Salting-out effect of sodium, potassium, carbonate, sulfite, tartrate and thiosulfate ions on aqueous mixtures of acetonitrile or 1-methyl-2-pyrrolidone: A liquid–liquid equilibrium

Page 1

Sa1

Ea

b

a

ARRAA

KA1CSA

1

fcptwtpbsi

btoftss

s

0h

Fluid Phase Equilibria 360 (2013) 357–366

Contents lists available at ScienceDirect

Fluid Phase Equilibria

jou rn al h om epage: www.elsev ier .com/ locate / f lu id

alting-out effect of sodium, potassium, carbonate, sulfite, tartratend thiosulfate ions on aqueous mixtures of acetonitrile or-methyl-2-pyrrolidone: A liquid–liquid equilibrium study

brahim Nemati-Kandea, Hemayat Shekaarib,∗

Young Researchers Club, Parsabad Mogan Branch, Islamic Azad University, Parsabad, IranDepartment of Physical Chemistry, University of Tabriz, Tabriz, Iran

r t i c l e i n f o

rticle history:eceived 23 May 2013eceived in revised form 22 August 2013ccepted 11 September 2013vailable online 20 September 2013

a b s t r a c t

The binodal curves, tie-line compositions and cloud point data as a function of temperature and concen-tration were measured for aqueous two phase systems composed of acetonitrile (ACN) + K2CO3 + H2O,ACN + Na2CO3 + H2O, 1-methyl-2-pyrrolidone (NMP) + K2CO3 + H2O and NMP + Na2CO3 + H2O. Addition-ally, salting-out ability of sodium sulfite, sodium thiosulfate and sodium tartrate was studied. The freeenergy, enthalpy and entropy of clouding point (CP) estimated using a simple method, and the driving

eywords:cetonitrile-Methyl-2-pyrrolidonearbonatealting-out

force of the two-phase formation process was discussed on the base of the estimated free energy values.Also, an empirical equation was modified as a function of organic solvents density and dielectric constantsand used for the simultaneous correlation of all experimental binodal data. Furthermore, e-NRTL and e-Wilson models were used for the correlation of tie-line compositions. The obtained results confirm thehigh performance of these models in the correlation of binodal and tie-line data.

TPS

. Introduction

The liquid–liquid extraction technique is a powerful methodor extraction and purification of significant chemical or biologi-al substances such as proteins, enzymes, nucleic acids or even cellarticles [1,2]. In this method a biphasic system was used to extracthe desired substance form one phase to the other one. Organic-ater biphasic system is one of the favorite systems utilizing in

hese extraction technique. The time, cost, scale of the processing,hysiochemical properties of the media and substance and also theioenvironmental factors are some of the factors that affecting theelection of the biphasic system to extract the substance from thenitial liquid phase [3,4].

Most of the organic solvents denature proteins and thereforeiochemists prefer to use the aqueous-organic mixed solvent ratherhan organic solvents in treatment of the proteins [5]. However,rganic-aqueous systems in some cases are not the proper systemsor extraction of biomolecules due to their unfavorable effects onhem [6]. Therefore, the influence of the organic-water biphasicystem on the extracted substrate should be considered in the

election of the extraction media.

Extraction efficiency can be improved by selection of differentolvents, solvent volume, pH and by using the “salting” effect. The

∗ Corresponding author. Tel.: +98 411 3393139; fax: +98 411 3340191.E-mail addresses: [email protected], [email protected] (H. Shekaari).

378-3812/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.fluid.2013.09.028

© 2013 Elsevier B.V. All rights reserved.

latest is more important, and is often referred to the “salting-out”or “salting-in” effects [7–10]. The salting-out effect, in fact, involvesthe addition of a “kosmotropic” (i.e. water structure making) salt toa miscible aqueous-organic mixture. In this respect, when the con-centration of the salt exceeds from a threshold concentration, thesystem becomes an immiscible biphasic one, in which, one phase is“water-rich phase” and the other phase is “organic-rich” [8]. Groverand Ryall [11] discussed different theories concern with the saltingeffect. However, a simple explanation of this effect can be madebased on the “hydration” nature of the ions. In this way, whena kosmotropic salt was added to the aqueous solution of a solu-ble organic solvent, the intermolecular interaction was affected byionization of the salt, whereas the stronger affinity between theions and the water molecules reflected by decreasing the availablewater molecules for the third component (i.e. organic solvent), andtherefore organic solvent enforced to increase its intermolecularinteractions, and subsequently, at threshold concentration of theionized species, the organic solvent was excluded from the rest ofthe solution as a separated phase [12].

Nowadays, salting effect found several applications in thenumerous separation processes, such as in rectification, to shiftfavorably azeotropic conditions, in extraction to alter miscibilitygaps and also in absorption and fractional crystallization to alter

the distribution coefficients [13]. In this respect, several groupswas studied the salting effect of different salts on different organic-aqueous systems based on the liquid–liquid equilibrium (LLE)measurements [14–19].

Page 2

358 E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366

Table 1Properties of the used pure chemicals.

Chemical Cas. Reg. No. Source Purity (in mass fraction) M (kg mol−1) d (kg m−3)a ε (C2 J−1 m−1)a

At T = 298.15 K

1-Methyl-2-pyrrolidone 872-50-4 Merck (Germany) >0.995 0.099133 1.0230 32.55Acetonitrile 75-05-8 Merck (Germany) >0.990 0.041052 0.7766 35.69K2CO3 584-08-7 Merck (Germany) >0.990 0.138206 – –Na2CO3 497-19-8 Merck (Germany) >0.995 0.105989 – –H2Ob 7732-18-5 Ghatreh (Iran) >0.999 0.018015 0.99704 78.30

0 �S c

wpmcboip[htfdiltesaa

m[attAct

tmlemestoc

tswcootveee

a Taken from Ref. [51].b The specific conductance of the double-distilled deionized water was about 0.7

Acetonitrile (ACN) and 1-methyl-2-pyrrolidone (NMP) areell-known medium-polarity solvents that are miscible in all pro-ortions in water and widely used in organic synthesis. Waterixtures of both solvents are also used for separation and purifi-

ation of organic compounds in organic synthesis. Both solventselong to the class of dipolar aprotic solvents [20–22]. ACN is alsone of the most preferred organic solvent or mobile phase in var-ous separation techniques because of its proper physicochemicalroperties like low viscosity, high resolution and low boiling point23]. However, ACN in acidic pH can yield hydrogen cyanide viaydrolysis in water [24]. Also, in the Federal Republic of Germany,he maximum concentration value at the workplace (MAK value)or ACN was suggested to be MAK ≤ 34 mg m−3 to prevent the skinamages [18]. Whereas, the toxicity of NMP is lower than ACN, and

t is suggested that small values of NMP can easily be used as solubi-izer cosolvent for medicinal agents at lower quantities comparedo other common organic cosolvents (ethanol, isopropanol, propyl-ne glycol and so on) [25]. Also, the value of MAK ≤ 80 mg m−3 wasuggested for NMP [18]. Similar properties of these two solventsnd the lower toxicity of NMP suggest that it may be a suitablelternative for ACN.

Also, sodium and potassium carbonate salts are nontoxic kos-otropic salts which are strongly hydrated with water molecules

26–28], and it is expected that have considerable effect on thequeous mixture of ACN or NMP. However, there is no report onhe study of the LLE of ternary water + salt + organic solvent sys-ems involving sodium carbonate or potassium carbonate salts andCN or NMP organic solvents. Therefore, the study these systemsan give reliable information about physicochemical properties ofhe consequent biphasic systems.

