Fractionation of orange essential oil using liquid–liquid extraction: Equilibrium data for model and real systems at 298.2 K.

• Fractionation of crude orange essential oil using hydroalcoholic solvents.
• Liquid–liquid equilibrium data of model and real systems were determined.
• High deviations for phase compositions calculation using parameters from literature.
• New NRTL and UNIQUAC parameters were adjusted to model citrus systems.
• The distribution of essential oil compounds depends on solvent hydration level.

Essential oils have many industrial applications and are basically composed of terpene hydrocarbons and oxygenated compounds. To improve their stability, sensorial quality, flavoring capacity and, consequently, their commercial value, the removal of hydrocarbon compounds is achieved by a fractionation process that can be performed by liquid–liquid extraction. In this work, a phase equilibrium study of real systems composed of crude orange essential oil and hydroalcoholic solvents (ethanol and water mixtures) was conducted and UNIQUAC and NRTL binary interaction parameters available in the literature were tested without success for the description of phase compositions. To obtain reliable parameters that provide a good description of the phase composition and can be used for computational simulation, liquid–liquid equilibrium data for model systems composed of limonene, linalool, ethanol and water at 298.2 K were determined and used to adjust new binary interaction parameters of UNIQUAC and NRTL thermodynamic models. Concerning the physical–chemical behavior of the system, the increase in the water content of solvent was observed to promote a decrease in the migration of oxygenated compounds to the solvent-rich phase and in the mutual solubility of the compounds. The increase in the composition of the oxygenated compounds in the system resulted in an increased solubility of these oxygenated compounds and a consequent decreased solvent selectivity. The new set of parameters provided a satisfactory representation of the experimental data, with global deviations of up to a 0.0031 mass fraction in the prediction process, which indicates that these parameters may be very useful in the optimization of the fractionation process by computational simulation.