Thermodynamic Properties of 1:1 Salt Aqueous Solutions with the Electrolattice Equation of State

— Thermodynamic Properties of 1:1 Salt Aqueous Solutions with the Electrolattice Equation of State — The electrolattice Equation of State (EOS) is a model that extends the Mattedi-Tavares-Castier EOS (MTC EOS) to systems with electrolytes. This model considers the effect of three terms. The ﬁrst one is based on a lattice-hole model that considers local composition effects derived in the context of the generalized Van der Waals theory: the MTC EOS was chosen for this term. The second and the third terms are the Born and the MSA contributions, which take into account ion charging and discharging and long-range ionic interactions, respectively. Depending only on two energy interaction parameters, the model represents satisfactorily the vapor pressure and the mean ionic activity coefﬁcient data of single aqueous solutions containing LiCl, LiBr, LiI, NaCl, NaBr, NaI, KCl, KBr, KI, CsCl, CsBr, CsI, or RbCl. Two methods are presented and contrasted: the salt-speciﬁc and the ion-speciﬁc approaches. Therefore, the aim of this work is to calculate thermodynamic properties that are extensively used to design, operate and optimize many industrial processes, including water desalination.


INTRODUCTION
Electrolyte solutions are encountered in many chemical industrial processes such as water desalination, distillation, extraction, solution crystallization, mineral scale formation in steam systems, gas scrubbing, hydrometallurgy and biotechnology according to Pitzer (1973), Loehe and Donohue (1997) and Myers et al. (2002).So, thermodynamic models for electrolyte systems have been developed for evaluating the properties needed to design and simulate processes involving electrolytes.
Compared to non-electrolyte solutions, electrolyte solutions are more difficult to model because of the presence of dissolved ions.Myers et al. (2002) explain these ions can interact with the solvent and with one another by long-range or short-range electrostatic interactions.
Both interactions are always present in electrolyte solutions.However, at low electrolyte concentrations, the long-range forces are dominant whereas, at high concentrations, the short-range forces are dominant.Therefore, an accurate model applicable to electrolyte solutions must consider all these interactions.Several reviews discuss in depth models for electrolyte solutions such as the publications of Friedman (1981), Loehe and Donohue (1997) and Anderko et al. (2002).Generally, these models can be classified in two types of engineering-oriented approaches: excess Gibbs energy (g E ) or Helmholtz energy.Some examples based on g E expressions are the models of: - Debye and Hu¨ckel (1923), who are pioneers in developing models to describe the behavior of electrolyte solutions.Their model considers the ions as charged hard spheres present in a continuum dielectric medium.The ions has fixed diameters values, and the concentration of the solutions must be lower than 0.1 molal (Guggenheim and Turgeon, 1955); - Pitzer (1973), which can predict the behavior of single solvent solutions; - Chen and Evans (1986) two different equations to account for the cationhydrated diameter, and correlated and predicted mean ionic activity coefficients insingle and mixed electrolyte solutions.Haghtalab and Mazloumi (2009) assert g E models are easily applicable to engineering design and also have good accuracy for different electrolyte solutions.However, such models are not suitable for calculating solution densities and activity coefficients as functions of pressure.Models for electrolyte solutions that express the Helmholtz energy as function of temperature, volume and species amounts do not suffer such shortcomings and therefore, provide a more complete description of the systems modeled.
In the case of models based on Helmholtz energy, Wu and Prausnitz (1998) extended the Peng-Robinson EOS using different terms: the SAFT term to account hydrogen bonding and a term to account the electric charging of ions and subsequent ion-ion and ion-solvent interactions.Liu et al. (1999) proposed an equation that includes 6 terms: interactions of ion-ion, ion-dipole, dipole-dipole, Lenard-Jones dispersion, hydrogen bond association accounted by the MSA equation and hard sphere repulsion.They investigated 30 aqueous electrolyte solutions, correlating mean ionic activity coefficient using only one adjustable parameter for each salt.Density and mean ionic activity coefficient in mixtures of salts were also predicted.Myers et al. (2002) determined mean ionic activity coefficients and osmotic coefficients at 298.15 K for 138 aqueous salt solutions by a three adjustable parameter equation containing a term based on the Peng-Robinson EOS, a Born term and a MSA term.Based on the perturbed-chain statistical associated fluid theory, Cameretti et al. (2005) used the equation named ePC-SAFT, which joins the PC-SAFT and the Dubye-Hu¨ckel term, to correlate the vapor pressure and the density of 12 salt solutions, considering two adjustable parameters for each ion.Held et al. (2008) extended the work done by Cameretti et al. correlating densities and mean ionic activity coefficients at 298.15 K and predicting vapor pressures.Inchekel et al. (2008) proposed an extension of the Cubic Plus Association (CPA) equation along with Born and MSA terms, in order to calculate the apparent molar volume, mean ionic activity coefficient and osmotic coefficient for 10 aqueous electrolyte solutions considering three ion-specific parameters.Held et al. (2008) studied the behavior of systems containing weak electrolytes, correlating densities and mean ionic activity coefficients, using ion-pairing parameters.Lee and Kim (2009) used the equation PC-SAFT along with the primitive MSA term and adjusted 4 ion-specific parameters for 26 aqueous electrolyte solutions, correlating density and mean ionic activity coefficient data at 298.15 K and 1 bar.