It is also valuable to note that the thermodynamic investiga-ion of the different organic + water + salt systems using reasonable

odels is an important part of the aforementioned studies. Theocal composition based models such as the NRTL [29] andlectrolyte-NRTL [30] models and the group contribution basedodels such as UNIQUAC [31] or UNIFAC [32] models have been

xtensively used for the correlation of the LLE data in such mixedolvent-electrolyte systems. Also, in previous works we extendedhe e-NRTL [33] and e-Wilson [12,19] models to represent the LLEf the systems composed of alcohol + salt + water systems and suc-essfully used these models to describe such systems.

In this paper the LLE of the aqueous solutions of ACN and NMP inhe presence of sodium carbonate and potassium carbonate weretudied. In this respect, binodal curves and tie-lines at T = 298.15 Kere measured. The effect of ACN, NMP, sodium and potassium

ations and also carbonate, sulfite, tartrate and thiosulfate anionsn the phase diagrams was discussed. The effect of temperaturen the phase-separation ability of these systems was studied usinghe experimental cloud point (CP) data as a function of organic sol-

ent mole fractions at T = 293.15–328.15 K with 5 K intervals. Thexperimental CP data was used to estimate the Gibbs free energy,nthalpy and entropy of clouding point (CP). The e-NRTL [33] and-Wilson [12,19] models were also used for the correlation of the

m−1.

tie-line compositions, and the binary interaction parameters werealso calculated.

2. Materials and methods

2.1. Materials

The physicochemical properties of the used chemicals weredescribed in Table 1. These chemicals were used without furtherpurification, and doubly distilled deionized water with specific con-ductance of about 0.70 �S cm−1 was used in all experiments.

2.2. Apparatus and procedure

The cloud point titration method was performed to collect thebinodal curve data. In this method, an appropriate amount of aque-ous solution of the salt solution or the organic solvent was placedin a double-wall glass cell, and the solution was stirred using amagnetic stirrer. The water at constant temperature was circulatedbetween the walls of the double-wall cell to control the tempera-ture of the cell. The temperature was controlled with an accuracyof ±0.03 K using a thermostat (JULABO model ED, Germany). Afterthe necessary rest time to establish the constant temperature, thedroplets of aqueous solution of another component (i.e. organic sol-vent, or salt) were added to the cell using a normal syringe, until thesolution was turned turbid. This point indicates that the system isin the biphasic region. Subsequently, the tiny droplets of the doubledistilled deionized water were added to the cloudy solution watch-fully, until the turbidity was vanished. This point indicates a nodeon the binodal curve. The mass changes were measured by ana-lytical balance (Sartorius model TE214S, Switzerland), and used tocalculate the composition of the organic solvent and the salt in thebinodal curve. The precision of the mass balance was ±1 × 10−7 kg.The procedure was repeated at least five times and the uncertaintyof the obtained binodal data was found to be better than ±0.0003(in mass fraction).

For determination of the tie-line compositions, appropriateamounts of the concentrate solutions of salt and the pure organicsolvent were mixed in glass cells and diluted by adding doublydistilled deionized water to form adequate feed samples (about10 cm3) which are in the biphasic region. These samples wereshaken (2400 cycles min−1) twice using a shaker (Labtron modelLS-100, Iran) for 3 min. After the first shaking, the samples wereimmersed in a thermostatted water bath at constant temperatureof T = 298.15 K for about an hour. Afterward, the samples wereshaken for the second time, and placed in the same bath to enrichthe equilibrium. Based on the equilibrium thermodynamics, in theequilibrium condition any macroscopic property of the systemsis stable, and therefore to ensure the occurrence of the thermo-

dynamic equilibrium the refractive index of some samples of theboth phases (i.e. water-rich and organic-rich) of several feed sam-ples was measured, using a refractometer (Atago, model DR-A1,Japan) with an uncertainty of the ±0.0002 in the refractive index

Page 3

id Phase Equilibria 360 (2013) 357–366 359

mnabTp

sdgcf

trofaiot

n

samciaomcwm

er

Table 2The coefficients of Eq. (1) for the studied organic solvents (m) + salt (ca) + water (w)systems at T = 298.15 K.

Systema n0w am 104SDm

b aca 104SDcab

m1 + ca1 + w 1.3325 0.1438 3.01 0.1611 2.68m1 + ca2 + w 1.3325 0.1426 3.08 0.2143 2.05m2 + ca1 + w 1.3325 0.0546 8.68 0.1645 2.88m2 + ca2 + w 1.3325 0.0542 6.77 0.2110 1.74

a m1, 1-methyl-2-pyrrolidone, m2, acetonitrile, ca1, potassium carbonate, ca2,sodium carbonate.

b SDj is the standard deviation between the calculated, cal, and experimental, exp,values of mass fraction, w, for component “j” (i.e. organic solvent (m) or salt (ca)) and

calculated ones using SDj =[∑

(wcal − wexp)2/n

]1/2. Moreover, n is the number of

TB

E. Nemati-Kande, H. Shekaari / Flu

easurement, at 1 h periods. Periodic measurements show that theecessary rest time to ensure the thermodynamic equilibrium isbout 3 h. However, the feed samples were immersed in the waterath for about (6–8) hours to enrich the equilibrium condition.he split phases were separated using long needle syringes andrepared by diluting for refractive index measurement.

After the separation of the two-phases, the concentrations ofodium or potassium carbonate in the top and bottom phases wereetermined by flame photometry (JENWAY model PFP7, U.K.). Theravimetric analyses reveal that the uncertainty of the obtained saltoncentration using this method was better than ±0.0003 (in massraction).

The refractive index measurement was performed to determinehe organic solvent concentration of the both split phases. In thisegard, the refractive indices of known solutions of the ternaryrganic solvent (m) + salt (ca) + water (w) systems in the massraction range of 0 ≤ wm ≤ 0.1 and 0 ≤ wca ≤ 0.05 was measuredt T = 298.15 K to found a proper relation between the refractivendex and organic solvent concentration. Satisfactorily result wasbtained when the experimental refractive indices data were fittedo the following simple relation:

D = n0w + amwm + acawca (1)

In this relation nD and n0w are the refractive indices of the ternary

olution and pure water at T = 298.15 K, respectively. Also, am andca are the constants, and acquired from the fitting of the experi-ental refractive indices of the standard solution to Eq. (1). These

onstants along with the relative standard deviations are reportedn Table 2, and the concentrations of the standard ternary solutionslong with the measured refractive indices are given in Table S1f Supplementary data associated with this article. It is proper toention that, all the unknown samples were diluted to be in the

alibration curve range (i.e. 0 ≤ wm ≤ 0.1 and 0 ≤ wca ≤ 0.05). Also, itas found that the accuracy of the calculation of the organic solvent

ass fraction using this method is better than ±0.0002.The cloud point titration method was also used to study the

ffect of temperature on the studied biphasic systems. In thisegard, an appropriate amount of the aqueous solution of the salt

able 3inodal curve data for organic solvents (m) + salt (ca) + water (w) ternary systems as a fun

100wm 100wca 100wm 100wca

1-Methyl-2-pyrrolidone (m) + potassium carbonate (ca) + water (w)5.53 ± 0.09 23.89 ± 0.27 18.80 ± 0.05 13.13 ± 0.03

7.63 ± 0.08 21.53 ± 0.16 19.99 ± 0.05 12.42 ± 0.02

9.37 ± 0.08 20.00 ± 0.16 20.76 ± 0.05 11.88 ± 0.02

12.55 ± 0.07 17.37 ± 0.07 22.07 ± 0.04 11.05 ± 0.02

14.96 ± 0.06 15.65 ± 0.04 23.59 ± 0.04 10.05 ± 0.01

1-Methyl-2-pyrrolidone (m) + sodium carbonate (ca) + water (w)6.85 ± 0.06 16.52 ± 0.10 13.78 ± 0.04 11.13 ± 0.02

