Effects of medium on decarboxylation kinetics: 3-Carboxybenzisoxazoles and their potential use as environmental probes in biochemistry

Article abstract of DOI:10.1021/jo9918798

The decarboxylation rate of the tetramethylguanidinium salt of 3- carboxy-6-nitrobenzisoxazole in 24 pure solvents and 36 dimethyl sulfoxide binary mixtures with diglyme, acetonitrile, benzene, dichloromethane, chloroform, and methanol was analyzed in the light of the SPP, SA, and SB pure solvent scales. The results allow one to rationalize the high sensitivity of this kinetics to the reaction medium and to assess the potential use of this compound as a probe in biochemical environments. The natural environment for comparison of this kinetics was found to be the gas phase rather than the aqueous medium. In the latter, the process is much faster owing to such high polarity, which, however, is strongly diminished by the high acidity of the medium. Based on our calculations, the rate constant for the decarboxylation kinetics in the gas phase must be in the region of 2 x 10^-10 s^-1 (i.e., 3 orders of magnitude smaller than in water).

Full text of DOI:10.1021/jo9918798

J . Org. Chem. 2000, 65, 3409-3415
3409
Effects of Med iu m on Deca r boxyla tion Kin etics:
3-Ca r boxyben zisoxa zoles a n d Th eir P oten tia l Use a s
En vir on m en ta l P r obes in Bioch em istr y
J . Catala´n,*,† C. D´ıaz,† and F. Garc´ıa-Blanco‡
Departamento de Quı´mica Fı´sica Aplicada, Universidad Auto´noma de Madrid, Cantoblanco,
E-28049 Madrid, and Departamento de Fı´sico-Quı´mica, Facultad de Farmacia, Universidad Complutense
de Madrid, E-28040 Madrid, Spain
Received December 7, 1999
The decarboxylation rate of the tetramethylguanidinium salt of 3-carboxy-6-nitrobenzisoxazole in
24 pure solvents and 36 dimethyl sulfoxide binary mixtures with diglyme, acetonitrile, benzene,
dichloromethane, chloroform, and methanol was analyzed in the light of the SPP, SA, and SB pure
solvent scales. The results allow one to rationalize the high sensitivity of this kinetics to the reaction
medium and to assess the potential use of this compound as a probe in biochemical environments.
The natural environment for comparison of this kinetics was found to be the gas phase rather
than the aqueous medium. In the latter, the process is much faster owing to such high polarity,
which, however, is strongly diminished by the high acidity of the medium. Based on our calculations,
the rate constant for the decarboxylation kinetics in the gas phase must be in the region of 2 ×
10-10 s-1 (i.e., 3 orders of magnitude smaller than in water).
In tr od u ction
carboxylate ion to give carbon dioxide and an organic
residue containing an unshared pair of electrons. As a
rule, the organic product stabilizes this electron pair by
delocalization, which is especially significant in enzymic
reactions, where an electron sink is usually provided by
a coenzyme (e.g., pyridoxal 5′-phosphate, thymine pyro-
phosphate), by a metal ion bound to an enzyme (typically
an oxidative decarboxylase), or by a prosthetic group (e.g.,
a lysine amino group or covalently bound pyruvate). That
decarboxylation reactions, whether catalyzed or other-
wise, are especially significant in the biological and
organic synthesis fields is currently a widespread percep-
tion. Consequently, any contribution to improving our
understanding of these processes will be equally signifi-
cant.
In 1934, Verhoek1 published the first of a series of
papers in which he demonstrated that the decarboxyla-
tion kinetics of trichloroacetic,1-3 trifluoroacetic,4 and
trinitrobenzoic acids5,6 and of sodium p-toluenesulfonyl
acetate7 are first-order in the concentration of the car-
boxylate ion and, especially, that the decarboxylation rate
is sensitive to the reaction environment. Thus, at 75 °C,
sodium p-toluenesulfonyl acetate is decarboxylated roughly
10 times faster in ethyleneglycol (k ) 0.481 × 10-5 s-1
)
than in water (k ) 0.0542 × 10-5 s-1); also, under these
conditions, trinitrobenzoate is decarboxylated about 1000
times faster in a 20:80 water/1,4-dioxane mixture (k )
2450 × 10-5 s-1) than in pure water (k ) 2.30 × 10-5
s-1). Doering and Pasternak8 found the decarboxylation
of 2-methyl-2(2-pyridyl)butyric acid to be faster in neutral
aqueous solutions than in strongly acidic or alkaline
solutions and thus identified the reactive species with a
zwitterion (I). A similar conclusion was reached by
Crosby et al.9 in relation to the decarboxylation of 2-(1-
carboxy-1-hydroxyethyl)-3,4-dimethylthiazolium chloride
(II), a prototype of enzymic decarboxylation of the pyru-
vate-thiamine complex.