The aim of this work is to present an EOS for electrolyte solutions, applicable to calculations of phase equilibrium and thermodynamic properties.This equation uses the Helmholtz approach, including three terms: one related to the Mattedi-Tavares-Castier EOS (MTC EOS) of Mattedi et al. (1998), based on the hole-lattice theory and whose partition function is obtained from the generalized van der Waals theory, along with the Born and MSA terms.Due to the combination of the hole-lattice theory with long-range interactions typical of electrolyte solutions, it is named electrolattice EOS.This model is used to calculate vapor pressures, mean ionic activity coefficients, densities and osmotic coefficients of 13 aqueous solutions of strong electrolytes containing salts of type 1:1 (LiCl, LiBr, LiI, NaCl, NaBr, NaI, KCl, KBr, KI, CsCl, CsBr, CsI and RbCl).Two approaches are presented: one considering ion-specific parameters, whose idea is to use parameters for the ions independently of the aqueous solutions they are inserted; the other considering salt-specific parameters, whose goal is to use ionic parameters dependent on the salt that constitutes the aqueous solution.Both kinds of parameters are analyzed in terms of the accuracy of the calculated thermodynamic properties compared to their experimental data.

ELECTROLATTICE EQUATION OF STATE
According to the Helmholtz energy (A) approach, it is possible to develop an EOS applicable to electrolytes following the suggestions of Myers et al. (2002).In general, they consider a path that starts from the reference state and passes through several intermediate states until reaching the final state that corresponds to the electrolyte solution.This procedure allows the formulation of a model using a path of constant temperature and volume.The change in Helmholtz energy between the final and initial states is equal to the residual Helmholtz energy computed by adding the Helmholtz energy changes between consecutive states along the path.The transition from each state to the next corresponds to adding another type of specific interaction to the system.These changes represent ion-solvent and solvent-solvent physical interactions of short range, association or solvation and long-range ion-ion interactions.Santos (2010) developed the electrolattice EOS following a four step thermodynamic path (Fig. 1), as follows: -step I.It is assumed that a reference mixture constituted by charged ions and molecules is in a hypothetical ideal gas state at temperature T and volume V.In the first step, the charges on all ions are removed.The change in Helmholtz energy is accounted by the Born equation for ions in a vacuum, DA Born disc ; -step II.The short-range attractive dispersion and repulsive forces due to excluded volume are turned on.Also, self-association of solvent molecules can occur.The MTC EOS is used to calculate the change in Helmholtz energy for this step, DA MTC ; -step III.The ions are recharged.The change in Helmholtz energy is accounted by the Born equation for ions in a dielectric solvent, DA Born chg ; -step IV.The long-range interactions among the ions in solution are taken into account using the Mean Spherical Approximation (MSA), and the corresponding change in the molar Helmholtz free energy is denoted by DA MSA .The residual Helmholtz energy for forming an electrolyte solution is given by: wherein Equation ( 1) can be arranged in order to obtain the residual Helmholtz energy in the following manner: In these equations, T is the temperature of the system, V is the volume, n is a vector of the number of moles of the species, A IGM is the Helmholtz energy of the ideal gas mixture that corresponds to the reference state and A R is the residual Helmholtz energy.The term A MTC is the contribution of the MTC EOS to the Helmholtz energy that includes the repulsive and attractive effects, association and the ideal gas contribution, whereas DA MTC is the contribution of the MTC EOS to the Helmholtz energy without ideal gas contribution.Expressions for the pressure and chemical potential can be obtained using standard techniques.According to Myers et al. (2002), it is possible to obtain the following expression for the fugacity coefficient: In the next subsections, the terms that form the electrolattice EOS are presented and discussed.

