J. Chem.

Thermodynamics 51 (2012) 7781

Contents lists available at SciVerse ScienceDirect

J. Chem. Thermodynamics
journal homepage: www.elsevier.com/locate/jct

Experimental study on (vapor + liquid) equilibria of ternary systems of

hydrocarbons/ionic liquid using headspace gas chromatography
Babak Mokhtarani , Leila Valialahi, Kurosh Tabar Heidar, Hamid Reza Mortaheb, Ali Shari,
Mojtaba Mirzaei
Chemistry and Chemical Engineering, Research Center of Iran, P.O. Box 14335-186, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: (Vapor + liquid) equilibrium (VLE) data for ternary systems of (hexane + benzene), (hexane + cyclohex-
Received 18 December 2011 ane), (benzene + cyclohexane), (1-hexene + cyclohexane), and (1-hexene + benzene) with an ionic liquid
Received in revised form 19 February 2012 were measured by headspace gas chromatography. The applied ionic liquid 1-methyl 3-octylimidazolium
Accepted 24 February 2012
thiocyanate, [Omim][SCN], acts as an entrainer. The comparison of the measured VLE data with the equi-
Available online 5 March 2012
librium data for the binary mixtures without ionic liquid show that [Omim][SCN] signicantly improves
the separation factor of these systems. The NRTL thermodynamic model is applied for correlating the
Keywords:
experimental data. The modeling results show the NRTL model can correlate the experimental data with
Ionic liquid
(Vapor + liquid) equilibria
a good accuracy.
Headspace gas chromatography 2012 Elsevier Ltd. All rights reserved.
NRTL model

1. Introduction Headspace gas chromatography (GC) is an efcient method for

Ionic liquids (ILs) are new solvents, which show very interesting
properties such as low vapor pressure, non-ammability, chemical
stability at high temperatures, and excellent solubility with organ-
ic and inorganic compounds. Due to negligible vapor pressure of
ILs, these substances can be used as substitute for toxic and volatile
organic compounds. As the ILs are composed of a cation and an an-
ion, their physical properties can be adjusted by appropriate selec-
tion of the cation or anion [1]. In spite of the promising properties
of ILs, only restricted applications are dened for these new sol-
vents in industry. This is partly because of the lack of their physi-
cochemical data either in pure forms or as mixture with organic/
inorganic compounds. In many industrial applications, it is essen-
tial to know how they interact with different solutes. Therefore,
knowing the thermodynamic behavior of IL in different solute
and having the reliable experimental data of phase equilibria have
a great importance.
As a new application for ILs, they can be used as an entrainer to
facilitate the separation of close-boiling binary hydrocarbons, and
can be appropriate alternatives for toxic and volatile organic sol-
vents such as sulfolane, N-methylpyrrolidone (NMP), and N-form-
ylmorpholine (NFM). In recent years, a number of experimental
VLE data for the ILs/hydrocarbons have been reported in several
publications [27].
Corresponding author. Tel.: +98 21 44580751; fax: +98 21 44580781.
E-mail address: mokhtaranib@ccerci.ac.ir (B. Mokhtarani).
measuring VLE data of binary hydrocarbons [8]. The composition
of the vapor phase is determined by gas chromatography. This
method saves time and reduces the potential errors that are com-
mon in other GC sample preparation techniques. The effect of IL
present in the hydrocarbon binaries has not widely studied by this
method. Liebert et al. [6] measured the (vapor + liquid) equilibria
(VLE) of (methylcyclohexane + toluene) and (1-octene + octane)
mixtures in the presence of 1-hexyl-3-methyl-imidazolium bis
(triuoromethyl-sulfonyl)imide [Hmim][BTI]. Westerholt et al. [7]
investigated the effect of 1-butyl-3-methylpyrrolidinium bis
(triuoromethyl-sulfonyl)imide [Bmpyr][BTI] and [Hmim][BTI] on
the separation (benzene + cyclohexane), (1-hexene + n-hexane),
and (2-propanol + water) binaries. The effect of different ILs,
1-octyl-3-methyl-imidazolium bis(triuoromethyl-sulfonyl)imide
[Omim][BTI], [Hmim][BTI], 1-octyl-3-methyl-imidazolium triuoro-
methanesulfonate [Omim][OTF], and 1-butyl-3-methyl-imidazolium
triuoro-methanesulfonate [Bmim][OTF] on the VLE of(hexane +
benzene), (hexane + cyclohexane), (benzene + cyclohexane), and
(ethanol + water) have been studied by Mokhtarani and Gmehling
[8]. Cehreli and Gmehling [9] reported the VLE data for the
mixture of (cyclohexene + benzene) with different types of ILs.
Recently, Matsuda et al . [10] investigated on the inuence of
1-ethyl-3-methylimidazolium ethyl sulfate [Emim][EtSO4] on the
VLE for two binary systems of (1-propanol + water) and (methyl
acetate + methanol). The literature data reveal that adding IL to
the binary systems of hydrocarbons improve the separation factors
in those systems.

