Correlation of the solubilities of alkali chlorides in mixed solvents: Polyethylene glycol þ H 2 O and Ethanol þ H 2 O Jorge A. Lovera a , Aldo P. Padilla a , He ´ ctor R. Galleguillos a,b,n a Department of Chemical and Mineral Process Engineering, University of Antofagasta, Av. Angamos 601, Antofagasta, Chile b Centro de Investigacio ´n Cientı ´fico y Tecnolo ´gico para la Minerı ´a (CICITEM), Av. Jose´ Miguel Carrera 1701, 41 piso, Antofagasta, Chile article info Article history: Received 18 November 2011 Received in revised form 24 March 2012 Accepted 24 March 2012 Available online 8 May 2012 Keywords: Experimental solubility Solubility correlation Modified Pitzer model abstract Solubility data for the LiCl þPEG 4000 þH 2 O system at 25 1C were obtained. These data, along with other data published in the literature for the NaCl þPEG 4000 þH 2 O, KCl þPEG 4000 þH 2 O, LiCl þ C 2 H 5 OHþH 2 O, NaCl þC 2 H 5 OHþH 2 O, and KCl þC 2 H 5 OH þH 2 O ternary systems at 25 1C, were correlated using a modified Pitzer model. The values of the solubilities calculated using the model are in good agreement with the experimental observations. Using the model parameterization established in this study, the yield of precipitated salt was calculated as a function of kg PEG 4000 or C 2 H 5 OH/100 kg of saturated solution. & 2012 Elsevier Ltd. All rights reserved. 1. Introduction One of the methods employed in crystallization to produce supersaturated solutions is termed drowning out. This separation method consists of adding an additional component that is miscible with the original solution to decrease the solubility of the salt of interest. This crystallization technique has a number of advantages compared with traditional evaporation or cooling procedures, including increased yields, operation at ambient temperature, higher purity of crystals, selectivity and others [1]. The prediction or correlation of salt solubility data in mixed solvents is an important tool for the design and simulation of the drowning-out crystallization process. To carry out this type of analysis, several models have been proposed in the literature based on the thermodynamics of the electrolytes. Farelo et al. [2] used the Pitzer mole fraction-based thermodynamic model to represent the phase equilibrium of the NaCl þ KCl þ water þ ethanol system from (298 to 323) K with up to 20 mass% ethanol in the solvent. The results obtained are very good—the model reproduced the experimental solubility data for both salts with a standard deviation of 70.023 mol/kg solvent. Zeng et al. [3] employed the extended BET model to predict the solubility of magnesium chloride in the HCl þ LiCl þ MgCl 2 þ H 2 O system, using HCl as a salting-out agent. The solubilities predicted at (273, 293, and 313) K are in good agreement with the experimental values, with standard deviations lower than 0.092 mol kg 1 . Marcilla et al. [4] and Reyes et al. [5] presented a method to simultaneously correlate the equilibrium data for all of the equilibrium regions present in ternary systems com- posed of water þ organic solvent þ salt. The NRTL model was used by the authors to formulate the liquid-phase activity. In general, the results are good when the salt crystallizes anhydrously, as is the case of sodium chloride at 298 K and potassium benzylpenicillin at 293 K [4]. However, the deviations obtained between the experi- mental and calculated data are higher for lithium chloride at 298 K. Because lithium chloride crystallizes in both monohydrate and anhydrous forms, there is a greater number of equilibrium zones, complicating the simultaneous correlation of the equilibrium data in all existing regions. The results improved when the Gibbs energy for monohydrated salt was considered as a fitting parameter [5]. Jimenez et al. [6] used the method of Kan et al. [7] to represent the solid–liquid equilibrium of potassium sulfate in different water þ or- ganic solvent mixtures at (288, 308 and 318) K. Because the activity coefficient of an electrolyte in a water þ organic solvent þ salt system depends simultaneously on the concentration and content of the organic co-solvent, Kan et al. [7] defined an overall activity coefficient to quantify these effects separately. The Pitzer model of virial coefficients was used to quantify the effect of the salt, whereas a similar equation from the Born model was used to quantify the co- solvent effect. In general, the results show a good agreement between experimental and calculated values; for example, a mean absolute deviation of 0.0063 mol  kg 1 is obtained for the potassium sulfateþ water þ 1-propanol system. The Pitzer model of virial coefficients has been used widely in the prediction and correlation of solubilities in systems that contain one salt (or more) in a single solvent, generally water Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/calphad CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 0364-5916/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.calphad.2012.03.002 n Corresponding author at: Department of Chemical Engineering, University of Antofagasta, Avenida Angamos 601, Antofagasta, Chile. Tel.: þ56 55 637344. E-mail address: hgalleguillos@uantof.cl (H.R. Galleguillos). CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry 38 (2012) 35–42