Short-Range Contribution (MTC EOS)
In the electrolattice EOS, the short-range interactions between species are accounted for by the MTC EOS, whose Helmholtz energy expression was developed by Mattedi et al. (1998).This EOS can be used in three forms: as a molecular model, as a conventional group-contribution model or, as done here, as a region-contribution model, by splitting each molecule in regions.Each molecular region interacts differently with the regions of its neighboring molecules (Santos et al., 2010).The expressions for the pressure and chemical potential derived from this expression for the Helmholtz energy are: In Equations ( 6) and ( 7), m a i is the number of regions of type a in a molecule of type i, ng is the total number of regions, nc is the number of components, Q a is the surface area of a region of type a, z is the coordination number (z = 10), W is a lattice constant (W = 1) and m* is volume of one mol of cells (m* = 5 9 10 À6 m 3 .molÀ1 ).
The number of external contacts (zq i ), the bulkiness factor (l i ) and the reduced volume (v˜) are: UNCHARGED IONS AND SOLVENT (IDEAL GAS)

UNCHARGED IONS AND SOLVENT MOLECULES IONS AND SOLVENT MOLECULES IONS AND SOLVENT MOLECULES
Path to the formation of an electrolyte solution at constant temperature and volume proposed by Myers et al. (2002).
The ions are assumed to be spherical and their bulkiness factors are calculated accordingly.We follow Abrams and Prausnitz (1975) in defining the volume (r) and surface area (q) parameters as ratios: wherein R i is the radius of a spherical ion i, V i is its volume and A i is its surface area.The corresponding symbols with an asterisk denote the same properties for a lattice cell, assumed to be spherical in Equations ( 13) and ( 14).It follows from these equations that, for a spherical ion: The bulkiness factor is obtained by substituting Equation (15) in Equation ( 9): For determining the volume parameter, the ratio between the ion volume and the cell volume is used as follows: in which r i is the diameter of the ion and N a is the Avogadro's number.Therefore, knowing the ion diameter, it is possible to calculate the volume parameter, the bulkiness factor and the surface area parameter by applying these three equations.Equations ( 13) to ( 17) are exclusively for calculating ion parameters, since the spherical shape is imposed.For the solvent molecule, no physical structure is assumed.Thus, no mathematical relation is proposed to connect the area and volume parameters for the solvent.
The average values of r and zq are: Other symbols present in previous equations are defined as follows: where u ma is the interaction energy between the m and the a regions, given as: The chemical potential of component i given by Equation ( 7) is the total contribution, including the ideal gas contribution, which is equal to: which is consistent with the MTC model.The difference between Equations ( 7) and ( 24) is the residual chemical potential of component i in the MTC model with only repulsive and attractive forces contributions included

Born Contribution
In the electrolattice EOS, the Born contribution term accounts for the variation in the Helmholtz energy for discharging an ion in vacuum (ideal gas) and charging it in a solvent.In it, the ions are considered as hard spheres immersed in a continuum with uniform dielectric constant.The Born contribution term to the pressure is: in which e is the unit of elementary charge, e 0 is the permittivity of vacuum, D is the relative dielectric constant of the solution and Z i is the number of charge of species i.
The chemical potential of component i due to the Born model is: in which: The Born term depends on the relative dielectric constant (D).We use the Pottel model for electrolyte solutions, which is presented by Zuo and Fu¨rst (1997) as: wherein: Here, D s is the pure solvent dielectric constant.To evaluate it for pure water, we used the following relation, reported by Zuo and Fu¨rst (1997), for temperatures in Kelvin: Many published models try to describe the long-range electrostatic contributions of charged species in solution.The MSA has been extensively used to describe such interactions, with different levels of approximation.Instead of solving implicit forms of the MSA iteratively, we opted for the explicit approximation with single effective ion diameter of Harvey et al. (1988), with diameter values for cations and anions obtained by Marcus (1988).This form of MSA was successfully used by Myers et al. (2002) along with the Peng-Robinson EOS for obtaining mean ionic activity coefficient of many salt solutions.The contribution of this MSA term to the pressure is: wherein, in which D is the dielectric constant presented in Equation (28).In these equations, j is the Debye screening parameter and U is the MSA shielding parameter.The contribution of this MSA term to the chemical potential is:

THERMODYNAMIC PROPERTIES
The relation between the fugacity coefficient and the activity coefficient for an ion of species i and for solvent s can be expressed, respectively, as: Þis the unsymmetrical activity coefficient of ion i on the mole fraction scale; u i (T, P, x) is the fugacity coefficient of ion I in the solution; u i (T, P, x s ? 1) is the fugacity coefficient of ion i at infinite dilution; u s (T, P, x) is the fugacity coefficient of solvent s in the solution and u s (T, P) is the fugacity coefficient of pure solvent s.