0021-9614/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jct.2012.02.032
78 B. Mokhtarani et al. / J. Chem. Thermodynamics 51 (2012) 7781

In the present research, the VLE data for the mixtures of (hex-
ane + benzene), (hexane + cyclohexane), (benzene + cyclohexane),
(1-hexen + benzene), and (hexane + 1-hexene) with 1-methyl 3-
octylimidazolium thiocyanate [Omim][SCN] is investigated by
using headspace gas chromatography. To our knowledge, there is
no report on the experimental VLE data for this type of IL in
literature. The NRTL thermodynamic model was then used for
correlating the experimental data.
2. Experimental
2.1. Chemicals
determined carefully. For this purpose, a mathematical relation
for the volume of the liquid was derived as a function of the height
of the liquid in the vial. The volume of the liquid phase can then be
determined by measurement of the height of liquid phase. Then,
the density of the liquid phase was measured using an Anton Paar
DMA-5000 digital densitometer with an uncertainty of less than
4.105 g  cm3. Afterward, the mass of liquid phase was calcu-
lated, from which the mass of vapor phase was determined by sub-
tracting the total initial mass and the mass of liquid phase. By
using the mass fractions in the vapor phase, the mass fraction of
each component in the vapor and liquid phase were then deter-
mined. The experimental uncertainty for the liquid mole fraction
was 0.002.

The purities of the chemicals and their suppliers are given in

table 1. The IL [Omim][SCN] was prepared in our laboratory from
[Omim][Cl] according to the procedure in literature [11].
[Omim][Cl] was prepared according to the method described in lit-
erature [12] and treated by KSCN in acetone to prepare
[Omim][SCN]. The prepared IL was then puried to remove unre-
acted initial materials. The puried [Omim][SCN] was dried and
degassed for 24 h at 343.15 K under vacuum, and was kept in bot-
tle under argon gas. The mass fraction of water in the dried prod-
uct, measured by a 684 Karl Fischer coulometer, was 7  104.
3. Results and discussion
The VLE data of (hexane + benzene), (hexane + cyclohexane),
(benzene + cyclohexane), (1-hexene + benzene), and (1-hexene +
cyclohexane) systems in the presence of [Omim][SCN] were mea-
sured by headspace gas chromatography. The equilibrium data
based on the solvent-free-basis are reported in tables 2 to 6. The
separation factors for each system were calculated using the
following relation:

