggregate and a higher
than expected rate ensues (Scheme 6). This is a reasonable
scenario since hexameric lithium aryl oxides have been previ-
ously shown to be up to 5.5 times more reactive than their
tetrameric counterparts.17 Similarly, the higher order in ether
(1.4(1)) suggests that the hexamer/tetramer reactivity ratio is
higher than in hexane. In THF, the order of 0.85(2) is consistent
with two scenarios, equilibrium of the tetramer with (a) a more
reactive dimer or (b) a less reactive hexamer. Both are possible
as precedence for a more reactive dimer exists in lithium aryl
oxide clusters.17 As shown in Figure 3, the log-log plot for
eq 7 is decidedly nonlinear in toluene. While this is most likely
due to relative aggregation and kinetic effects, the structural
data to deconvolute these contributions is not yet available and
we include it for completeness.
Comparison of second-order rate constants for the addition
of metal alkoxides to esters is instructive. In a closely related
system, Jackman measured rate constants that varied from 2 ×
M-1 s-1 to 13 × 10-4 M-1 s-1 in various solvents at 35 °C for
the process in eq 2. Rates as high as 2.9 × 10-3 M-1 s-1 at 35
°C for dimers were also observed.17 Buncel has measured rate
constants for the addition of alkali-metal ethoxides to aryl
benzoates in ethanol and found that the metal effect on rate
was not large (0.31 (Li+), 0.55 (Na+), 0.58 (K+), and 0.48 (Cs+)
M-1 s-1; 25 °C).30 Under these conditions, monomeric contact
ion pairs are in equilibrium with reactive free ethoxide (0.13
M-1 s-1) and the metal enhances the rate of transesterification.
(32) While such a scenario has been assumed (reasonably) for many
years, clear evidence for a prebound ester was found in the case of
organolanthanide catalysts for the ring opening of lactones. In this case,
the intermediate bound lactone could be isolated and spectroscopically
characterized (NMR) and its conversion to a new alkoxide via alkoxide
migration monitored. Yamashita, M.; Takemoto, Y.; Ihara, E.; Yasuda, H.
Macromolecules 1996, 29, 1798-1806.
(33) (a) Shambayati, S.; Crowe, W. E.; Schreiber, S. L. Angew. Chem.,
Int. Ed. Engl. 1990, 29, 256-272. (b) Lefour, J.-M.; Loupy, A. Tetrahedron
1978, 34, 2597-2605.
(34) A further complication is the unkown effect on reactivity of the
structural consequences (on the cluster) of a change in ionic radius in the
cations (Li+ (0.76 Å), Cs+ (1.67 Å)).25

5988 J. Am. Chem. Soc., Vol. 120, No. 24, 1998
Stanton et al.
Scheme 7
In the present case, the solvation terms in the Eisenman
equation are expected to be minimal for two reasons: (1) use
of a relatively nonpolar solvent and (2) the aggregation of the
metal alkoxides which shield the cations from solvation effects.
The equation for the competition of two monovalent cations
with a single anion thus reduces to ∆G°1/2 ≈ 1/(ra + r1) - 1/(ra
+ r2), where ra is the radius of the anion and r1 and r2 are the
radii of cations 1 and 2, respectively (Coulomb’s law). The
concentrated (small) charge in the alkoxide anion maximizes
the difference between the smallest and largest cations and leads
to strong ground-state stabilization by the smaller metals
(Scheme 7).
On the other hand, the negative charge in the transition state
for tert-butoxide addition will be spread among the three oxygen
atoms and result in a larger, more diffuse anion as compared to
the ground-state alkoxides. The effect of the larger anion in
the transition state is to compress the energy difference between
the smallest and largest cations (Scheme 7). This transition-
state compression hypothesis is consistent with the observed
cation reactivities (Cs+ > Rb+ > K+ > Na+ > Li+) and
moreover predicts the linear correlation between log kobs and
1/rionic (Figure 4).36 While this correlation has been observed
by several groups,37 it was Buncel who first applied the
Eisenman theory to the concept of transition-state stabiliza-
tion.30,38
Figure 4. log kobs (one half-life) for eq 7 vs reciprocal six-coordinate
ionic radius of the metal cation25 (0.8 M in each ester, 40 mM MOtBu
(5 mol %), THF).
Reaction orders in catalyst and the known aggregation behavior
of these alkoxides implicate tetrameric clusters as the primary
reactants. These clusters have been shown to be modified under
the reaction conditions to clusters containing mixed tert-butoxy
and methoxy substituents which appear to have enhanced
activities compared to the tetrameric all-tert-butoxide cluster.
The good correlation between log(rate) and 1/rionic suggests a
scenario wherein strong electrostatic interactions occur in the
ground state and are diminished in the transition state, leading
to the observed ordering in rate (Cs+ > Rb+ > K+ > Na+
>
Li+).
We propose that the roles of the cluster in facilitating the
rate of transesterification, are
(1) To supply multiple cations within the cluster framework
to serve as charge neutralizing sites. These sites stabilize
developing negative charge in the transition state.
(2) To provide scaffolding for first precoordinating the ester,
potentially activating it, and then delivering the alkoxide
nucleophile via a six-membered transition state along the
stereoelectronically preferred Bu¨rgi-Dunitz angle.
Our efforts are currently focused on identifying and structur-
ally characterizing cluster intermediates. These results will be
reported in due course.