7.47 ± 0.06 15.95 ± 0.09 14.90 ± 0.04 10.37 ± 0.02

8.36 ± 0.06 15.12 ± 0.07 15.87 ± 0.04 9.84 ± 0.01

9.18 ± 0.05 14.44 ± 0.06 17.2 ± 0.4 9.11 ± 0.01

10.43 ± 0.05 13.38 ± 0.03 17.83 ± 0.03 8.73 ± 0.01

12.14 ± 0.05 12.21 ± 0.02 18.91 ± 0.03 8.15 ± 0.01

Acetonitrile (m) + potassium carbonate (ca) + water (w)9.96 ± 0.08 14.63 ± 0.06 15.32 ± 0.04 9.70 ± 0.02

10.61 ± 0.06 13.87 ± 0.05 16.89 ± 0.04 8.58 ± 0.01

11.79 ± 0.05 12.61 ± 0.04 18.52 ± 0.03 7.67 ± 0.01

12.63 ± 0.05 11.74 ± 0.03 19.79 ± 0.03 6.93 ± 0.01

13.70 ± 0.05 10.93 ± 0.02 20.73 ± 0.03 6.46 ± 0.01

Acetonitrile (m) + sodium carbonate (ca) + water (w)4.35 ± 0.07 17.52 ± 0.19 10.1 ± 0.04 11.02 ± 0.03

5.78 ± 0.06 15.61 ± 0.11 11.62 ± 0.04 9.74 ± 0.02

6.75 ± 0.05 14.45 ± 0.08 12.22 ± 0.04 9.25 ± 0.02

7.62 ± 0.05 13.36 ± 0.06 13.11 ± 0.03 8.63 ± 0.02

8.49 ± 0.05 12.51 ± 0.05 13.8 ± 0.03 8.16 ± 0.01

9.32 ± 0.04 11.76 ± 0.04 14.52 ± 0.03 7.73 ± 0.01

i j,i j,i

measured refractive indices data.

was titrated with tiny droplets of pure organic solvent until thesolution was turned turbid, and the cell temperature was changedat 5 intervals until the cloudiness disappeared. Afterward, dropletsof the organic solvent were added to the solution until the solu-tion became turbid. As can be inferred, in this method the molefraction of organic solvent was changed whereas the relative saltto water concentration (i.e. mole fraction ratio) was remainedconstant.

3. Results and discussion

3.1. Phase diagrams

The experimental binodal curves and tie-line compositions foracetonitrile (ACN) + K2CO3 + H2O, ACN + Na2CO3 + H2O, 1-methyl-2-pyrrolidone (NMP) + K2CO3 + H2O and NMP + Na2CO3 + H2Oternary systems at T = 298.15 K are reported in Tables 3 and 4,respectively. The uncertainties of data were evaluated using the

recommended method of NIST [34]. Furthermore, as examples, thephase diagrams of ACN + K2CO3 + H2O and NMP + Na2CO3 + H2Osystems are shown in Figs. 1 and 2, respectively.

ction of mass fractions at T = 298.15 K.

100wm 100wca 100wm 100wca

24.90 ± 0.04 9.33 ± 0.01 30.25 ± 0.03 6.39 ± 0.0026.09 ± 0.03 8.65 ± 0.01 31.74 ± 0.02 5.62 ± 0.0027.03 ± 0.03 8.08 ± 0.01 32.63 ± 0.02 5.12 ± 0.0027.94 ± 0.03 7.58 ± 0.01

29.1 ± 0.03 6.94 ± 0.00

19.40 ± 0.03 7.87 ± 0.01 25.10 ± 0.02 5.09 ± 0.0020.22 ± 0.03 7.48 ± 0.01 26.05 ± 0.02 4.74 ± 0.0021.65 ± 0.03 6.74 ± 0.00 26.9 ± 0.02 4.32 ± 0.0022.76 ± 0.02 6.21 ± 0.00 27.75 ± 0.02 3.99 ± 0.0023.91 ± 0.02 5.67 ± 0.00 28.58 ± 0.02 3.63 ± 0.0024.62 ± 0.02 5.35 ± 0.00 29.33 ± 0.01 3.28 ± 0.00

21.76 ± 0.02 6.08 ± 0.00 27.41 ± 0.02 4.17 ± 0.0022.49 ± 0.02 5.81 ± 0.00 28.38 ± 0.02 3.88 ± 0.0023.90 ± 0.02 5.27 ± 0.00 30.13 ± 0.01 3.49 ± 0.0024.78 ± 0.02 5.01 ± 0.00 32.44 ± 0.01 3.15 ± 0.0025.83 ± 0.02 4.64 ± 0.00 33.34 ± 0.01 2.97 ± 0.00

15.54 ± 0.03 7.15 ± 0.01 20.66 ± 0.02 4.87 ± 0.0016.66 ± 0.03 6.60 ± 0.01 21.64 ± 0.02 4.49 ± 0.0017.61 ± 0.02 6.18 ± 0.01 22.43 ± 0.02 4.28 ± 0.0018.26 ± 0.02 5.80 ± 0.01 23.51 ± 0.02 4.00 ± 0.0019.18 ± 0.02 5.49 ± 0.0019.96 ± 0.02 5.16 ± 0.00

Page 4

360 E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366

Table 4Tie-line data for organic solvent (m) + sodium thiosulfate (ca) + water (w) systems as a function of mass fractions at T = 298.15 K.

Total composition Top phase Bottom phase

wm wca wm wca wm wca

1-Methyl-2-pyrrolidon (m) + potassium carbonate (ca) + water (w)20.72 13.92 34.77 ± 0.56 4.56 ± 0.45 5.66 ± 0.02 24.04 ± 0.4521.84 14.91 39.39 ± 0.96 2.98 ± 0.45 3.64 ± 0.01 27.05 ± 0.4422.83 16.00 42.97 ± 1.48 2.11 ± 0.44 2.78 ± 0.01 29.53 ± 0.4523.80 16.97 46.43 ± 2.14 1.59 ± 0.26 1.95 ± 0.00 31.86 ± 0.4524.80 18.01 49.44 ± 2.13 1.15 ± 0.25 1.83 ± 0.01 33.69 ± 0.4525.78 18.90 52.19 ± 2.47 0.85 ± 0.28 1.54 ± 0.00 35.42 ± 0.45

1-Methyl-2-pyrrolidon (m) + sodium carbonate (ca) + water (w)16.97 10.51 25.78 ± 0.39 4.50 ± 0.32 5.12 ± 0.02 18.62 ± 0.3117.67 10.97 28.93 ± 0.49 3.62 ± 0.31 3.66 ± 0.01 20.32 ± 0.3117.99 11.51 31.15 ± 0.67 2.89 ± 0.31 3.20 ± 0.00 21.2 ± 0.3118.69 11.99 33.68 ± 0.88 2.42 ± 0.31 2.55 ± 0.00 22.44 ± 0.3119.55 13.01 37.77 ± 0.91 1.69 ± 0.31 1.27 ± 0.01 24.55 ± 0.3120.76 13.95 41.49 ± 0.94 1.22 ± 0.31 0.62 ± 0.00 26.47 ± 0.31

Acetonitrile (m) + potassium carbonate (ca) + water (w)34.89 3.22 57.65 ± 1.19 0.54 ± 0.09 26.48 ± 0.22 4.19 ± 0.1336.08 3.78 61.82 ± 1.23 0.37 ± 0.07 21.99 ± 0.13 5.70 ± 0.1437.67 4.43 66.30 ± 1.36 0.20 ± 0.07 18.14 ± 0.09 7.34 ± 0.1239.42 5.54 71.24 ± 1.42 0.12 ± 0.06 14.96 ± 0.05 9.79 ± 0.1341.03 6.37 74.07 ± 1.49 0.09 ± 0.03 13.34 ± 0.03 11.73 ± 0.12