In some cases, the rate of the decarboxylation process
depends very strongly on the particular reaction environ-
ment. Thus, the decarboxylation rate of the pyruvate-
thiamine complex at 26 °C is 9000 times greater in
ethanol (k ) 0.180 min-1) than in water (k ) 2 × 10-5
min-1).9 Kemp and Paul10 demonstrated that the decar-
boxylation rate of the tetramethylguanidinium salt of
3-carboxy-6-nitrobenzisoxazole at 30 °C increases by a
factor of 94 000 000 in going from water (k ) 7.4 × 10-6
s-1) to hexamethylphosphoramide (k ) 7.0 × 102 s-1); the
rate of this process relative to water can be adjusted by
using appropriate solvents. The extremely interesting
results reported by Kemp and Paul led some to consider
the decarboxylation of 3-carboxy-6-nitrobenzisoxazole a
highly suitable process for probing not only solvents but
also such varied media as micelles, bilayers, macrocyclic
In its most simple form, the decarboxylation process
involves the cleavage of a carbon-carbon bond in a
† Universidad Auto´noma de Madrid.
‡ Universidad Complutense de Madrid.
(6) Trivich, D.; Verhoek, F. H. J . Am. Chem. Soc. 1943, 65, 1919.
(7) O’Connor, D. J .; Verhoek, F. H. J . Am. Chem. Soc. 1958, 80, 288.
(8) Doering, W. v. E.; Pasternak, V. Z. J . Am. Chem. Soc. 1950, 72,
143.
(9) Crosby, J .; Stone, R.; Lienhard, G. E. J . Am. Chem. Soc. 1970,
92, 2891.
(1) Verhoek, F. H. J . Am. Chem. Soc. 1934, 56, 571.
(2) Verhoek, F. H. J . Am. Chem. Soc. 1945, 67, 1062.
(3) Hall, G. A.; Verhoek, F. H. J . Am. Chem. Soc. 1947, 69, 613.
(4) Auerbach, I.; Verhoek, F. H.; Henne, A. L. J . Am. Chem. Soc.
1950, 72, 299.
(5) Verhoek, F. H. J . Am. Chem. Soc. 1939, 61, 186.
(10) Kemp, D. S.; Paul, K. G. J . Am. Chem. Soc. 1975, 97, 7305.
10.1021/jo9918798 CCC: $19.00 © 2000 American Chemical Society
Published on Web 05/09/2000

3410 J . Org. Chem., Vol. 65, No. 11, 2000
Catala´n et al.
Sch em e 1. Deca r boxyla tion of th e
Tetr a m eth ylgu a n id in iu m Sa lt of
3-Ca r boxy-6-n itr oben zixosa zole
romethane, chloroform, and methanol. They also studied
the decarboxylation of four other tetramethylguani-
dinium salts of 3-carboxy-6-X-benzisoxazole (with X )
NH2, MeO, H, and Cl) to obtain additional information
on the influence of various electronic effects of the
benzisoxazole derivative on the decarboxylation process.
All of this information is thoroughly analyzed in this
work with a view to its rationalization and to assessment
of the potential of this process for studying catalytic and
noncatalytic biochemical environments.
The amazing results of Kemp and co-workers10,28,29 for
the decarboxylation of the tetramethylguanidinium salt
of 3-carboxy-6-nitrobenzisoxazole allowed three factors
to be identified as the sources of the increased reaction
rates relative to water: (a) the decreased acidity of the
medium, which is bound to accelerate the reaction
through a less extensively interaction with the anion (the
reactive species); (b) the ion-pair interactions arising from
the presence of the anion and the tetramethylguani-
dinium cation, which will also be modulated by the
nature of the solvent; and (c) the stabilization of the
transition state of the process through dispersion interac-
tions. Kemp and Paul10 tried to rationalize their rates
by using the single-parameter solvent scales available
at the time, the ET(30) scale of Reichardt30 and the Z scale
of Kosower,31 but found no correlation. In response,10 they
introduced a new scale based on the solvatochromism of
the decarboxylation product of 2-cyano-5-nitrophenolate,
which they called the “H scale”. However, the 24 solvents
studied clustered in three nearly parallel arrangements
with no physical significance, at least not in relation to
the nature of the solvent in each group.
hosts, and polymers with respect to water.11-22 In addi-
tion, some polymers accelerate the reaction in organic
solvents.23,24 Likewise, the reaction rate can be increased
by a factor about 19 000 relative to water inside a binding
pocket in a catalytic antibody.25 The ease with which
antibodies can be genetically engineered makes these
systems highly suitable for studying the effects of sol-
vents as their catalytic properties are dictated by the
protic character of the solvent.
The monoclonal antibody 21D8 has been used as a very
simple system to study the effect of solvation on enzyme
catalysis.25 By comparison with the rate constant relative
to water, the occurrence of a hydrogen bond in the
binding site between the carboxylate group and lysine
or arginine was concluded. The presence of this bond
decreases the rate of decarboxylation.
The efficiency for proton transfer catalysis in enzymes
and models is deeply analyzed by Kirby.26 Hollfelder et
al.27 reported that serum albumin proteins catalyze the
conversion of Kemp’s reaction, but they use a lysine side
chain as the catalytic general base rather than the
carboxylate group, thereby allowing the contribution of
the medium effect for this catalysis.