K i yi =xi
2.2. Apparatus and procedure aij ; 1
K j yj =xj

The concentrations of vapor contents in the VLE of ternary where Ki is the ratio of mole fraction of component i in the vapor (yi)
systems were determined by headspace gas chromatography. to that in the liquid phase (xi).
The term of headspace refers to the vapor space above the As the results show, the separation factors in all theses systems
liquid sample placed in a vial [6]. The apparatus consists of a are higher than 1, which means that the IL can act as an appropri-
headspace sampler (HS-sampler) from Perkin Elmer Company ate entrainer for these systems.
(turbo matrix 40), a gas chromatograph (Varian, cp 3800), and a
computer where a FID detector and capillary column (Chrompack,
25 m  0.32 mm  1.2 lm) were used for GC analysis. TABLE 2
Experimental equilibrium data of liquid mole fraction (xs1 ) and vapor mole fraction
The ternary mixtures of hydrocarbons and IL were prepared by
(ys1 ) based on solvent free basis and separation factors a12 for the system of {hexane
mass using a laboratory balance with the precision of 104 g. The (1) + benzene (2)} in the presence of 80 mol% [Omim][SCN] at 323.2 K.
mole fraction of the [Omim][SCN] was xed at 0.8 in each mixture.
xs1 ys1 a12
This mole percent was selected because the mixture of the hydro-
carbon and ionic liquid is not homogeneous in mole fraction less 0.048 0.210 5.27
than 0.75, and a two-phase system is observed. The mixtures were 0.109 0.325 3.94
0.145 0.436 4.56
transferred into 20 cm3 vials and the vials were closed tightly with 0.255 0.558 3.68
the cap and septum. The vials were inserted in the oven of head- 0.328 0.634 3.55
space sampler and heated to the specied temperature. After 3 h, 0.437 0.676 2.69
when the equilibrium conditions were attained in the vials, a sam- 0.601 0.762 2.12
0.673 0.836 2.48
ple of the vapor phase was taken by the headspace sampler and
0.800 0.900 2.24
analyzed by the gas chromatography. Since the vapor pressure of 0.872 0.968 4.42
the IL is negligible, the vapor phase consists of hydrocarbons. 0.904 0.996 26.84
Therefore, in order to measure, the compositions of the vapor
phase; calibrations for the binary systems were required. The
samples of the binary hydrocarbons with compositions within
the entire range of concentrations were prepared and injected TABLE 3
directly into the gas chromatograph. The experimental uncertainty Experimental equilibrium data of liquid mole fraction (xs1 ) and vapor mole fraction
(ys1 ) based on solvent free basis and separation factors a12 for the system of {hexane
for the vapor phase mole fraction was less than 0.001.
(1) + cyclohexane (2)} in the presence of 80 mol% [Omim][SCN] at 323.2 K.
In order to determine the composition of the liquid phase, the
volume of the liquid phase at equilibrium conditions was xs1 ys1 a12
0.009 0.091 11.05
0.028 0.170 7.12
TABLE 1 0.100 0.300 3.86
The purities and suppliers of the chemicals. 0.214 0.441 2.90
0.294 0.517 2.57
Chemical name Supplier Mass fraction purity
0.378 0.611 2.58
Hexane Merck >0.99 0.491 0.702 2.45
Benzene Merck >0.99 0.603 0.789 2.46
Cyclo hexane Merck >0.99 0.672 0.856 2.90
1-Hexene Fluka >0.99 0.818 0.948 4.07
[Omim][SCN] Prepared in lab >0.99 0.915 0.985 6.13
 B. Mokhtarani et al. / J. Chem. Thermodynamics 51 (2012) 7781 79