Acetonitrile (m) + sodium carbonate (ca) + water (w)13.02 12.00 67.72 ± 1.23 0.07 ± 0.00 8.72 ± 0.07 12.92 ± 0.0715.51 14.92 74.22 ± 1.34 0.03 ± 0.01 4.61 ± 0.05 17.69 ± 0.07

3d

bptlpsm

F(l

dc

17.66 15.98 76.60 ± 1.63

17.69 17.07 77.71 ± 1.7519.50 17.87 79.49 ± 1.86

.2. Effect of different solvents, anions and cations on the phaseiagrams

Because the binodal curve represents the phase separationoundary between the one-phase and two-phase regions the com-arison of the binodal curves may give valuable information abouthe salting-out ability of the two-phase systems. In other words, theower the concentration of the components needs to separate two

hases the higher the two-phase formation ability. This compari-on would be more precise if the comparison of these curves wereade in a molecular (or molar) frame. In many previous papers,

ig. 1. Experimental and calculated phase diagram for acetonitrile (m) + K2CO3

ca) + water (w) system at T = 298.15 K. (�): experimental binodal data; (—): calcu-ated binodal curve from modified Merchuk equation; (–©–): experimental tie-line

ata; (---×--- ) calculated tie-line compositions using e-NRTL model; (•••×••• ) cal-ulated tie-line compositions using e-Wilson model; (©) initial total compositions.

0.03 ± 0.00 3.06 ± 0.05 19.98 ± 0.070.03 ± 0.00 1.72 ± 0.01 21.49 ± 0.060.02 ± 0.01 1.24 ± 0.00 23.24 ± 0.07

the salting-out ability of different ATPSs is compared by the binodalcurves plotted in mass fractions basis, however, the mole fractionbasis is a suitable choice to have a comparison in the molecularframe. In this respect, the binodal curves for ACN + K2CO3 + H2O,ACN + Na2CO3 + H2O, NMP + K2CO3 + H2O and NMP + Na2CO3 + H2Oternary systems in mole fractions at T = 298.15 K are representedin Fig. 3. The mole fractions were calculated using the followingsimple relation:

w /M

xi = i i∑

j(wj/Mj)(2)

Fig. 2. Experimental and calculated phase diagram for 1-methyl-2-pyrrolidone(m) + Na2CO3 (ca) + water (w) system at T = 298.15 K. (�): experimental binodaldata; (—): calculated binodal curve from Eq. (5); (–©–): experimental tie-line data;

(---×--- ) calculated tie-line compositions using e-NRTL model; (•••×••• ) calculatedtie-line compositions using e-Wilson model; (©) initial total compositions.

Page 5

E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366 361

Fig. 3. Comparison of the mole fraction based binodal curves of different ATPSs atT = 298.15 K. (—�—) 1-Methyl-2-pyrrolidone (m) + potassium carbonate (ca) + water((t

wvc

oActmtN

tsarosdoltod(tlmtiotoa

nis

Fig. 4. Mole fraction based binodal curves for the ATPSs composed of acetoni-

w); (—�—) 1-methyl-2-pyrrolidone (m) + sodium carbonate (ca) + water (w);- - ♦ - -) acetonitrile (m) + potassium carbonate (ca) + water (w); (- - © - -) ace-onitrile (m) + sodium carbonate (ca) + water (w).

here i and j subscripts represents salt, water or the organic sol-ent. Also, wi and Mi are the mass fraction and the molar mass ofomponent i, respectively.

From Fig. 3 it is obvious that, the two-phase formation abilityf the systems composed of NMP is more than the ones for ACN.lso, from this figure, it can be concluded that the threshold con-entration of the Na2CO3 need to separate the two phases is lesshan K2CO3 and therefore the salting-out ability of the Na2CO3 is

ore than K2CO3. In other words, because the carbonate anion ishe common anion for these two salts, the salting-out ability of thea is more than K.

This observation can be discussed considering the changes inhe intermolecular interactions between the components of theystem when an electrolyte and an organic solvent are mixed inqueous solution. The “hydration theories” can depict a simpleepresentation of the salting-out phenomenon. Based on these the-ries [11], when the electrolyte solved in a solvent the strongerolvent–ion interactions overcomes the ion–ion interactions; theistance between the ions are increased, and consequently, eachf the constitutive ions of the electrolyte are surrounded by aayer of solvent. The stronger solvent–ion interactions immobilizehe solvent molecules and suppress their role as solvents. In anrganic + aqueous + electrolyte system the solvent molecules andissolved ions compete with each other to detain more solventi.e. water) molecules around their respective hydration shells. Inhis challenge the ionic species, which have more intermolecu-ar interactions with water molecules, can achieve more water

olecules and proportionally the other component is enforcedo decrease the interactions with water molecules and thereforets self-intermolecular interactions increased. Subsequently, therganic solvent is “salted-out” and when the amount of the elec-rolyte is increased from the specific threshold concentration therganic solvent molecules excluded from the remained solution as

separated phase.

It is obvious that in this representation of the salting-out phe-

omenon, the lower organic solvent–water interactions resultsn the lower concentration of the ionic species to salt-out theolvent (i.e. two-phase formation process). The sum of all possible

trile (ACN) + salt + water. (▬○▬ ): ACN + sodium carbonate + water; (▬♦▬ ):

ACN + sodium sulfite + water; (—◊— ): ACN + sodium thiosulfate + water; (–�–):ACN +sodium tartrate + water.

interaction forces between the molecules of solvent and solute canbe related to the so-called “polarity” of solvent and solute [20]. Inother words, the polar substances tend to dissolve in polar solventsand vice versa. Reichardt [20] tabulated the normalized solventpolarity parameters, EN

T , for a selection of 288 solvents based on thelong-wavelength UV/Vis charge-transfer absorption data. In histabulation, water is the most polar solvent with EN

T=298.15 = 1.000,and the values of 0.460 and 0.355 were reported for EN

T=298.15 ofACN and NMP, respectively. Comparison of the EN

T=298.15 valuesand the two-phase formation ability of the ACN and NMP revealsthat the lower polarity of the NMP leads to weaker NMP–waterintermolecular interaction, and subsequently, the more two-phase formation ability of NMP, which is in agreement with theprediction of the hydration theories.

Additionally, in this work, the slating-out ability of some othersodium slats was studied by measuring the binodal curves of theATPSs obtained from the addition of the aqueous solution of theseslats on ACN. The selected salts are some uni-bivalant slats includ-ing: sodium thiosulfate, sodium sulfite and sodium tartrate, that weare reported the ATPSs composed of them with some aliphatic alco-hols in our previous works [12,19,35]. The obtained binodal curvesalong with the one for Na2CO3 and K2CO3 are represented in Fig. 4.Also, the experimental binodal curve data for each of these ATPSscan be found in Table S2 of Supplementary data associated with thisarticle. The Na is the common cation in these series, and therefore,from Fig. 4 it can be concluded that, the salting-out ability of theconstituent anions of these salts is as follows:

carbonate > sulfite > thiosulfate > tartrate

The slating ability of different slats can be attributed to the Gibbsfree energy of hydrogen, �Ghyd, of the constituent ions [33]. �Ghydis the change in the free energy from an isolated naked ion in thegas phase to the aqueous hydrated ion in solution, and therefore,the ions with higher kosmotropicity have a more negative �Ghyd,due to the resulting more structured water ‘lattice’ around the ion

[8].

