No abstract

Solubility is a fundamental physico-chemical property with important applications to biological, chemical, pharmaceutical and environmental industries. Reliable solubility data require careful experimental measurements that are tedious, time consuming and costly, and this research area needs further investigation.1) Solubility data for drugs in solvent mixtures at several temperatures are scarce, in particular in solvent mixtures.2-7) These data allow to obtain thermodynamic magnitudes that provide a better understanding of the co-solvent action and solute-solvent interactions. [2][3][4][5][6]8) Aqueous mixtures are often used to increase the solubility of drugs. Non-aqueous mixtures have application in synthesis, re-crystallization and purification of drugs and in some techniques of microencapsulation. 9,10) The solubility data also serve to construct mathematical models that help to optimize solvent composition selection in pharmaceutical technology. Many drug candidates are so hydrophobic that its bioavailability depends on techniques to enhance the aqueous solubility.11) During the earlier stages of drug development the amount available of drug is often restrictive. Therefore it is desirable to have a model for correlating and predicting drug solubility from a small amount of data being at the same time easy to apply. Solubility prediction is very difficult because it depends on a large number of factors such as chemical structure, solute-solvent interactions and temperature. The drugs usually form non-ideal solutions with the solvents of pharmaceutical interest, and semi-empirical or empirical approaches are often needed to obtain useful solubility prediction. Solubility modelling may significantly reduce the experimental work in solvent selection.Predictive methods often require a number of physicochemical parameters and relatively complicate computation, and low prediction capability has been reported for the universal functional group activity coefficient (UNIFAC) method in some compounds.1) The log-linear model of Yalkowsky is not applicable for drugs showing a solubility maximum, and a model was developed to account for deviations from linearity in ethanol-water mixtures.12) In fact, when solubility behaviour of drugs is studied within a wide polarity range, curves with one or two solubility peaks may be found. 6)Mota et al. 11) used the Non-random Two Liquid Segment Activity Coefficient (NRTL-SAC) model for predicting the solubility in pure solvents and solvent mixtures at 25-40°C. The model requires the melting data from the solid phase, T f , DH f and DC p . Limited experimental DC p values are available for drugs and this contribution is frequently ignored in solubility calculations.11) The predictions did not agree well with the experimental data of paracetamol at high ethanol-in water ratios, a fact that was attributed to the appearance of a peak in the solubility profile, as was earlier reported. 13,14) Bustamante and co-workers proposed a model to predict the solubility mole fraction ln X 2 of ...