TABLE 4 Dg ij g ij  g jj : 5
Experimental equilibrium data of liquid mole fraction (xs1 ) and vapor mole fraction
(ys1 ) based on solvent free basis and separation factors a12 for the system of In order to correlate the experimental data, the following equa-
{cyclohexane (1) + benzene (2)} in the presence of 80 mol% [Omim][SCN] at 323.2 K. tion is used as an objective function:
xs1 ys1 a12  exp 
Pn y1  ycal
1
0.045 0.105 2.52 F ; 6
i1 yexp
1 i
0.093 0.204 2.50
0.204 0.340 2.01
0.323 0.445 1.68 where n is the number of data points, y is the vapor mole fraction,
0.467 0.576 1.55 and superscripts of exp and cal referred to experimental and calcu-
0.503 0.603 1.50 lated values, respectively.
0.587 0.687 1.54 The experimental data were correlated by the Microsoft Excel
0.710 0.780 1.45
program (Solver). The solver program minimizes the objective
0.851 0.896 1.51
0.940 0.975 2.51 function through changing the interaction parameters of the NRTL
0.943 0.991 6.58 model with trial and error. The method of the solution is based on
the newton method. At the minimum point the nal NRTL interac-
tion parameters are determined. The nonrandomness parameter, a,
TABLE 5 in NRTL model was set to 0.3 for all the systems and the interaction
Experimental equilibrium data of liquid mole fraction (xs1 ) and vapor mole fraction parameters are reported in table 7. The root mean square deviation
(ys1 ) based on solvent free basis and separation factors a12 for the system of {1-hexene
is dened as follows:
(1) + cyclohexane (2)} in the presence of 80 mol% [Omim][SCN] at 323.2 K.
 1=2
xs1 ys1 a12 P
n
exp 2
rmsd yexp
1  ycal
1 =y 1 i =n  1  100: 7
0.026 0.124 5.41 i1
0.054 0.304 7.61
0.101 0.402 6.00 As seen in table 7, the NRTL model can correlate the VLE data
0.179 0.468 4.04 with a good accuracy.
0.295 0.570 3.18 The experimental results and the literature data [1821] for the
0.421 0.666 2.74 hydrocarbon binaries are compared with the correlated data by the
0.546 0.771 2.80
0.654 0.849 2.97
NRTL model on the solvent free basis in gures 1 to 4. As the g-
0.723 0.878 2.76 ures show, adding [Omim][SCN] as the entrainer improves the sep-
0.839 0.948 3.47 aration factor of these systems. The gures also show that the
0.934 0.976 2.80 correlated data by the NRTL model represent good agreements
with the experimental data.
In order to evaluate the performance of [Omim][SCN] as the en-
TABLE 6 trainer for these binary systems, the experimental VLE data and
Experimental equilibrium data of liquid mole fraction (xs1 ) and vapor mole fraction separation factors can be compared with those obtained by using
(ys1 ) based on solvent free basis and separation factors a12 for the system of {1-hexene other ILs reported in a previous work [8]. The comparison of the
(1) + benzene (2)} in the presence of 80 mol% [Omim][SCN] at 323.2 K.
VLE data with those obtained by using [Omim][BTI] for the binary
xs1 ys1 a12 systems of (hexane + benzene), (hexane + cyclohexane), and
0.071 0.201 3.30 (cyclohexane + benzene) are shown in gures 1 to 3. From gure
0.084 0.240 3.44 1, it can be seen that the [Omim][BTI] is a more suitable entrainer
0.142 0.368 3.53
0.341 0.646 3.52
0.369 0.701 4.02 TABLE 7
0.441 0.779 4.47 The interaction parameters (Dg) for NRTL model obtained by correlating VLE data and
0.486 0.821 4.85 the root mean square deviation (rmsd).
0.591 0.889 5.54
0.743 0.943 5.70 Component ij NRTL parameter rmsd
Dgij (kJ  mol1) Dgji (kJ  mol1)
{Hexane (1) + benzene (2) + [Omim][SCN] (3)}
4. Thermodynamic correlation 12 85.108 105.080
13 4.806 11.972 2.28
The experimental VLE data were correlated using the Non 23 16.700 3.327
Random Two Liquid (NRTL) model [13]. This model has been {Hexane (1) + cyclohexane (2) + [Omim][SCN] (3)}
widely used in modeling both (vapor + liquid) and (liquid + liquid) 12 16.372 13.102
equilibria [1417]. In this model, the activity coefcients are calcu- 13 11.986 29.446 1.58
23 5.657 30.038
lated as follows:
Pn   Pn 
s
j1 ji xj Gji P
n xj Gij m1 smi xm Gmi
{Cyclohexane (1) + benzene (2) + [Omim][SCN]} (3)
ln ci Pn Pn sij  P n ; 2 12 1.162 568.904
k1 xk Gki j1 k1 xk Gkj k1 xk Gkj 13 76.105 3.983 2.10
23 4.011 20.951
where
{1-Hexene (1) + cyclohexane (2) + [Omim][SCN]} (3)
Gij expa  sij ; 3 12 3.625 568.904
13 76.049 17.138 1.83
Dg ij 23 2.562 18.425
sij ; 4
RT {1-Hexene (1) + benzene (2) + [Omim][SCN]} (3)
where x is the mole fraction, R is the gas constant, a is the nonran- 12 9.458 3.120
13 4.775 6.072 0.73
domness parameter, and Dg ij is the energy parameter which is
23 11.321 1.999
dened as:
80 B. Mokhtarani et al. / J. Chem. Thermodynamics 51 (2012) 7781