Marcus [36] reported the experimental values of −365 and−295 kJ mol−1 for �Ghyd of Na and K cations, respectively. Also,the values of −1315 and −1295 kJ mol−1 reported by Marcus [36]

Page 6

362 E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366

Table 5Cloud-point (CP) data along with the calculated �Gc , �Hc and �Sc for the organic solvent (m) + salt (ca) + water (w) systems as a function of mole fraction of the relevantNMP or ACN at temperature range T = 293.15–328.15 K.

T (K) 293.15 298.15 303.15 308.15 313.15 318.15 323.15 328.15

System 1-Methyl-2-pyrrolidon (m) + potassium carbonate (ca) + water (w) (xca/xw = 0.00883)xm 0.098 0.100 0.103 0.105 0.107 0.109 0.111 0.113�Gc (kJ mol−1) −5.66 −5.71 −5.73 −5.77 −5.82 −5.86 −5.91 −5.95�Hc (kJ mol−1) −3.26T�Sc (kJ mol−1) 2.4 2.44 2.48 2.52 2.56 2.61 2.65 2.69

System 1-Methyl-2-pyrrolidon (m) + sodium carbonate (ca) + water (w) (xca/xw = 0.00891)xm 0.073 0.075 0.078 0.08 0.083 0.085 0.087 0.089�Gc (kJ mol−1) −6.38 −6.42 −6.43 −6.47 −6.48 −6.52 −6.56 −6.6�Hc (kJ mol−1) −4.63T�Sc (kJ mol−1) 1.75 1.78 1.81 1.84 1.87 1.9 1.93 1.96

System Acetonitrile (m) + potassium carbonate (ca) + water (w) (xca/xw = 0.00886)xm 0.134 0.143 0.152 0.163 0.176 0.187 0.199 0.209�Gc (kJ mol−1) −4.9 −4.82 −4.75 −4.65 −4.52 −4.44 −4.34 −4.27�Hc (kJ mol−1) −10.41T�Sc (kJ mol−1) −5.49 −5.59 −5.68 −5.78 −5.87 −5.96 −6.06 −6.15

System Acetonitrile (m) + sodium carbonate (ca) + water (w) (xca/xw = 0.00893)xm 0.127 0.134 0.143 0.153 0.164 0.172 0.179 0.183

−1 −4

−4

ftoafrsas

3

mitTts

FvN

�Gc (kJ mol ) −5.03 −4.98 −4.9

�Hc (kJ mol−1)

T�Sc (kJ mol−1) −3.82 −3.89 −3.95

or carbonate and sulfite anions, respectively. Furthermore, usinghe method recommended by Marcus [36] we estimated the valuef −1160 and −1010 kJ mol−1 for �Ghyd of thiosulfate and tartratenions, respectively. For calculation of the tartrate and thiosul-ate ionic radii’s we used partial molar volume at infinite dilutioneported in Refs. [37,38]. Comparison of �Ghyd values and thealting-out abilities represented above, shows that the more neg-tive value of �Ghyd of the cation or anion results in the morealting-out ability of the ion.

.3. Cloud-point (CP) data and the free energies of the CP

The cloud-point (CP) temperature as a function of NMP or ACNole fractions at temperature range of T = 293.15–328.15 K at 5 K

ntervals was measured to study the effect of temperature on

he investigated systems. The experimental CP data, reported inable 5 and plotted in Fig. 5, shows the NMP or ACN concen-ration dependence of CP for the same concentration of aqueousodium or potassium carbonate solutions for each of the studied

ig. 5. Effect of temperature on cloud points, CP, as a function of the organic sol-ent mole fraction for the studied ATPSs. (—�—): NMP + Na2CO3 + H2O; (—�—):MP + K2CO3 + H2O; (–�–): ACN + Na2CO3 + H2O; (–�–): ACN + K2CO3 + H2O.

.81 −4.71 −4.66 −4.62 −4.63−8.84

.02 −4.08 −4.15 −4.21 −4.28

systems. As shown in Fig. 5, in the case of all ATPSs, the con-centration of NMP or ACN required to achieve a phase separationslightly decreases with increasing the temperature. In other words,the phase-separation ability of all studied ATPSs in the temper-ature range of T = 293.15–328.15 K increased with increasing thetemperature. On the other point of view, by increasing the tempera-tures the electrostatic interactions between ionic species and watermolecules more precisely can overcome the organic solvent–waterinteractions, and therefore the phase separation process more eas-ily can be occurred.

Recently, Dan et al. [39], considering CP as the point of phaseseparation (or the solubility limit) estimated the free energy, �Gc,enthalpy, �Hc, and entropy, �Sc, of phase-separation to obtainsome information about the driving force of the aqueous two-phase formation process. Following Dan et al. [39], the free energyof phase separation or clouding (�Gc) can be calculated from thefollowing relation:

�Gc = RT ln Xm (3)

where Xm is the mole fraction concentration of the organic solventat CP. The �Gc values calculated using the experimental CP valuesof NMP or ACN for the same slat to water mole fraction ratios, andare reported in Table 5. The �Gc values at different CP (or T) wereprocessed according to the equation given below to get �Hc fromthe slope of the linear (least squares) plot of (�Gc/T) against (1/T)for each of the studied ATPSs:

�Hc = d(�Gc/T)d(1/T)

(4)

The calculated �Hc values are presented in Table 5 which arenegative for all studied ATPSs. The negative values of �Hc forthese systems demonstrate that the phase-separation process isan exothermic process.

Moreover, the following Gibbs–Helmholtz equation was used tocalculate the entropy changes:

�Gc = �Hc − T�Sc (5)

The T�Sc values calculated from Eq. (5) are also reported in

Table 5, which are positive in the case of ATPSs composed of NMP.While on the contrary, the estimated T�Sc values for the ATPSscomposed of ACN are negative. The positive entropy and the neg-ative enthalpy values for the ATPSs composed of NMP propose

Page 7

E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366 363

Table 6Parameters of Merchuk equation as a function of density and dielectric constant used for the simultaneously correlation of the binodal curves of the studied ATPSs atT = 298.15 K.

System a0 a1 a2 a3 b0 b1 b2 b3 c0 c1 c2 c3 sda

40.1522 0.7439 −28.1397 −0.9991 −58.6756 0.439 −48.9853 0.9791 17,930.15 518.3362 −21,007.2 −524.404

NMP + Na2CO3 + H2O 0.03NMP + K2CO3 + H2O 0.16ACN + Na2CO3 + H2O 0.09ACN + K2CO3 + H2O 0.06Overall 0.10

n of re

a

tmeob

3

3

csscsisssttoostisceoftwoeceMfif

w

wsToa

a sd =((∑

i(100wexp

m,i− 100wcal

m,i))

/N)0.5

, where wm represent the mass fractio

nd “cal” stand for the experimental and calculated values, respectively.

hat both entropy and enthalpy factors are suitable to enforce theixed-solvent electrolyte systems to became a biphasic one. How-

ver, in the case of the ATPSs composed of ACN the driving forcef two phase formation process may be the decrease of enthalpy,ecause of their negative enthalpy and entropy values.