Kemp and co-workers10,28,29 examined the decomposi-
tion of the tetramethylguanidinium salt of 3-carboxy-6-
nitrobenzisoxazole to form carbon dioxide and the cor-
responding 2-cyano-5-nitrophenolate (Scheme 1) in a
series of 24 pure solvents including polar and nonpolar
solvents with protic, nonprotic, and amphiprotic connota-
tions, as well as in binary mixtures of dimethyl sulfoxide
(DMSO) with diglyme, acetonitrile, benzene, dichlo-
To the authors’ knowledge, there have been only three
subsequent attempts at rationalizing the behavior of the
tetramethylguanidinium salt of 3-carboxy-6-nitroben-
zisoxazole in the solvent series studied by Kemp and
Paul.10 In 1993, Grate et al.32 used parameters π*, δ, R,
â, and δ2H in the solvent scheme of Taft and Kamlet33,34
with 20 of the 24 solvents formerly studied by Kemp and
Paul.10 In 1994, Drago et al.35 analyzed the 24 solvents
examined by these authors, using parameters EB, E′A,
CB, C′A, and S′ in their own solvent scheme. Finally, also
in 1994, Famini and Wilson36 analyzed the solvents on
(11) Bunton, C. A.; Minch, M. J .; Hidalgo, J .; Sepu´lveda, L. J . Am.
Chem. Soc. 1973, 95, 3262.
(12) Bunton, C. A.; Kamego, A. A.; Minch, M. J .; Wright, J . L. J .
Org. Chem. 1975, 40, 1321.
(13) Germani, R.; Ponti, P. P.; Savelli, G.; Spreti, N.; Cipiciani, A.;
Cerichelli, G.; Bunton, C. A. J . Chem. Soc., Perkin Trans. 2 1989, 1767.
(14) Germani, R.; Ponti, P. P.; Romeo, T.; Savelli, G.; Spreti, N.;
Cerichelli, G.; Luchetti, L.; Mancicni, G.; Bunton, C. A. J . Phys. Org.
Chem. 1989, 42, 553.
the basis of their solvent descriptors Mmc, π1, δ2 , ꢀB, ꢀA,
H
q-, and q+. As shown below, these analyses led them to
classify the solvents in different families outside which
the fitting of their results was rather poor.
(15) Kunitake, T.; Shinkai, S.; Klotz, I. M. J . Org. Chem. 1977, 42,
306.
(16) Kunitake, T.; Okahata, R. A.; Shinkai, S.; Hirakawa, S. J . Am.
Chem. Soc. 1980, 102, 7877.
(17) Schmidtchen, F. P. J . Chem. Soc., Perkin Trans. 2 1986, 135.
(18) Shad, S. C.; Smid, J . J . Am. Chem. Soc. 1978, 100, 1426.
(19) Smid, J . J . Am. Chem. Soc. 1975, 97, 5932.
(20) Straub, T. S.; Bender, M. L. J . Am. Chem. Soc. 1972, 102, 7877.
(21) Suh, J .; Scarpa, I. S.; Klotz, I. M. J . Am. Chem. Soc. 1976, 98,
7060.
(22) Hollfelder, F.; Kirby, A. J . J . Am. Chem. Soc. 1997, 119, 9578.
(23) Shirai, M.; Smid, J . J . Polym. Sci., Polym. Lett. Ed. 1980, 18,
659.
As shown in discussing the reaction model below, the
decarboxylation process is complicated by the occurrence
of side effects such as the presence or absence of an ion-
pair; this depends on the particular nature of the solvent,
which, among others, dictates the concentration of the
active species (the anion). Because the reactants and
products in the presence of solvent are linked by several
simultaneous equilibria, the solvent will exert various
modulating effects; fortunately, the effects will be gov-
(24) Smid, J .; Varma, A.; Shah, S. C. J . Am. Chem. Soc. 1979, 101,
5764.
(30) Reichardt, C. Chem. Rev. 1994, 94, 2319 and references therein.
(31) Kosower, E. M. J . Am. Chem. Soc. 1958, 80, 3253.
(32) Grate, J . W.; McGill, R. A.; Hilvert, D. J . Am. Chem. Soc. 1993,
115, 8577.
(33) (a) Kamlet, M. J .; Taft, R. W. J . Am. Chem. Soc. 1976, 98, 377.
(b) Taft, R. W.; Kamlet, M. J . J . Am. Chem. Soc. 1976, 98, 2886.
(34) Kamlet, M. J .; Abboud, J . M. L.; Abraham, M. H.; Taft, R. W.
J . Org. Chem. 1983, 48, 2877.
(25) (a) Lewis, C.; Kramer, T.; Robinson, S.; Hilvert, D. Science 1991,
253, 1019. (b) Thorn, S. N.; Daniels, R. G.; Auditor, M.-T. M.; Hilvert,
D. Nature 1995, 373, 228. (c) Blackburn, G. M.; Dalta, A.; Denham,
H.; Wenworth, P., J r. Adv. Phys. Org. Chem. 1998, 31, 249.
(26) Kirby, A. J . Acc. Chem. Res. 1997, 30, 290.