Summary
We have shown that alkali-metal alkoxide clusters function
as highly active catalysts for the ester interchange reaction.
Mechanistic investigations implicate a stepwise process for our
test reaction wherein tert-butoxide addition to methyl benzoate
is turnover limiting and is sensitive to para substituents.
Experimental Section
Materials and Methods. All solvents were prepared by first purging
reagent grade solvents with dry nitrogen and then dried by passing
over a column of activated alumina,39 with the exception of DME,
tBuOMe, 2-Me-THF, and 2,5-Me2-THF, which were dried and vacuum
distilled from purple sodium/benzophenone ketyl solutions. Esters were
purchased from Aldrich and distilled over CaH2. Alkali-metal alkoxide
catalysts were prepared from the corresponding metal and tert-butyl
alcohol, purified by sublimation, and stored in an inert atmosphere
glovebox. Potassium tert-butoxide solutions (1 M in THF) were
purchased from Aldrich. Reactions were carried out in an inert argon
atmosphere using standard Schlenk line techniques. Para-substituted
tert-butyl benzoates for GC calibration were synthesized as described
in ref 3b.
(35) The complete Eisenman equation for the competition of two cations
for a single anion is ∆G°1/2 ) e2/(ra + r2) - e2/(ra + r1) - (∆G2 - ∆G1),
where e is the electronic charge, ra is the radius of the anion, r1 and r2 are
the radii of cations 1 and 2, respectively, and ∆G1 and ∆G2 are the solvation
energies of cations 1 and 2 (Coulomb’s law).
(36) While 1/rionic is not mathematically equivalent to the simplified
Eisenman equation, substituting various pairs of reasonable anion sizes into
this expression yields families of curves which correlate closely to 1/rionic
(R ∼ 0.99).
(37) (a) Kurts, A. L.; Sakembaeva, S. M.; Reutov, O. A. J. Org. Chem.
USSR (Engl.) 1974, 10, 1572-1580. (b) Powell, D. G.; Warhurst, E. Trans.
Faraday Soc. 1962, 58, 953-956. (c) Mathias, A.; Warhurst, E. Trans.
Faraday Soc. 1962, 58, 948-952.
Gas chromatography was performed on a Hewlett-Packard 6890 GC
equipped with an HP-5 column (30 m × 0.32 mm) and an HP 6890
series autosampler. 1H NMR spectra were obtained at 200 MHz on a
(38) Attempts to fit log(kobs) vs [1/(rionic + ralkoxide) - 1/(rionic + rTS)]
(the rigorous anion size dependence in the electrostatic term) to extract
information about the size of the alkoxide (ralkoxide) and transition-state anions
(rTS) were unsuccessful. The flat potential energy surfaces were not sensitive
to changes in anion sizes and did not satisfactorily converge on reasonable
ralkoxide and rTS values.
(39) Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.;
Timmers, F. J. Organometallics 1996, 15, 1518-1520.

“New” Catalysts for the Ester-Interchange Reaction
J. Am. Chem. Soc., Vol. 120, No. 24, 1998 5989
Bruker WP 200 instrument. 13C NMR spectra were obtained at 50
MHz on a Bruker WP 200 instrument.
Linear Free Energy Relationships. Rate measurements were as
described in Typical Kinetic Experiment using sodium tert-butoxide
(5 mol %) as catalyst, and initial rates were determined from the first
half-life. Integration percentages were corrected for response factor
differences between the substituted methyl benzoate and the corre-
sponding tert-butyl benzoate using authentic samples of the substituted
tert-butyl benzoates.3b
Typical Kinetic Experiments. A typical kinetic measurement was
run as follows: Methyl benzoate (5 mmol, 0.63 mL) and tert-butyl
acetate (5 mmol, 0.67 mL) were combined in a round-bottomed flask
and diluted with 2.5 mL of the desired solvent. In a separate flask,
MOtBu (0.025 mmol) was dissolved in 2.5 mL of the desired solvent.
The esters (0.8 M in each ester) were transferred via syringe to the
catalyst solution. Aliquots were withdrawn via syringe, quenched in
saturated aqueous NH4Cl solution, diluted with ethyl acetate, and
analyzed by GC to monitor the conversion of methyl benzoate to tert-
butyl benzoate. Integration percentages were corrected for response
factor differences between tert-butyl benzoate and methyl benzoate,
i.e. %PhCO2Me (GC)/%PhCO2tBu (GC) ) 1.46 × %PhCO2Me
(actual)/% PhCO2tBu (actual).
Solvent Effects on Rate. Experimental conditions for carrying out
these measurements were identical to those described above except for
the substitution of THF with the solvent of choice.
Determination of Activation Energies. Rate measurements were
made as described above. The temperature of the reaction was
controlled using a Neslab cryobath, and the kinetic data were extracted
using the initial rate to 1 half-life and fitted to the linearized form of
the Eyring equation.
Acknowledgment. Support from the University of North
Carolina at Chapel Hill is gratefully acknowledged. M.R.G. is
an NSF CAREER Award recipient.
Supporting Information Available: Kinetic plot highlight-
ing the deviations from linearity for a simple first-order
relaxation analysis, log-log plot to determine the kinetic order
in esters, Eyring plot for the process in eq 7, and a Hammett
plot for the substrates in eq 8 (2 pages, print/PDF). See any
current masthead page for ordering information and Web access
instructions.
JA980584T