Prediction of Xanthine Solubilities Using Statistical Techniques

Mixture response-surface methodology can be used as a technique to predict solubility in mixed solvent systems. The present report shows that if the intent is to predict solubility in nonideal solutions, mixture response-surface methodology is a better technique than one which assumes a particular mechanism to hold true. This is demonstrated by comparing the predictive ability of the mixture response-surface model with that of an extended Hildebrand approach to nonideal solutions. The nonideal systems are those used by Martin and co-workers involving the solubility of theobromine, caffeine, and theophylline in dioxane–water mixtures.