.4. Correlation of the phase diagrams

.4.1. Correlation of the binodal curvesSeveral empirical or semi-empirical equations were used to the

orrelation of the binodal data [40–45]. However, the literatureurvey shows that, there is no attempt to obtain an equation forimultaneously correlation of the binodal curves of the systemsomposed of mixed solvents until now. In a simple description, theolvent can be considered as a continuous medium characterized byts physical properties [46]. Melting and boiling point, vapor pres-ure, heat of vaporization, index of refraction, density, viscosity,urface tension, dipole moment, dielectric constant, polarizability,pecific conductivity, etc. are some important physical propertieshat can be used in the classification of solvents [18]. However,he properties of the mixed solvent depend on the compositionf the constituent solvents, and therefore, significant descriptionf a mixed solvent can be made using composition dependence ofome of the physical properties of the subsequent solvents. Fromhe physical properties of solvents the dielectric constant, ε, is anmportant one that can be considered as a factor describing theolvent polarity. The importance of this property is such that inontinuum models the ε is solely characterizing the electrostaticffects [46]. The density of the medium was also effectively dependsn the concentration and therefore may be added in the correlationunction to have an efficient description of the mixed solvent withhe minimum number of parameters. In this respect, in previousorks [47,48], considering the solvent composition dependence

f the density and dielectric constant, we have modified Merchukquation to obtain a simple empirical relation for the simultaneousorrelation of the binodal data for the systems composed of differ-nt solvents. In the same manner, in this work we have modifiederchuk equation in slightly different solvent dependence form to

t all of the binodal data simultaneously. Merchuk equation has theollowing form [49]:

m = a exp(bw0.5ca − cw3

ca) (6)

here wca and wm denote the mass fraction of salt and organicolvent in the binodal curve, and a, b and c are fitting parameters.o obtain an appropriate equation for simultaneously correlationf all binodal data the fitting parameters of Eq. (6) are expressed as

nonlinear function of dielectric constant and density, as follows:

−0.5 −0.5

a = a0εs + a1 lnεs + a2ds + a3 ln ds

b = b0ε−0.5s + b1 ln εs + b2d−0.5

s + b3 ln ds

c = c0ε−0.5s + c1 ln εs + c2d−0.5

s + c3 ln ds

(7)

levant organic solvent, N is the number of binodal data, and also superscripts “exp”

where a, b and c are the parameters of Eq. (6) and ai, bi and ci (i = 0,1 or 3) are new parameters as a function of dielectric constant anddensity. Furthermore, εs and ds are the mixed solvent dielectricconstant and density, respectively. εs and ds values are calculatedfrom the simple composition average mixing rules adopted by Chenet al. [50], as follows:

1ds

=∑

m

xm∑m′ xm′

1dm

(8)

εs =∑

m

xmMm∑m′ xm′ Mm′

εm (9)

where Mm and xm are the molecular weight and the mole fractionof the solvent m.

Required densities and dielectric constants were obtained fromRef. [51]. The experimental binodal data reported in Table 3 for thestudied systems were fitted by non-linear least-square’s regressionmethod to the obtained equation, and the fitting parameters alongwith the corresponding standard deviation for each of the studiedsystems are given in Table 6. Furthermore, the experimental andcalculated binodal curves are shown in Fig. 3. On the base of thestandard deviations (sd) reported in Table 6 and the results shownin Fig. 3 it can be concluded that the obtained equation as a functionof density and dielectric constant can be successfully used for thesimultaneous correlation of all experimental binodal data.

3.4.2. Correlation of the tie-line compositionsIn recent years, several models have been developed for the

correlation of phase equilibrium of ATPSs [12,19,29–33]. Also, wehave used e-NRTL and e-Wilson models for the correlation of ATPSscomposed of aliphatic alcohol + salt + water systems in our previousworks [12,19], and in this work, we decided to further examine theperformance of e-NRTL and e-Wilson models in the correlation ofthe tie-lines obtained in this work. The model development for thee-NRTL and e-Wilson models can be found in details in our previ-ous works [12,19]. Here, we only presented general expressions ofthese models.

Both e-NRTL and e-Wilson models are build up from two con-tributions for the excess Gibbs energy, GE, a Pitzer–Debye–Hückel(PDH) contribution [52] to account the long-range interaction con-tribution, GE,PDH, and a local composition based contribution (i.e.NRTL or Wilson models) to account the short-range interactioncontribution, GE,LC:

GE = GE,PDH + GE,LC (10)

Accordingly, appropriate derivation of Eq. (10) gives the activitycoefficient of component j as follows:

ln � = ln �PDH + ln �LC (11)

j j j

In the case of LLE of organic solvent + salt + water systems thedifference in the dielectric constants and the densities of the sol-vents will be large, and a physically correct description will require

Page 8

364 E. Nemati-Kande, H. Shekaari / Fluid Phase Equilibria 360 (2013) 357–366

Table 7Values of restricted parameters of e-NRTL and e-Wilson models for the studied organic solvents (m) + salt (ca) + water (w) systems at 298.15 K. The highlighted values areobtained from the correlation of the binary aqueous data.

Systema �w,cab �ca,w

b 104SDbc �w,m �m,w 103SDb

c �ca,m �m,ca SDtd

e-NRTL modelm1 + ca1 + w −14.7087 6.548 12.81 −0.9413 −0.9924 – 4.7946 −3.6852 0.29m1 + ca2 + w −17.1736 2.8885 7.21 9.3443 20.4752 – −0.6317 −12.8643 0.21m2 + ca1 + w −14.7087 6.5480 12.81 1.5468 1.2702 6.83 1.4228 −0.4072 0.05m2 + ca2 + w −17.1736 2.8885 7.21 1.5468 1.2702 6.83 52.6876 −8.5752 0.30

System Hw,caa Hc,aw

a 104SDbc Hw,m Hm,w 103SDb

c Hca,m Hm,ca SDtd

e-Wilson modelm1 + ca1 + w 8.41800 0.5492 15.15 1.0133 1.0982 – −0.0569 0.3335 0.22m1 + ca2 + w 14.0704 0.03700 3.12 0.4418 0.9403 – 3.2124 −61.6431 0.10m2 + ca1 + w 8.41800 0.5492 15.15 0.891 0.8981 20.77 0.6190 −0.5863 0.04m2 + ca2 + w 14.0704 0.03700 3.12 0.891 0.8981 20.77 −2.0150 39.0642 0.10

a m1, 1-methyl-2-pyrrolidone; m2, acetonitrile; ca1, potassium carbonate; ca2, sodium carbonate.b The highlighted parameters are obtained from the correlation of the binary water activity data obtained from Refs. [54–56].c SDb =

((∑i(awexp

i− awcal

i)2)

/N)0.5

, and stands for the standard deviation of the binary water activity data. In this relation aw represent the activity of water, N is the

number of data, and also superscripts “exp” and “cal” stand for the experimental and calculated values, respectively.

dard d

I alt or

tdssrscw

vtmopv�biaoa

wm

tccrt(a

O

w“usb

1 ˛

d SDt =∑

p

∑l

∑j

[((100wcal

p,l,j,T− 100wexp

p,l,j,T)2/6N)

]0.5, and stands for the stan

n this relation wp,l,j , is the weight fraction of the component j (i.e. organic solvent, s

he use of a solvent composition-dependent dielectric constant andensity [17]. A possible way to consider the properties of the mixedolvent is to account the dependency of dielectric constant and den-ity of mixed-solvent on the composition of any solvent. In thisegard, Chen et al. [50] adopted Eqs. (8) and (9) as simple compo-ition average mixing rules to calculate the density and dielectriconstant of the mixed solvent, respectively, and the obtained valuesere used in the desired models.

For correlation of the tie-line data, in the PDH contribution, thealue of � = 14.9 has been used, that has been frequently used forhe aqueous electrolyte solutions [12,17,30,33,50,53]. The e-NRTL

odel has nine interaction parameters. ˛ij (where i,j = water (w),rganic solvent (m) or slat (ca) and stands for binary couples of com-onents) are the non-randomness factors that usually set in a fixedalue, and the binary interaction parameters, �w,ca, �ca,w, �m,w, �w,m,m,ca and �ca,m. The first four binary interaction parameters cane obtained from the correlation of the experimental water activ-

ty, osmotic coefficient or vapor liquid equilibrium (VLE) data forqueous salt or organic solvent solutions. The two remaining salt-rganic solvent, �ca,m, and organic solvent–salt, �m,ca, parametersre usually determined from the fitting of LLE data.