The activity coefficient of ion i on the molality scale is given by: wherein x s is the molar fraction of the solvent; v is the sum of stoichiometric coefficients of the ions in the salt molecules, in this work, v = 2 since the employed salts are 1:1; m is the molality; and M s is the molecular weight of the solvent in kg.mol À1 .Expressions for the mean ionic activity coefficient and for the osmotic coefficient are available in several references (e.g., Myers et al., 2002).

PARAMETER FITTING
Solutions containing aqueous single strong 1:1 electrolytes are formed by three species: solvent, cation (C) and anion (A).Since the MTC EOS term is present in the electrolattice EOS, according to Ehlker and Pfenning ( 2002) and Mattedi et al. (1998), it is convenient to split the solvent molecule in three regions: an electron-donor (a), an electron-acceptor (b) and a dispersion region (D).
Once the aim of this work is to determine thermodynamic properties of electrolyte solutions, the first step toward this goal is to obtain the parameters for pure water.As in the MTC EOS, the electrolattice EOS uses seven parameters to represent pure polar solvents.These parameters were obtained by simultaneous fit of liquid density data and pressure at saturation, obtained from the DIPPR correlation (American Institute of Chemical Engineers, 2007), in the temperature range from 292.84 to 489.67 K. To reduce the number of adjustable solvent parameters, all region-region interactions, except that between a and b regions, are assumed to be dispersion interactions, numerically equal (u 0 D-D /R).These interactions are assumed to be temperature dependent, with B a-a = B a-D = B b-b = B b-D = B D-D .Also for reducing the number of adjustable parameters, the interaction between a and b regions, which refers to the hydrogen bond, is considered to be temperature independent (B a-b = 0).Table 1 shows the parameters for water used in this paper.
When an electrolyte solution of a single salt and a solvent is modeled by the electrolattice EOS, there are five regions that interact with one another (three for the solvent (a, b, D), one for the cation (C) and one for the anion (A)).Each of these interactions is, in principle, supposed to follow Equation ( 23) and there would be 18 adjustable parameters in such a solution: 9 parameters (u 0 ma /R) and 9 parameters B ma .This is inconvenient in an engineering-oriented model and we adopted simplifications to reduce the number of adjustable parameters.We assumed the interactions between water regions and each charged species are equal, i.e.: These interactions are assumed to be temperature independent; therefore, the B ma term of Equation ( 23) is neglected as follows: Two assumptions are made about the interactions between ionic species.Following Zuo et al. (2000), the short-range interaction between ions with the same charge is neglected once the repulsion between the species is strong, therefore: Also, according to Lee and Kim (2009), it is possible to neglect the short-range interactions between ions of different charges in solution, i.e.: With these simplifications, each ion has only one adjustable parameter and yet, the electrolattice EOS gives good results.
To calculate the vapor pressure of electrolyte solutions, it is assumed the vapor phase only contains water.Also, all the salts are strong electrolytes, which fully dissociate in solution into cations and anions.Then, for each data point, there is only one phase equilibrium  equation, the isofugacity equation for water, which is solved to find the vapor pressure at a given temperature.In this work, the experimental data set for vapor pressure is composed by : Hubert et al. (1995), who obtained data for NaCl in a temperature range from 293.15 to 363.15 K; for LiCl, LiBr and LiI, the work of Patil et al. (1990) was used and, for other salts, the work of Patil et al. (1991) was used, both in the temperature range from 303.15 K to 343.15 K.For mean ionic activity coefficients and osmotic coefficients, the data were obtained from Robinson and Stokes (1949) and Lobo and Quaresma (1989) at atmospheric pressure and 298.15 K.
