Handbook of Solvents - George Wypych - ChemTech - Ventech!

14.21.2 Predicting cosolvency 1007
Cosolvent N a b R 2
log Kow range Ref.
Propylene glycol 47 0.03 0.89 0.99 -5 ~ 7 61
Propylene glycol 8 0.77 0.62 0.96 n.a. 24
PEG400 10 0.68±0.43 0.88±0.16 0.79 -0.10 ~ 4.18 80
Butylamine 4 1.86±0.30 0.64±0.10 0.96 -1.69 ~ 4.49 80
*estimated from Figure 3 in Reference 24.
n.a. = not available.
This empirical approach using equations [14.21.2.8], [14.21.2.5], and [14.21.2.2] can
produce acceptable estimates of log (Sm/Sw) only if the solubilization exhibits a roughly
log-linear pattern, such as in some HOC/water/methanol systems. In addition, it is important
to limit the use of equations [14.21.2.5] and [14.21.2.8] within the ranges of log Kow used in obtaining the corresponding parameters.
14.21.2.6 Predicting cosolvency in non-ideal liquid mixtures
Deviations from the log-linear model
Most solubilization curves, as shown in Figure 14.21.2.1, exhibit significant curvatures
which are not accounted for by the log-linear model. A closer look at the solubilization
curves in Figure 14.21.2.1 reveals that the deviation can be concave, sigmoidal, or convex.
In many cases, especially with amphiprotic cosolvents, a negative deviation from the
end-to-end log-linear line is often observed at low cosolvent concentrations, followed by a
more significant positive deviation as cosolvent fraction increases.
The extent of the deviation from the log-linear pattern, or the excess solubility, is measured
by the difference between the measured and the log-linearly predicted log Sm values:
i ( Sm Sm) Sm ( Sw ifi) log / = log − log +∑σ [14.21.2.9]
The values of log (S m/S i m) for naphthalene, benzocaine, and benzoic acid in selected
binary solvent mixtures are presented in Figures 14.21.2.2-a, -b, and -c, respectively.
The log-linear model is based on the presumed ideality of the mixtures of water and
cosolvent. The log-linear relationship between log (S m/S w) and f is exact only if the
cosolvent is identical to water, which cannot be the case in reality. Deviation is fortified as
any degradation, solvation, dissociation, or solvent mediated polymorphic transitions of the
solute occur. The problem is further compounded if the solute dissolves in an amount large
enough to exert significant influence on the activity of solvent components. Due to the complexity
of the problem, efforts to quantitatively describe the deviations have achieved only
limited success.
A generally accepted viewpoint is that the deviation from the log-linear solubilization
is mainly caused by the non-ideality of the solvent mixture. This is supported by the similarities
in the patterns of observed log S m and activities of the cosolvent in solvent mixture,
when they are graphically presented as functions of f. Based on the supposition that solvent
non-ideality is the primary cause for the deviation, Rubino and Yalkowsky 87 examined the
correlations between the extent of deviation and various physical properties of solvent mix-