The e-Wilson model has only six binary interaction parameters,hich are represented by Hij, where subscripts i and j has the sameeaning as described for e-NRTL model.In this work, the salt–water, and water–salt, binary interac-

ion parameters of e-NRTL or e-Wilson models obtained from theorrelation of the water activity data at T = 298.15 K for sodiumarbonate + water and potassium carbonate + water binary systemseported by Peiper et al. [54] and Sarbar et al. [55], respectively. Inhe correlation procedure the following binary objective functionOfb) was minimized using Levenberg–Marquardt optimizationlgorithm.

fb =∑

i

(aexpw,i

− acalw,i)

2(12)

here, aw,i represents the i’th water activity data, and superscripts

˛Na2CO3,w = ˛w,Na2CO3 = 0.

exp” and “cal” stand for the experimental and calculated val-es, respectively. Also, the “water–organic solvent” and “organicolvent–water” binary interaction parameters for ACN + waterinary systems was obtained from the correlation of water

eviation between the calculated, cal, and experimental, exp, tie-line compositions.

water) in the phase p for lth tie-line, and N represents the number of tie-line data.

activity data reported by French [56] using the same procedure (i.e.minimization of Eq. (12)).

All the binary interaction parameters of e-NRTL and e-Wilsonmodels that are obtained from the correlation of the binary wateractivity data are reported in Table 7 and are highlighted to distin-guish from the remaining ones. In should be noted that, in the caseof e-NRTL model the best results were obtained when the followingnon-randomness factor was used:

K2CO3,w = ˛w,K2CO3 = 0.01 ˛Acetonitrile,w = ˛w,Acetonitrile = 0.4

Also, C parameter of e-Wilson model treated as a fixed value andthe value of C = 10 was used.

Based on the binary standard deviations (SDb) values reportedin Table 7, it can concluded that the performance of both e-NRTLand e-Wilson models in the correlation of the binary water activitydata for the mentioned systems are excellent.

It is also reliable to mention that, we cannot access any binarydata for NMP + water binary systems at T = 298.15 K, and therefore,in this case the NMP–water and water–NMP interaction parameterswere obtained from the correlation of the LLE data.

The aforementioned binary interaction parameters were fixedin the studied ATPSs, and the other remaining restricted parametersare obtained from the correlation of the LLE data using the followingprocedure.

The requirement for thermodynamic equilibrium is that theGibbs free energy is at a minimum. It can be shown from classi-cal thermodynamics that a two-phase system at constant pressureand temperature containing component i (organic solvent, salt orwater) at top (top) and bottom (bot) phases will obey the followingconstraints at equilibrium [33,47]

(xj�j)top = (xj�j)

bot (13)

where x and � represent the mole fraction and the activity coeffi-cient, respectively. The interaction parameters are evaluated fromthe fitting of experimental LLE data to Eq. (13) using the followingternary objective function (Oft):

Oft =∑

p

∑

l

∑

j

(xexpp,l,j

− xcalp,l,j)

2(14)

where xp,l,j is the mole fraction of the component j in the phasep for lth tie-line and the superscripts “cal” and “exp” refer to thecalculated and experimental values, respectively. In Eq. (14) thespecies j can be the organic solvent, salt or water.

Page 9

id Pha

mdmtftsciNdtaeaaorttf

4

Kwtect

tsaoitttimrot

oo

oiGbsot

dud

[

E. Nemati-Kande, H. Shekaari / Flu

The obtained restricted binary interaction parameters using thisethod are also reported in Table 7 along with the ternary standard

eviation (SDt), which is refers to the ability of e-NRTL or e-Wilsonodels in the correlation of the ternary LLE data. It should be noted

hat, in the case of e-NRTL model the remaining non-randomnessactors were fixed at 0.25. Furthermore, to show the reliability ofhe mentioned models in the correlation of the tie-line compo-itions, as examples, comparison between the experimental andalculated phase equilibrium data using the parameters reportedn Table 7 are shown in Figs. 1 and 2 for ACN + K2CO3 + H2O andMP + Na2CO3 + H2O systems, respectively. Also, all the calculatedata using both models are reported in Table S3 of the suppor-ing information along with the absolute deviations percent andbsolute relative deviation percent between the calculated andxperimental data for each of the tie-lines. The total absolute aver-ge deviation percent of 0.5141 and 0.2527 for all of 22 tie-linesre obtained for e-NRTL and e-Wilson models, respectively. Basedn the obtained results, it can be concluded that both models canepresent tie-line compositions of the investigated systems usinghe restricted binary interaction parameters excellently. However,he performance of e-Wilson model, that has no non-randomnessactors, is slightly better than e-NRTL model.

. Conclusions

Liquid–liquid equilibrium of NMP + Na2CO3 + H2O, NMP +2CO3 + H2O, ACN + Na2CO3 + H2O and ACN + K2CO3 + H2O ATPSsere studied. The binodal curve data and tie-line compositions for

he investigated systems were reported at T = 298.15 K. Also, theffect of temperature on the ATPSs was studied by measuring theloud point data as a function of organic solvent compositions andemperature.

Also, binodal curves at T = 298.15 K for sodium sulfite, sodiumhiosulfate and sodium tartrate + ACN + water systems were mea-ured. The obtained data shows that the two-phase separationbility of NMP is more than ACN. This result can be discussed basedn the hydration theory of the salting-out phenomenon. Accord-ng to this theory and molecular point of view, it concluded that,he molecules of the solvent with lower polarity (i.e. NMP), andherefore the weaker water-solvent intermolecular interactions, inhe lower concentration of the salt loses the competition againstons, that has the stronger electrostatic interactions with water

olecules, to detain more water molecules and therefore moreapidly salted-out as a separated phase. Also, the experimental bin-dal data showed the following order of the salting-out ability forhe studied cations and anions, respectively:

Na > Kcarbonate > sulfite > thiosulfate> tartrate.

This trend was attributed to the Gibbs free energy of hydrationf these ions. It was obtained that, the more negative value of �Ghydf the ion results in the more salting-out ability of the ion.

The measured CP data showed that the phase formation abilityf all studied ATPSs in the temperature range of T = 293.15–328.15 Kncreased with increasing the temperature. Also, the calculatedibbs free energy, enthalpy and entropy of CP data inform thatoth entropy and enthalpy factors are suitable to form a biphasicystem in the case of ATPSs composed of NMP, whereas, decreasingf enthalpy is the driving force of two-phase formation process forhe ATPSs composed of ACN.

Moreover, an empirical equation was modified as a function ofensity and dielectric constant of NMP and ACN and successfullysed for the simultaneous correlation of all experimental binodalata. The e-NRTL and e-Wilson models were also used for the

[[[[

se Equilibria 360 (2013) 357–366 365

correlation of the tie-line compositions. It was found that, the per-formance of both models in the correlation of LLE data is excellent;however, the ability of e-Wilson model which has less parameterwas slightly better than e-NRTL model.

List of symbolsa, b, and c parameters of Eq. (6)ACN acetonitrileC the parameter of Wilson model fixed at C = 10am constants of Eq. (1) for organic solventca slatd density (kg m−3)GE excess Gibbs energy (J mol−1)Hij Wilson binary interaction parameter between compo-

nent i and jNMP 1-methyl-2-pyrrolidonen0

w refractive index of pure waterOf objective functionSD standard deviationT temperature (K)w mass fractionx mole fraction

Greek letters˛ nonrandomness factors of e-NRTL model� activity coefficientε dielectric constant (=4� 8.85 × 10−12εr) (C2 J−1 m−1)� closest approach parameter

Subscripts and superscriptscal calculated valueca saltexp experimental valuem organic solventn component numbertop upper(i.e. organic solvent reach) phasebot lower (i.e. water reach) phasePDH Pitzer–Debye–Hückel contributionLC local composition contribution

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.fluid.2013.09.028.

