Experimental determination of solubility and ternary phase diagram of chiral compound are of tedious and time consuming tasks, and in many cases, there is not enough experimental data for different enantiomeric compositions to access the experimental ternary phase diagram. Using thermodynamic models with predictive capability, having less dependency on experimental data, affords a great advantage for ternary phase diagram modeling of these compounds. In this study, NRTL-SAC model, as one of the known and famous predictive models in the solubility modeling of pharmaceutical compounds, has been applied to ternary phase diagram modeling of chiral medetomidine salts in alcohols. NRTL-SAC molecular parameters of each chiral compound were calculated just by using racemic point solubility data before being used for the solubility prediction of other enantiomeric compositions. The results of thermodynamic modeling and the predictive data are compared with those of the experimental data. Root mean square error (RMSE) is also calculated for each compound. It was observed that, there is a small deviation between predictive and experimental ternary phase diagrams. The eutectic points of chiral compounds in both racemic and conglomerate forms are also well predicted through this method.

The NRTL-NRF (non-random two liquid-non-random factor) model proposed by Haghtalab mid Vera is used to represent thr excess Gibbs free energy of the binary electrolyte systems. The model is successfully applied for the prediction of the mean ionic activity coefficient for the whole range of electrolyte concentrations, from dilute solutions up to saturation. The model is generalized to represent the excess Gibbs free energy of aqueous multielectrolyte systems using only the binary interaction energy parameters. It is shown, furthermore, that the mean activity coefficients of the ternary aqueous electrolyte systems can be predicted fairly well without additional parameters. In comparison with other models, better results are obtained and the experimental values are reproduced from the dilute region up to saturation.

The non-electrolyte NRTL-NRF model has been modified to study electrolyte solutions. The modified model for electrolytes is composed of short-range parts expressed by the modified nonelectrolyte NRTL-NRF and the Pitzer-Debye-Hü ckel equation to represent the long-range interactions of ions in the solution. In this work, a salt-specific parameter is used. Various types of experimental data including binary and ternary activity and osmotic coefficients, solid and gas solubilities in aqueous NaCl, and also aqueous Methyldiethanolamine (MDEA) data at wide temperature and pressure ranges have been implemented to check the performance of the present model. The overall relative standard deviation of 0. 046 has been achieved for 130 strong aqueous binary electrolytes by the new model in fitting the experimental data of activity coefficients. The percent of absolute average deviations of the modified model for CO2+ MDEA+ H2O and H2S+ MDEA+ H2O are 30. 3% and 24. 8%, respectively. The results show the good capabilities of the model for electrolyte solutions.

In this study, NRTL and UNIQUAC thermodynamic models were used to predict the composition of ternary mixtures of solvents+m/o/p-cresol+water in organic and aqueous phases. Various solvents are used for the separation of cresols from water. In this study, methyl propyl ketone, methyl isopropyl ketone, methyl butyl ketone, and methyl isobutyl ketone solvents were investigated. Intermolecular interaction parameters were considered to be a function of temperature. The binary interaction parameters of water and cresols (m, o, or p) in the presence of each type of solvent were considered to be the same. Also, regardless of the type of cresol (m, o, or p), the parameters of binary interaction between water and each solvent were considered to be the same. The results proved the accuracy of the presented models, though the parameters of binary interaction parameters were considered to be the same. The root mean square deviation for NRTL and UNIQUAC models was 0.0086 and 0.0089, respectively.