(27) Hollfelder, F.; Kirby, A. J .; Tawfik, D. S. Nature 1996, 383, 60.
(28) Kemp, D. S.; Cox, D. D.; Paul, K. G. J . Am. Chem. Soc. 1975,
97, 7312.
(35) Ferris, D. C.; Drago, R. S. J . Am. Chem. Soc. 1994, 116, 7509.
(36) Famini, G. R.; Wilson, L. Y. J . Chem. Soc., Perkin Trans. 2 1994,
1641.
(29) Kemp, D. S.; Reczek, J .; Vellaccio, F. Tetrahedron Lett. 1978,
8, 741.

Effects of Medium on Decarboxylation Kinetics
J . Org. Chem., Vol. 65, No. 11, 2000 3411
Sch em e 2
erned by the same interactions with the solvent, so a
precise enough one can be accomplished by using pure-
solvent scales such as the SPP,37,38 SB,39 and SA.40,41
This work was undertaken with the aim of rational-
izing the decarboxylation of the trimethylguanidinium
salt of 3-carboxy-6-nitrobenzisoxazole not only in the 24
pure solvents studied by Kemp and Paul10 but also in
binary mixtures of DMSO with diglyme, acetonitrile,
benzene, dichloromethane, chloroform, and methanol.
Exp er im en ta l Section
All solvents used were of the highest available purity and
purchased from Merck in Uvasol grade. Solvent mixtures were
prepared from freshly opened bottles, using Brand II 25.00
mL burets to transfer the liquids.
Polarity (SPP), basicity (SB), and acidity parameters (SA)
were obtained from the wavenumbers of the UV-vis absorp-
tion maxima for the following probe/homomorph couples: 2-(di-
methylamino)-7-nitrofluorene/2-fluoro-7-nitrofluorene,37,38 5-ni-
troindoline/1-methyl-5-nitroindoline,39 and o-tert-butylstilba-
zolium betaine dye/o,o′-di-tert-butylstilbazolium betaine dye.40
The last couple was replaced with the probe 3,6-diethyl-1,2,4,5-
tetrazine41 in the DMSO/methanol mixture at DMSO concen-
trations below 8 M.
UV-vis measurements were made on a Shimadzu 2100
spectrophotometer, the monochromator of which was cali-
brated by using the 486.0 and 656.1 nm lines from a deuterium
lamp. The instrument was routinely checked for wavelength
accuracy by using holmium oxide and didymium filters. All
spectral measurements were performed at 25 °C, using a
matched pair of quartz cells of 1 cm path length.
F igu r e 1. Plot of the log k values for the decarboxylation of
the tetramethylguanidinium salt of 3-carboxy-6-nitrobenzisox-
azole in 24 pure solvents against the corresponding ET(30)
values (O, protic solvents).
tion step involves charged species, the reaction rate can
be expected to vary with increasing solvent polarity. In
summary, step b will be clearly unfavored by an in-
creased solvent acidity and very likely favored by an
increased polarity.
Step a is quite important with a view to rationalizing
the rate of the decarboxylation process; in fact, it
produces the active species (the carboxylate ion, which
is initially stabilized by formation of an ion pair with the
protonated tetramethylguanidine salt). This step will
obviously be favored by the solvent cleaving the ion pair,
both through its polarity and through its acidity, which
will help stabilize the carboxylate ion. A similar effect
will be exerted by the solvent basicity, which will stabilize
the protonated form of tetramethylguanidine.
In summary, the solvent increases the rate of the
decarboxylation process through its polarity and basicity
but also decreases it in sequestering the active species
(the carboxylate ion) through its acidity. As noted earlier,
some effects of the solvent acidity favor the process,
whereas others deactivate it. On the basis of Palit’s rule,42
however, the deactivating effect must prevail, since those
solvents that block the active site of a reactant suppress
its reactivity. On the basis of the previous reasoning, the
decarboxylation rate of the tetramethylguanidinium salt
of 3-carboxy-6-nitrobenzisoxazole must be increased by
increased solvent polarity and basicity and decreased by
an increased acidity.
Deca r boxyla tion Ra te ver su s Solven t Sca les. In -
ter p r eta tion Ba sed on Sin gle-P a r a m eter Sca les.
Kemp and Paul10 used two single-parameter scales to
rationalize the decarboxylation rate of 3-carboxy-6-ni-
trobenzisoxazole, viz., the ET(30) scale of Reichardt30 and
the Z scale of Kosower.31 Figure 1 shows the variation of
the logarithmic constants in the 24 pure solvents studied
by these authors10 against the corresponding ET(30)
values recently compiled by Reichardt.30 As formerly
Resu lts a n d Discu ssion
Solven t Sen sit ivit y of t h e Deca r b oxyla t ion
Sch em e. The decarboxylation of the tetramethylguani-
dinium salt of 3-carboxy-6-nitrobenzisoxazole studied by
Kemp and co-workers10,28 can be schematized as follows
(Scheme 2):
In step a , the carboxylate ion (the reactive species) is
released from the ion pair to an extent dependent on the
equilibrium constant of the process; in step b, the
decomposition of the carboxylate ion produces carbon
dioxide and the organic anion. Because of that, it is
important to determinate the influence of the environ-
ment on steps a and b or, in other words, on the overall
decarboxylation process.