1 1.0

0.8 0.8

0.6 0.6
y1

y1
0.4 0.4

0.2 0.2

0 0.0
0 0. 2 0. 4 0. 6 0. 8 1 0. 0 0. 2 0. 4 0. 6 0. 8 1.0
x1 x1

FIGURE 1. Mole fraction of vapor against mole fraction of liquid (IL-free basis) for FIGURE 4. Mole fraction of vapor against mole fraction of liquid (IL-free basis) for
the ternary system of {hexane (1) + benzene (2) + 80 mol% IL} at T = 323.2 K, () the ternary system of {1-hexene (1) + benzene (2) + 80 mol% [Omim][SCN]} at
ternary system with [Omim][SCN]; (j) binary system without IL [14]; (N) ternary T = 323.2 K, () ternary system with [Omim][SCN]; (j) binary system without IL at
system with [Omim][BTI] at T = 322.8 K [8]; and () NRTL correlation. T = 313.1 K [17]; and () NRTL correlation.

and (cyclohexane + benzene) but for the binary system of (hex-

1.0
ane + cyclohexane), [Omim][SCN] is more suitable. For the binary
systems containing benzene, the interaction between hydrogen
0.8
atoms in benzene having double bonds and oxygen atoms in sulfo-
nyl groups of [BTI] is higher than that with [SCN]. This interaction
0.6
may decrease the interactions between hexane (or cyclohexane)
y1

and benzene, and therefore facilitates the separation of hexane.

0.4
On the other hand, for the binary system of (hexane + cyclohex-
ane), the interaction between the hydrogen atoms of cyclohexane,
0.2
which have single bonds with carbon, and the oxygen atoms in sul-
fonyl groups of [BTI] is not as much as that in the case of benzene.
0.0
0. 0 0. 2 0. 4 0. 6 0. 8 1.0
At the same time, as [SCN] has a smaller molar volume [22], its
interaction with cyclohexane may facilitate the separation of
x1 hexane.
FIGURE 2. Mole fraction of vapor against mole fraction of liquid (IL-free basis) for
the ternary system of {hexane (1) + cyclohexane (2) + 80 mol% IL} at T = 323.2 K, () 5. Conclusions
ternary system with [Omim][SCN]; (j) binary system without IL [15]; (N) ternary
system with [Omim][BTI] at T = 322.8 K [8]; and () NRTL correlation.
VLE data for the systems of (hexane + benzene), (hexane + cyclo-
hexane), (benzene + cyclohexane), (1-hexene + cyclohexane), and
(1-hexene + benzene) with IL [Omim][SCN] were measured using
1.0
headspace gas chromatography method at 323.15 K. The compari-
son of VLE experimental results with those of the binary systems,
0.8
shows signicant enhancement in separation factor of these sys-
tems due to presence of the IL as an entrainer. The experimental
0.6
data were correlated by the NRTL model with a good accuracy
y1

