C. A. BUNTON, N. D. GILLITT AND A. KUMAR
227
The corresponding relationships for reactions of OHϪ
with (PhO)2POOAr, Ph2POOAr and Ph2POSPh are
kinetic solvent effects. The dielectric constant of t-BuOH
(12·47) is much lower than those of MeCN (35·94) and
NMP (32·2), but the corresponding ENT values are 0·389,
0·460 and 0·355.3 Based on solvatochromic scales of pure
solvents, t-BuOH is a better hydrogen bond donor and
acceptor than MeCN, but is less polarizable.4
log(k2/k02)=0·25 log ␥s =Ϫ0·33 log(␥OH/␥s)
and for reaction with Ph2POSEt:
log(k2/k02)=0·35 log ␥s =Ϫ0·54 log(␥OH/␥s)
(11)
(12)
For reactions with OHϪ values of k2/k02 for given mole
fractions of t-BuOH or MeCN are similar, although rate
minima appear at lower mole fractions of t-BuOH than
MeCN. The behavior of NMP is different, because the
values of k2/k02 increase sharply with NMP >0·1, especially
with the p-nitrophenoxy derivatives (Table 7).
For most of these reactions the relationships between rate
constants and relative activity coefficients of the anionic
nucleophiles and the transition states are similar, despite
marked differences in the structures of the nucleophiles. The
coefficients of log ␥s and log(␥Nuc/␥ϯ ) are numerically
similar, but of opposite sign, which means that addition of
MeCN inhibits reaction by lowering the free energies of the
esters and accelerates it by increasing ␥Nuc/␥ϯ . This partial
compensation is not very sensitive to the structures of the
esters or the nucleophiles, despite marked differences in
their solvation requirements. However, a decrease in the
water content of the solvent always stabilizes the bulky
anionic transition state relative to the anionic nucleophile,
regardless of the structure of the latter, as predicted by
qualitative and quantitative treatments.2, 3 The failure of
these simple treatments to fit the kinetic solvent effects is
due to their neglect of medium effects on the free energies
of the esters.
In reactions of BDO, MeCN and t-BuOH affect k2/k20
similarly, but the rates increase much more sharply in the
drier solvents than for reactions of OHϪ (Tables 5 and 8).
As for reactions with OHϪ , the rates increase very sharply
with increasing amount of NMP (Table 7). The ability of
NMP to increase nucleophilicity is understandable because
it, like DMF and DMSO, should be a strong hydrogen bond
acceptor,3, 4, 7, 29 and reduce the ability of H2O to deactivate
nucleophilic anions. The only limitation in the use of NMP
as a kinetic solvent is that we had to use freshly distilled
material in obtaining consistent kinetic data (Experi-
mental).
Values of k2/k02 in the drier solvents depend on the
nucleophile, the leaving group and the organic solvent. For
given reactants these values are always larger in H2O–NMP
and H2O–MeCN than in H2O–t-BuOH, reflecting the ability
of t-BuOH to hydrogen bond, albeit weakly, to anions.4 The
consequent deactivation should be more important for
reactions of OHϪ than of the oximate ions, owing to
differences in hydrogen bond acceptance which make it
more difficult to desolvate OHϪ than an oximate ion.
Hydrogen bonding to the leaving oxide or thiolate ion
(electrophilic assistance) is probably not very important,
because although arene thiols are much more acidic than the
corresponding phenols, this difference is not reflected in the
reactivities of the phosphorus(V) esters. Electron-with-
drawing substituents increase the acidities of the phenols or
thiols much more than reactivities of the esters.16–18
Terrier and co-workers have observed non-linear Brøn-
sted plots in deacylations by oximate ions11 and in reactions
of aryloxide ions with bis(4-nitrophenyl)phenylphosphonate
in mixed solvents.28 They explained these results in terms of
a required partial desolvation of the nucleophile, based on
modified Brønsted plots which take medium effects into
account.27 The solvation term, as in our experiments,
depends on the nucleophile and the organic cosolvent.
In the drier solvents desolvation of the anionic nucleo-
philes gives sharp increases in k2, depending on the effect of
the organic solvent on the ability of H2O to hydrogen bond
to the anion. In these solvents we expect that ␥s and ␥ϯ will
become similar because the transition state is a bulky, low-
charge density, anion and solvation by organic solvents
should not be very sensitive to charge.
The similar behaviors of OHϪ and oximate ions of
different structures may arise because the charge in the
latter is largely on oxygen and strongly charge-delocalized
oxyanions may behave differently. The rate minima are less
evident for reactions with oximate ions than with OHϪ ,
except for reactions with Ph(Et)POOAr (Tables 3–5 and
Ref. 1). These qualitative observations accord with the
slopes, Ϫa, of plots of log(␥Nuc/␥ϯ ) against log ␥s, tending
toward unity (Nuc=OHϪ or oximate) as the ester becomes
more and the nucleophile less hydrophilic. There is then
more compensation between the solvation requirements of
the reactants and the transition state until in the drier
solvents there is dominant desolvation and destabilization of
the anionic nucleophile (increase in ␥Nuc). The rates then
increase very sharply and deviations from predictions based
on the qualitative solvent rules decrease.2, 3
ACKNOWLEDGEMENT
Kinetic effects of organic solvents
Support by the US Army Research Office is gratefully
acknowledged.
Second-order rate constants for reactions in aqueous t-
BuOH and NMP are given in Tables
4 and 7–9.
Qualitatively these solvents and MeCN behave similarly,
but there are differences in the extents of rate minima and
increases in the drier solvents.
REFERENCES
1. C. A. Bunton, N. D. Gillitt and A. Kumar, J. Phys. Org. Chem.
9, 145 (1996).
A number of physical properties are used as indicators of
© 1997 by John Wiley & Sons, Ltd.
JOURNAL OF PHYSICAL ORGANIC CHEMISTRY, VOL. 10, 221–228 (1997)