The decarboxylation step, b, will obviously be deter-
mined by the solvent acidity, which will have two effects,
the solvation of the carboxylate ion (the active species)
and its charge dispersion. This dual effect eventually
results in a decrease of the decarboxylation rate when
the solvent acidity is increased. Also, because this reac-
(37) Catala´n, J .; Lo´pez, V.; Pe´rez, P.; Mart´ın-Villamil, R.; Rodr´ıguez,
J .-G. Liebigs Ann. 1995, 241.
(38) Catala´n, J .; Lo´pez, V.; Pe´rez, P. Liebigs Ann. 1995, 793.
(39) Catala´n, J .; D´ıaz, C.; Lo´pez, V.; Pe´rez, P.; de Paz, J .-L. G.;
Rodr´ıguez, J .-G. Liebigs Ann. 1996, 1785.
(40) Catala´n, J .; D´ıaz, C. Liebigs Ann. 1997, 1941.
(41) Catala´n, J .; D´ıaz, C. Eur. J . Org. Chem. 1999, 885.
(42) Palit, S. R. J . Org. Chem. 1947, 12, 752.

3412 J . Org. Chem., Vol. 65, No. 11, 2000
Catala´n et al.
noted by Kemp et al., there is absolutely no correlation
between these two data sets. However, it should be noted
that, if chloroform and dichloromethane are excluded, the
solvents cluster in two groups. One group exhibits an
increase in decarboxylation rate with increase in solvent
polarity, whereas the other encompasses protic solvents
and reflects a marked decrease in reaction rate with
increasing solvent acidity. This is consistent with the fact
that ET(30) is virtually a 50-50 combination of solvent
polarity and acidity.41 Similar results were obtained by
plotting the data for the 19 solvents with reported
Kosower’s Z values.
In view of the poor results obtained by Kemp and
Paul10 with these two solvent scales, they used the
similarity principle to represent the effect of the solvent
on the decarboxylation kinetics in terms of the influence
of the former on the organic anion resulting from the
decarboxylation, as measured by the energy, in kcal/mol,
of its first electronic transition, which they called H. The
end result was that the 24 solvents clustered in three
nearly parallel straight lines with no physical signifi-
cance, at least not in relation to the nature of the solvents
in each group. This treatment is thus seemingly useless
with a view to rationalizing the decarboxylation data.
In conclusion, the single-parameter scales appear not
to be flexible enough to provide an accurate description
of the complicated decarboxylation process studied by
Kemp and Paul.10
F igu r e 2. Plot of ln k in 20 solvents as a function of S ′, E,
and C parameters.
The positive signs of the polarity and basicity coef-
ficients and the negative signs of the polarizability and
acidity coefficients are quite consistent with the above-
described solvation model.
In ter p r eta tion Ba sed on Mu ltip a r a m eter Sca les.
Grate et al.32 used the multiparameter model of Taft and
Kamlet33,34 to account for this kinetics, using the follow-
ing general equation:
Ferris and Drago35 examined the problem in the light
of the following scheme:
∆ø ) (EΑ*EΒ) + CΑ*C Β + S′P + W
(4)
∆ø ) constant + (Sπ* + dδ) + aR1 +bâ1 + hδ2H (1)
where the solvent shift, ∆ø, can be described in terms of
electrostatic (EA × EB), covalent (CA × CB), or polarity
contributions (S′). These authors developed an elegant
treatment by which they divided the 24 solvents into
three groups. In one, which included the highly polar and
basic solvents (hexamethylphosphoramide, N-methylpyr-
rolidine, N,N-dimethylacetamide, N,N-dimethylforma-
mide, DMSO, and sulfolane), the ion pair is assumed to
be virtually completely dissociated and ln k to be de-
scribed by polarity interactions through the equation ln
k ) - 8.24S′ + 27.24. In the second group, which
comprises the scarcely polar, scarcely basic solvents
(tetrachloromethane, benzene, ether, dimethoxymethane,
1,4-dioxane, tetrahydrofuran, diglyme, acetone, benzo-
nitrile, acetonitrile, and nitromethane), the ion pair is
assumed to be only partially dissociated and the rate to
be described by the equation ln k ) -1.86EB + 3.75CB +
5.15S′ - 14.35. Finally, in the third group, formed by the
protic solvents (chloroform, dichloromethane, ethanol,
methanol, water, formamide, and N-methylformamide),
the decarboxylation rate conforms to the equation ln k
) 1.03EA′ - 6.74CB′ + S′P + W. However, application of
this treatment to the 20 solvents for which definitive
parameter values are available provides the following fit
where parameters π* and δ, R1, â1, and δ2 represent the
H
dipolarity and polarizability, the acidity, basicity, and
cohesive energy of the solvents. They applied this model
to 20 of the 24 solvents studied by Kemp and Paul,10
excluding N-methylformamide, sulfolane, dimethoxy-
methane, and diglyme, and found the data to fit the
equation
log k ) 5.45((0.89)π* - 1.46((0.45)δ -
3.03((0.65)R + 1.80((0.66)â -
1.06((0.25)δ2H -2.97((0.52) (2)
with n ) 20, r ) 0.976, and SD ) 0.58.