References

[1] P.G. Mazzola, A.M. Lopes, F.A. Hasmann, A.F. Jozala, T.C. Penna, P.O. Magal-haes, C.O. Rangel-Yagui, A. Pessoa Jr., J. Chem. Technol. Biotechnol. 83 (2008)143–157.

[2] C.M. Ballard, J.P. Dickinson, J.J. Smith, Biochim. Biophys. Acta 582 (1979)89–101.

[3] M. Martínez-Aragón, S. Burghoff, E.L.V. Goetheer, A.B. de Haan, Sep. Purif. Tech-nol. 65 (2009) 65–72.

[4] B.Y. Zaslavsky, Aqueous Two-phase Partitioning, Physical Chemistry and Bio-analytical Applications, Marcel Dekker, New York, 1995.

[5] T. Burgi, A. Baiker, J. Am. Chem. Soc. 120 (1998) 12920–12926.[6] Y.G. Wu, M. Tabata, T. Takamuku, A. Yamaguchi, T. Kawaguchi, N.H. Chung,

Fluid Phase Equilib. 192 (2001) 1–12.[7] K. Griebenow, A.M. Klibanov, J. Am. Chem. Soc. 118 (1996) 11695–11700.[8] M.T. Zafarani-Moattar, S. Hamzehzadeh, Fluid Phase Equilib. 304 (2011)

110–120.[9] R. De Santis, L. Marrelli, P.N. Muschetta, Chem. Eng. J. 11 (1976) 207–214.10] N. Bourayou, A.H. Meniai, Desalination 185 (2005) 473–481.

11] P.K. Grover, R.L. Ryall, Chem. Rev. 105 (2005) 1–9.12] E. Nemati-Knade, H. Shekaari, S.A. Jafari, Fluid Phase Equilib. 321 (2012) 64–72.13] A. Hasseine, A.H. Meniai, M. Korichi, Desalination 242 (2009) 264–276.14] H.N. Solimo, C.M. Bonatti, J.L. Zurita, M.B. Gramaio, Fluid Phase Equilib. 137

(1997) 163–172.

Page 10

3 id Pha

[

[[

[[[

[[

[[

[

[[[[[[[[

[

[

[[[[[[

[[

[

[[[

[[[[

[

66 E. Nemati-Kande, H. Shekaari / Flu

15] J.L. Zurita, M.B. Gramajo, C.M. Bonatti, H.N. Solimo, J. Chem. Eng. Data 43 (1998)1039–1042.

16] E.A. Macedo, P. Skovborg, P. Rasmussen, Chem. Eng. Sci. 45 (1990) 875–882.17] G.H. van Bochove, G.J.P. Krooshof, T.W. de Loos, Fluid Phase Equilib. 171 (2000)

45–58.18] M.L. Desai, E.O. Eisen, J. Chem. Eng. Data 16 (1971) 200–202.19] E. Nemati-Kande, H. Shekaari, S.A. Jafari, Fluid Phase Equilib. 329 (2012) 42–54.20] C. Reichardt, Solvents and Solvent Effects in Organic Chemistry, third ed.,

WILEY-VCH, Weinheim, 2003.21] L. Dinana, J. Harmathab, R. Lafont, J. Chromatogr. A 935 (2001) 105–123.22] I.M. Smallwood, Handbook of Organic Solvent Properties, Halsted Press, New

York, 1996.23] M. Rubinstien, Anal. Biochem. 98 (1979) 1–7.24] M. Greenberg, Toxicological Review of Acetonitrile, Environmental Protection

Agency, Washington, DC, USA, 1999.25] A. Jouyban, M.A.A. Fakhree, A. Shayanfar, J. Pharm. Pharmaceut. Sci. 13 (2010)

524–535.26] A. Salabat, M. Hashemi, J. Chem. Eng. Data 51 (2006) 1194–1197.27] M. Hu, L. Jin, Q. Zhai, S. Li, Z. Liu, Fluid Phase Equilib. 232 (2005) 57–61.28] Y. Wang, S. Hu, Y. Yan, W. Guan, Calphad 33 (2009) 726–731.29] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135–144.30] B. Mock, L.B. Evans, C.C. Chen, AIChE J. 32 (1986) 1655–1664.31] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116–128.32] A. Fredenslund, R.L. Jones, J.M. Prausnitz, AIChE J. 21 (1975) 1086–1099.

33] M.T. Zafarani-Moattar, E. Nemati-Kande, A. Soleimani, Fluid Phase Equilib. 313

(2012) 107–113.34] B.N. Taylor, C.E. Kuyatt, Guidelines for the Evaluation and Expression of Uncer-

tainty in NIST Measurement Results, NIST Technical Note 1297, GovernmentPrinting Office Washington, Gaithersburg, MD, U.S., 1994.

[[[

[

se Equilibria 360 (2013) 357–366

35] E. Nemati-Kande, H. Shekaari, S.A. Jafari, J. Chem. Eng. Data 57 (2012)2336–2342.

36] Y. Marcus, J. Chem. Soc. Faraday Trans. 87 (1991) 2995–2999.37] M.T. Zafarani-Moattar, B. Asadzadeh, J. Chem. Eng. Data 53 (2008) 1000–1006.38] I.V. Olofsson, J.J. Spitzaer, L.G. Hepler, Can. J. Chem. 56 (1978) 1871–1873.39] A. Dan, S. Ghosh, S.P. Moulik, J. Phys. Chem. B 112 (2008) 3617–3624.40] T.A. Graber, M.E. Taboada, J. Chem. Eng. Data 45 (2000) 182–184.41] G. Tubio, L. Pellegrini, B.B. Nerli, G.A. Pico, J. Chem. Eng. Data 51 (2006)

209–212.42] M.T. Zafarani-Moattar, A.A. Hamidi, J. Chem. Eng. Data 48 (2003) 262–265.43] M. Foroutan, N. Heidari, M. Mohammadlou, J. Chem. Eng. Data 53 (2008)

242–246.44] S.M. Waziri, B.F. Abu-Sharkh, S.A. Ali, Fluid Phase Equilib. 205 (2003)

275–290.45] Y. Guan, T.H. Lilley, T.E. Treffry, Macromolecules 26 (1993) 3971.46] C.J. Cramer, D.G. Truhlar, Chem. Rev. 99 (1999) 2161–2200.47] E. Nemati-Kande, H. Shekaari, M.T. Zafarani-Moattar, J. Chem. Eng. Data 57

(2012) 1678–1688.48] E. Nemati-Kande, H. Shekaari, J. Solution Chem. 41 (2012) 1649–1663.49] J.C. Merchuk, B.A. Andrews, J.A. Asenjo, J. Chromatogr. B 711 (1998) 285–293.50] C.C. Chen, Y. Song, AIChE J. 50 (2004) 1928–1941.51] D.R. Lide, CRC Handbook of Chemistry and Physics, 87th ed., Taylor and Francis,

Boca Raton, FL, 2007.52] K.S. Pitzer, J. Am. Chem. Soc. 102 (1980) 2902–2906.

53] J.M. Simonson, K.S. Pitzer, J. Phys. Chem. 90 (1986) 3009–3013.54] J.C. Peiper, K.S. Pitzer, J. Chem. Thermodyn. 14 (1982) 613–638.55] M. Sarber, A.K. Covington, R.L. Nuttal, R.N. Goldberg, J. Chem. Thermodyn. 14

(1982) 695–702.56] H.F. French, J Chem. Thermodyn. 19 (1987) 1155–2116.