Considering the parameters obtained for the pure solvent and the set of simplifications for the ions, the number of adjustable parameters in a single-salt aqueous solution is reduced to two: ðu solventÀC 0 =RÞ and ðu solventÀA 0 =RÞ.Different approaches are used to determine these energy interactions: the first considers the parameters as salt-specific, in which, these two parameters are simultaneously obtained for each salt; the second approach considers the parameters as ion-specific, which means that the energy interaction between water and an ion must always be the same, independently of the salt it takes part.The ion-specific approach is more complex compared to the salt-specific approach since more parameters need to be fitted at the same time.Following Held et al. (2008), we first determined the parameters for a basic set of salts, composed by three cations (Na + , K + , Li + ) and three anions (Br À , Cl À , I À ), corresponding to nine single-salt aqueous solutions.In this initial fitting, 6 parameters were simultaneously obtained using 368 data points for vapor pressure and 208 data points for mean ionic activity coefficient.Then, the anion parameters obtained previously were fixed to determine the Cs + parameters, using experimental data of CsBr, CsCl and CsI aqueous solutions.Likewise, the Rb + parameters were determined by correlating the experimental data of RbCl.
For both approaches previously mentioned, the adjustable parameters were determined using the Excess Gibbs Energy Models and Equations of State (XSEOS) package, an Excel Ò add-in developed by Castier and Amer (2011), by minimizing the Objective Function (OF): in which P calc i and c calc AEi denote calculated values for vapor pressure and mean ionic activity coefficient, respectively; P exp i and c exp AEi are the corresponding experimental values; Np is the number of points.

RESULTS AND DISCUSSION
The volume, the bulkiness and the surface area parameters for ions are calculated using Equations (15-17).Marcus (1988) determined the ionic diameter of several cations and anions in aqueous solutions, which are used instead of Pauling-type crystal ionic diameters (Pauling, 1927).The diameter values used in our model are presented in Table 2.
The parameters fitted by the salt-specific and ionspecific approaches are shown in Tables 3 and 4,  respectively.In Table 5, the values for the Average Relative Deviation (ARD(%)), calculated by Equation ( 45), for vapor pressure, mean ionic activity coefficient and osmotic coefficient of each salt solution are presented.These outcomes are shown in Table 5 for both fitting approaches, as SSP for salt-specific parameters, and ISP for ion-specific parameters.
In Equation ( 45), val calc i and val exp i represent the calculated and experimental values of vapor pressure, mean ionic activity coefficient or osmotic coefficient.Table 5 shows the electrolattice EOS with salt-specific and ionspecific parameters can correlate vapor pressure satisfactorily, with average of ARD values equal to 1.74% and 1.84%, respectively.A comparison of our results with those obtained by Held et al. (2008) for the same salts shows that their average ARD for vapor pressure is 4.48%.Figures 2 and 3 present the vapor pressures of aqueous solutions containing KI and KBr, respectively, as function of molality in the temperature range from 303.15 K to 343.15 K, using ion-specific parameters.The results with the salt-specific parameters, not shown in these figures, are similar.With both fitting approaches, the results of the model are in good agreement with the experimental data.
For the mean ionic activity coefficient data (Tab.5), in general, the salt-specific parameters correlate the experimental data more adequately than the ion-specific parameters, except for RbCl.The average ARD for such approaches are 2.06 % and 3.71 %, respectively.The results obtained for LiBr and CsI are higher compared to other salts modeled by ion-specific parameters, and their ARD values are 6.92% and 10.88 %, respectively.In Figure 4, it is possible to observe that the mean ionic activity coefficient is underestimated for KCl and CsCl at 298.15 K and at atmospheric pressure, when ion-specific parameters are used.Also in Figure 4, it is possible to observe that for salts whose anion is Cl À , the larger is the diameter of the cation, the lower is the mean ionic activity coefficient in all the molality range.In Figure 5, salts containing Br À as the anion are presented.Also in this set of salts, the mean ionic activity coefficient decreases as the cation size increases, in the sequence NaBr > KBr > CsBr. Figure 5 also shows that the electrolattice EOS can match the experimental mean ionic activity coefficient satisfactorily with the parameters fitted by both approaches.
The osmotic coefficient data of aqueous solutions, using the ion-specific and the salt-specific methods, are presented in Figure 6.The average ARD is 1.36% for salt-specific approach and 2.23% for ion-specific approach.This thermodynamic property was not directly used for parameter fitting but the mean ionic activity coefficient is analytically related to the osmotic coefficient via the Gibbs-Duhem equation (a complete formulation about this relation can be found in Prausnitz et al., 1999).For aqueous solutions containing NaBr, KBr and RbCl, results using parameters fitted Mean ionic activity coefficient of NaCl (h), KCl (D) and CsCl (s) in water at 1 atm and 298.15 K, obtained from Lobo and Quaresma (1989).The lines represent the electrolattice model considering salt-specific parameters (ÀÁÀ) and ion-specific parameters (-).
with the salt-specific and ion-specific approaches are in good agreement with experimental data.For RbCl at high concentrations, the parameters adjusted by both approaches cause underestimation of the osmotic coefficients compared to the experimental data.