However, for consistency with subsequent comparisons,
which were to include all 24 solvents, we added the four
solvents initially excluded from the fitting, using the
parameter values reported by Marcus.43 By exception,
those for dimethoxymethane, which were unavailable,
were replaced with the values for diethoxyethane. The
δ2H values for these four solvents were obtained from
Famini and Wilson.36 The data fitted the resulting
equation
ln k ) - 0.71((1.74)S′ + 6.07((1.96)E -
2.56((2.44)C - 6.75((4.58) (5)
log k ) 5.23((0.95)π* - 1.58((0.62)δ -
3.77((0.65)R +1.36((0.74)â - 0.84((0.27)δ2H -
2.80((0.56) (3)
with n ) 20, r ) 0.649, and SD ) 4.42, which does not
reproduce the experimental results (see Figure 2). It
should be noted that the uncertainty in the polarity
coefficient is more than twice the coefficient itself, so its
values make no sense.
with n ) 24, r ) 0.959, and SD ) 0.70.
(43) Marcus, Y. Chem. Soc. Rev. 1993, 22, 409.

Effects of Medium on Decarboxylation Kinetics
J . Org. Chem., Vol. 65, No. 11, 2000 3413
Ta ble 1. Resu lts Obta in ed by F ittin g Da ta for 3-Ca r boxy-6-X-ben zisoxa zole Der iva tives to SP P , SA, a n d SB Sca les in 14
Solven ts Accor d in g to log k ) a SP P + bSA + cSB + log k0
X )
NO2
Cl
H
OMe
NH2
a
b
c
11.19 ( 4.12
-6.57 ( 0.78
0.90 ( 1.16
-9.40 ( 4.05
14
11.72 ( 4.23
-6.50 ( 0.80
0.71 ( 1.19
-10.20 ( 4.15
14
11.22 ( 4.21
-6.12 ( 0.80
0.50 ( 1.19
-10.67 ( 4.14
14
11.72 ( 4.04
-6.14 ( 0.77
0.50 ( 1.14
-10.84 ( 3.97
14
11.60 ( 4.13
-5.80 ( 0.78
0.21 ( 1.17
-11.22 ( 4.05
14
log k0
n
r
0.966
0.963
0.959
0.963
0.957
SD
0.77
0.79
0.79
0.76
0.77
Famini and Wilson36 use a linear solvation energy
relation similar to that of Taft and Kamlet33 but based
on a series of descriptors including Vmc, π1, ꢀB, q-, ꢀA, q+,
and δ2H, which denote the molecular volume, polariz-
ability, covalent basicity, electrostatic basicity, covalent
acidity, electrostatic acidity, and cohesive energy, many
of which were evaluated on a semiempirical level using
the MNDO method.44 The equation obtained by fitting
the values for the 24 solvents was
At this point one will wonder why the decarboxylation
rate in water is so small. On the basis of the previous
fitting, it must be a result of the high polarity of water
(SPP ) 0.962) increasing the rate by 10 logarithmic units
and its high acidity (SA ) 1.062) decreasing it by more
than 6 logarithmic units. It therefore seems inappropriate
to continue to use water as the reference solvent for the
decarboxylation rate of 3-carboxy-6-nitrobenzisoxazole
when it seems clear that its small value in water is a
result of the two different properties. Consequently, it
seems advisable to refer rates to the natural medium for
comparing solvent effects, i.e., the absence of solvent (gas
phase).
log k ) 0.26((1.23)Vmc - 25.07((54.81)π1 +
48.09((31.37)ꢀB + 4.87((2.07)q- +
27.21((27.87)ꢀA - 18.14((6.48)q+ -
Kemp and Paul10 analyzed the decarboxylation rate of
four other 3-carboxy-6-X-benzisoxazoles (with X ) NH2,
MeO, H, and Cl) in 14 different solvents; if these data
are supplemented with those for the nitro derivative, a
set of five derivatives with electronic substituent effects
ranging from a typical electron-releasing function (the
amino group) to a typical electron acceptor (the nitro
group) becomes available. Table 1 gives the results
obtained by fitting the data for these five derivatives in
the 14 solvents to the SPP, SA, and SB scales. The
decarboxylation of the five derivatives is similarly sensi-
tive to the solvent: the rate of the process increases with
increasing polarity and basicity and decreases with
increasing acidity, the prevailing effect being that of
polarity, followed by that of acidity. Because some of
these derivatives contribute active sites to the solvating
effect, the small changes in some of these terms are
rather difficult to rationalize. Also, because, as noted
earlier, the independent term (log k0) represents the rate
of the process in the gas phase and is thus an intrinsic
value, changes in the term must be related to the
electronic properties of the substituents. As can be seen
from Table 1, the independent term varies markedly with
the nature of the solvent. Also, as shown in Figure 3,
these terms vary linearly with parameter σp for the
substituent. This behavior confirms that our scales
accurately reproduce the behavior in the gas phase.