0.4
0.2
0.0
0. 0 0. 2 0. 4 0. 6 0. 8
x1
FIGURE 3. Mole fraction of vapor against mole fraction of liquid (IL-free basis) for
the ternary system of {cyclohexane (1) + benzene (2) + 80 mol% IL} at T = 323.2 K,
() ternary system with [Omim][SCN]; (j) binary system without IL [16]; (N)
ternary system with [Omim][BTI] at T = 332.5 K [8]; and () NRTL correlation.
for the separation of (hexane + benzene). However, the separation
factors of [Omim][SCN] at the azeotropic point is higher than
[Omim][BTI]. As gure 2 shows, [Omim][SCN] is an appropriate en-
trainer for the hexane + cyclohexane system where the separation
factors for this IL are higher than those for [Omim][BTI]. For the
(cyclohexane + benzene) system (gure 3), [Omim][BTI] is a more
suitable entrainer. This comparison shows, the [Omim][BTI] is a
suitable entrainer for the binary systems of (hexane + benzene)
and compared with the VLE data of the systems with [Omim][BTI]
as the entrainer. The comparison indicates that both ILs are appro-
priate entrainer for binary systems of hydrocarbons but [Omim][B-
TI] is a more suitable entrainer for (hexane + benzene) and
(cyclohexane + benzene) systems. For the binary system of (hex-
ane + cyclohexane), the inuence of [Omim][SCN] on the VLE is
1.0
more than that of [Omim][BTI]. It seems that more experimental
data are required for the inuence of the different ILs on the VLE
of the hydrocarbon systems over the entire concentration range.
References
[1] B. Mokhtarani, A. Shari, H.R. Mortaheb, M. Mirzaei, M. Ma, F. Sadeghian, J.
Chem. Thermodyn. 41 (2009) 323329.
[2] B. Mokhtarani, A. Shari, H.R. Mortaheb, M. Mirzaei, M. Ma, F. Sadeghian, J.
Chem. Eng. Data 55 (2010) 39013908.
[3] R. Kato, J. Gmehling, Fluid Phase Equilib. 226 (2004) 3744.
[4] S. Nebig, V. Liebert, J. Gmehling, Fluid Phase Equilib. 277 (2009) 6167.
[5] S. Nebig, R. Blts, J. Gmehling, Fluid Phase Equilib. 258 (2007) 168178.
[6] V. Liebert, S. Nebig, J. Gmehling, Fluid Phase Equilib. 268 (2008) 1420.
[7] A. Westerholt, V. Liebert, J. Gmehling, Fluid Phase Equilib. 280 (2009) 5660.
[8] B. Mokhtarani, J. Gmehling, J. Chem. Thermodyn. 42 (2010) 10361038.
[9] S. Cehreli, J. Gmehling, Fluid Phase Equilib. 295 (2010) 125129.
[10] H. Matsuda, K. Tochigi, V. Liebert, J. Gmehling, Fluid Phase Equilib. 307 (2011)
197201.
 B. Mokhtarani et al. / J. Chem. Thermodynamics 51 (2012) 7781
[11] H.Y. Huang, H.M. Wang, R.S. Hou, H.T. Cheng, L.C. Chen, J. Chin. Chem. Soc. 55
(2008) 12041207.
[12] J.G. Huddleston, A.E. Visser, W.M. Reichert, H.D. Willauer, G.A. Broker, R.D.
Rogers, Green Chem. 3 (2001) 156164.
[13] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135144.
[14] E.J. Gonzlez, N. Calvar, E. Gomez, A. Domnguez, J. Chem. Thermodyn. 42
(2010) 104109.
[15] E.J. Gonzlez, N. Calvar, B. Gonzlez, A. Domnguez, J. Chem. Eng. Data 55
(2010) 633638.
[16] E.J. Gonzlez, N. Calvar, B. Gonzlez, A. Domnguez, J. Chem. Thermodyn. 42
(2010) 752757.
81
[17] T.G. Mark, A. Jongmans, B. Boelo Schuur, A.B. de Haan, J. Chem. Thermodyn. 47
(2012) 234240.
[18] I.P.C. Li, Y.W. Wong, S. Chang, B.C.Y. Lu, J. Chem. Eng. Data 17 (1972) 492498.
[19] J.B. Ott, K.N. Marsh, R.H. Stokes, J. Chem. Thermodyn. 12 (1980) 11391148.
[20] K. Kurihara, M. Uchiyama, K. Kojima, J. Chem. Eng. Data 42 (1997) 149154.
[21] T.M. Lesteva, G.I. Logunova, V.I. Chernaya, Zh. Fiz. Khim. 53 (1979) 11801183.
[22] A.B. Pereiro, J.M.M. Arajo, J.M.S.S. Esperana, I.M. Marrucho, L.P.N. Rebelo, J.
Chem. Thermodyn. 46 (2012) 228.
JCT 11-562