It is not possible to observe a trend in the salt-specific parameters of Table 3.Generally, all the parameters have negative values, except the ones obtained for the salts whose cation is Cs + , in which the energy interaction between the solvent and the anion is positive.Osmotic coefficient of NaBr (h), KBr (D) and RbCl (s) in water at 1 atm and 298.15 K, obtained from Lobo and Quaresma (1989).The lines represent the electrolattice model considering salt-specific parameters (ÀÁÀ) and ionspecific parameters (-).Mean ionic activity coefficient of NaBr (h), KBr (D) and CsBr (s) in water at 1 atm and 298.15 K, obtained from Lobo and Quaresma (1989).The lines represent the electrolattice model considering salt-specific parameters (ÀÁÀ) and ion-specific parameters (-). (1)Experimental data obtained from Lobo and Quaresma (1989).
property, the prediction is acceptable in the temperature range of each salt.Predictions of vapor pressure considering binary salt solutions were also analyzed.For solutions composed by NaCl + KCl, KBr + KCl and KI + KCl the experimental data were from Apelblat and Korin (2009), obtained at saturation.For NaBr + KCl and NaCl + KBr, the data were obtained from Hsu et al. (2003) at a molality range from 1 to 4 molal and temperature range from 303.15 K to 333.15 K. Table 8 sums up the results for such predictions and Figures 7-9 display the corresponding plots.For the three first solutions, the ARD values are slightly higher than for the other two since the molality is high throughout the temperature range and the model tends to be not so accurate at such condition.For the other two solutions, the predictions are very good in the temperature range from 303.15 to 333.15 K.It should be noted that no parameter fitting is needed when applying the electrolattice EOS to predict vapor pressures of aqueous solutions of multiple salts.

CONCLUSIONS
In this work, the electrolattice equation of state was used to model the thermodynamic properties of aqueous solutions of strong 1:1 electrolytes.The ion volume and surface area parameters used in this EOS were evaluated from ionic diameter data.Experimental data for vapor pressure and mean ionic activity coefficients were used Vapor pressure (Pa) Temperature (K) Vapor pressure prediction of saturated solutions of KCl + KBr (h), NaCl + KCl (D) and KCl + KI (s) in water, obtained from Apelblat and Korin (2009).The line represents the electrolattice model considering ion-specific parameters (-). to fit the water-ion parameters of the model by using the salt-specific and ion-specific approaches.The results for these two properties are generally in good agreement with experimental values.The results for osmotic coefficient, which was not directly used in the objective function, are also in good agreement with experimental data.Predictions of density, mean ionic activity coefficient and osmotic coefficient at temperatures other than 298.15K and vapor pressure of binary salt solutions, considering ion-specific parameters, were fairly accurate.Comparing all the thermodynamic properties, the salt-specific approach tends to produce more accurate correlations.This approach can be useful for specialized applications involving a single salt.However, to develop the model for general applications, the ion-specific approach is the way forward because its parameters can be used for different salts.It is more challenging because the parameters of several ions, cations and anions, are fitted at the same time, often using hundreds of data points.Also, these parameters represent a compromise solution that tries to capture the behavior of multiple properties of multiple systems.In such calculations, even with appropriate minimization methods, started from different initial estimates, the occurrence of local minima in the objective function cannot be ruled out and it is not possible to guarantee a perfect, simultaneous fit of all the thermodynamic properties for all salts.Thus, their ARD values can be low for some salts and high for others.Overall, these results are promising and motivate additional investigations of this model's performance in future work.
, who extended the Non-Random Two Liquid model (NRTL) to electrolyte solutions; -Papaiconomou et al. (2002), who used the e-NRTL equation, combining the NRTL model with the Born term and the restricted primitive MSA term, in order to describe systems formed by electrolytes and multiple solvents; -Nasirzadeh et al. (2005), who proposed an equation considering the sum of contributions of the Mean Spherical Approximation (MSA) and NRTL, and applied it to obtain the vapor-pressure and osmotic coefficients of aqueous solutions of lithium hydroxide; -Salimi et al. (2005), who combined the Ghotbi-Vera Mean Spherical Approximation (GV-MSA) with Figure 2Vapor pressure of KI in water at 303.15 K (h), 313.15K (D), 323.15K (s), 333.15K (9) and 343.15K (e), obtained byPatil et al. (1991).The full line (-) represents the electrolattice model considering ion-specific parameters.