Kemp and Paul10 also reported the decarboxylation
rates for the tetramethylguanidinium salt of 3-carboxy-
6-nitrobenzisoxazole in a series of binary mixtures of
DMSO with diglyme, acetonitrile, benzene, dichlo-
romethane, chloroform, and methanol. Figure 2 in their
paper10 shows the variation of log k with the DMSO
concentration in each mixture. Such a figure warrants
some interesting comments. Thus, as the DMSO (SPP )
1, SB ) 0.647, SA ) 0.072) is enriched with acetonitrile
(SPP ) 0.895, SB ) 0.286, SA ) 0.044), the reaction rate
undergoes a slight, monotonic decrease resulting from the
lower basicity and polarity of the latter. On the other
hand, enriching the DMSO with chloroform (SPP )
0.786, SB ) 0.071, SA ) 0.047) results in a marked
0.95((1.06)δ2H - 9.08((4.44) (6)
with n ) 24, r ) 0.876, and SD ) 1.27. It should be noted
that uncertainty in the polarity in this equation is more
than twice the coefficient itself, and so is the case with
the molar volume (Vmc), whereas uncertainty is of the
same order for the solvent cohesive energy (δ2H), i.e., the
fitting results in a high uncertainty in log k.
The analysis of the decarboxylation rates in the 24
solvents based on the SPP, SA, and SB pure solvent
scales based on the scheme
log k ) a‚SPP + b‚SA + c‚SB + log k0
(7)
proposed by Catala´n et al.38,40,41 results in the following
fitting:
log k ) 10.37((1.47)SPP - 5.93((0.58)SA +
2.59((0.74)SB -9.74((1.17) (8)
with n ) 24, r ) 0.951, and SD ) 0.73, which reproduces
the sensitivity of the decarboxylation rate to the pure
solvents studied by Kemp and Paul more than acceptably.
In addition, it clearly shows that the decarboxylation rate
increases dramatically with increase in the solvent
polarity and also, to a lesser extent, with its basicity; by
contrast, the rate decreases considerably with increasing
solvent acidity. This behavior is quite consistent with the
solvent effect scheme proposed at the beginning of this
section.
In addition, this fitting provides highly interesting
information. In fact, its independent term, log k0, which
would correspond to the logarithmic decarboxylation rate
in the gas phase (SPP ) 0, SA ) 0, and SB ) 0), is -9.74,
so the rate would be k ) 1.82 × 10-10 s-1. At this point,
it is interesting to examine the reason the decarboxyla-
tion rate in water reported by Kemp and Paul10 was
about 103 times greater (k ) 7.4 × 10-6 s-1).
(44) Dewar, M. J . K.; Thiel, W. J . Am. Chem. Soc. 1977, 99, 4899.

3414 J . Org. Chem., Vol. 65, No. 11, 2000
Catala´n et al.
Ta ble 2. SP P , SB, a n d SA Va lu es for Dim eth oxym eth a n e
a n d Bin a r y Mixtu r es of DMSO w ith Diglym e,
Aceton itr ile, Ben zen e, Dich lor om eth a n e, Ch lor ofor m ,
a n d Meth a n ol
solvent
[DMSO] (M) SPP
SB
SA
dimethoxymethane
DMSO/diglyme
0.648 0.359
0
1.1
7.3
9.0
10.5
1.1
2.8
7.3
10.7
0.11
0.28
1.1
2.2
3.9
5.6
9.0
0.56
1.1
2.8
3.9
5.6
8.4
0.56
1.6
3.4
5.6
8.4
0.873 0.616 0.008
0.936 0.642 0.043
0.952 0.634 0.056
0.970 0.639 0.057
0.915 0.404 0.052
0.923 0.530 0.060
0.938 0.606 0.070
0.957 0.628 0.072
DMSO/acetonitrile
DMSO/benzene
0.694 0.202
0.710 0.553
0
0
0.790 0.637 0.005
0.861 0.629 0.020
0.901 0.637 0.034
0.911 0.638 0.044
0.953 0.645 0.056
0.882 0.261 0.033
0.881 0.399 0.027
0.901 0.577 0.041
0.906 0.635 0.051
0.919 0.667 0.063
0.942 0.670 0.076
0.837 0.230 0.028
0.869 0.459 0.041
0.892 0.620 0.067
0.904 0.665 0.093
0.919 0.665 0.107
0.942 0.643 0.090
0.900 0.543 0.489a
0.909 0.546 0.467a
0.934 0.581 0.425a
0.935 0.599 0.416a
0.959 0.611 0.299a
0.953 0.608 0.285b
0.952 0.617 0.279b
0.969 0.625 0.267b
0.976 0.618 0.260b
DMSO/dichloromethane
DMSO/chloroform
DMSO/methanol
F igu r e 3. Plot of the independent term (log k0) of different
fits versus the σp parameters.
decrease in the reaction rate owing to the much less polar
and basic character of the latter solvent, the decrease
being especially prominent when the mixture contains
chloroform almost exclusively. Worth special emphasis
in this respect is the mixture with methanol (SPP )
0.857, SB ) 0.545, SA ) 0.605). Because this solvent is
much less basic and polar and much more acidic than
DMSO, raising its content in the mixture causes a
sustained decrease in the decarboxylation rate. The fact
that the rate apparently varies linearly with the DMSO
concentration is, to our minds, coincidental.
11.5
2.5
4.5
6.6
7.7
8.7
9.4
10.2
11.3
11.9
a
This peculiar variation of the decarboxylation rate with
the mixture composition and the strong dependence of
this behavior on the nature of the solvent mixed with
the DMSO (note that gradually changing from DMSO to
acetonitrile decreases the rate by a factor of only 3,
whereas changing to methanol reduces it 40 000 times)
prompted us to determine SPP, SB, and SA for the
mixtures studied by Kemp and Paul10 (see Table 2). It is
very interesting and significant that, adding these data
for the mixtures to those for the 24 above-mentioned pure
solvents provides the following global fitting:
Obtained from the wavenumbers of the UV-vis absorption
maxima for the probe DETZ.41 Obtained from the wavenumbers
b
of the UV-vis absorption maxima for the TBSB/DTBSB couple.40
(d) The reference medium to be used to compare
decarboxylation rates for this system is the gas phase,
the only one where the highly influential effects of
polarity, basicity, and acidity are completely absent.
Following the behavior described by the eq 9, it can be
predicted that, whereas the trifluoroethanol addition to
water does not change the constant rate for decarboxy-
lation (i.e., for a mixture of 0.5 molar fraction log k results
in -0.53), an equivalent amount of DMSO added to water
changes greatly the decarboxylation rate constant (i.e.,
for a 0.5 molar fraction log k is equal to +0.67).
log k )10.03((1.05)SPP - 5.73((0.40)SA +
2.41((0.49)SB - 9.58((0.83) (9)
Use of th e Deca r boxyla tion P r ocess a s a n En vi-
r on m en ta l P r obe. According to Smid et al.,24 poly(4-
vinylbenzo-18-crown-6) catalyzes the decarboxylation of
6-nitrobenzisoxazole-3-carboxylate in water. The process
involves the formation of a complex between the polymer
and the anion. In this situation, the decarboxylation of
6-nitrobenzisoxazole takes place at a rate similar to that
observed in benzene (i.e., 9.3 × 107 times greater than
in the gas phase).
In the presence of cation-binding crown polymers, the
rate of the process increases dramatically (by a factor up
to 5.7 × 108 relative to the gas phase). The effect can be
ascribed to the formation of tight ion pairs within the
polymer core, where the distance between pairs increases,
as does the activity of carboxylate ion, the active species,
with n ) 60, r ) 0.933, and SD ) 0.60, very similar to
that previously established for the pure solvents. Equa-
tion 9 warrants some interesting comments (see Figure
4):
(a) This is the first global explanation for the experi-
mental data of Kemp and Paul10 including all their pure
solvents and mixtures.
(b) Our probes provide accurate descriptions for the
behavior of the binary mixtures of DMSO with diglyme,
acetonitrile, benzene, dichloromethane, chloroform, and
methanol.
(c) The fact that our scales allow solvent mixtures to
be examined as if they were pure solvents is very
interesting and appealing.

Effects of Medium on Decarboxylation Kinetics
J . Org. Chem., Vol. 65, No. 11, 2000 3415
The use of this kinetics to derive information about the
environment of the process and the factors that influence
it have been discussed by several authors.25,32,35
In any case, from eq 9 it follows that the rate constant
depends qualitatively and quantitatively differently on
SA, SPP, and SB, which hinders its use as an environ-
mental probe. Thus, although the most influential pa-
rameter is SPP, its effect is appreciably modulated by
the basicity.
Acidity has the opposite effect, consistent with its
sequestering action on carboxylate ion (the active spe-
cies).10,24,35 In summary, the interpretation of k values,
the sole experimental data for the reaction, rests on three
parameters that vary with the particular environment.
A deeper, more comprehensive description of the decar-
boxylation process requires the knowledge of the presence
and sign of charges in the biochemical environment that
triggers the reaction.
However, the fact that eq 9 applies to both pure
solvents and solvent mixtures could help to explain the
catalytic mechanism for a reaction and/or boosting the
efficiency of existing biochemical sites.
F igu r e 4. Plot of log k values for the decarboxylation rate of
the tetramethylguanidinium salt of 3-carboxy-6-nitrobenzisox-
azole versus the values predicted by eq 9 corresponding to 24
pure solvents and 36 binary mixtures of DMSO.
Ack n ow led gm en t. The authors are grateful to
Spain’s DGICYT for funding this research within the
framework of Project PB93-0280 and PB98-0063. C.
D´ıaz also wishes to thank Comunidad de Madrid for
additional support in the form of a postdoctoral grant.
as a result. The net effect is that the decarboxylation rate
is much greater than that for the tetramethylguani-
dinium salts, which form ion pairs, but similar to the
values found in nonprotic dipolar solvents.
J O9918798