ISSN 1810-2328, Journal of Engineering Thermophysics, 2010, Vol. 19, No. 3, pp. 170–183.


c Pleiades Publishing, Ltd., 2010.

Activity Coefficient Models to Describe Isothermal Vapor–Liquid

Equilibrium of Binary Systems Containing Ionic Liquids
J. A. Lazzús* and J. Marı́n

Departamento de Fı́sica, Universidad de La Serena, Casilla 554, La Serena, Chile

Received January 19, 2010

Abstract—Isothermal vapor–liquid equilibrium data of binary mixtures containing ionic liquids are
correlated using three activity coefficient models: UNIQUAC, Wilson, and NRTL. Twenty binary
systems taken from the literature were selected for this study. A genetic algorithm is used to
determine the interaction parameters for the three models. The results given by the three models
have been compared with experimental data, and show that the UNIQUAC model is the best method
to correlate and predict the vapor–liquid equilibrium of this type of systems.

DOI: 10.1134/S1810232810030070

1. INTRODUCTION

In recent years, room temperature ionic liquids (RTILs) or just ionic liquids (ILs) have come into
focus because of their potential as alternatives for several engineering applications [1]. The ionic liquids
are typically composed of a large organic cation and an inorganic polyatomic anion. There is virtually no
limit in the number of possible ionic liquids since there is a large number of cations and anions that can
be combined [2]. At ambient room temperature, they exist as liquids and have a wide variety of unique
properties (for instance, a negligible vapor pressure, favorable salvation behavior, low viscosity, and high
reactivity and selectivity) [3].

The increasing utilization of ILs in chemical and industrial processes requires reliable and systematic
thermophysical properties such as activity coefficients, heats of mixing, densities, solubilities, vapor–
liquid equilibria (VLE), and liquid–liquid equilibria (LLE). In addition, the transport properties are also
needed (viscosity, electric conductivity, mutual diffusion coefficients, etc.) [4]. For better understanding
of their thermodynamic behavior and for development of thermodynamic models, reliable experimental
phase equilibrium data are required [5]. Nevertheless, the number of published VLE data sets is still very
limited. But, in particular, reliable VLE data would allow one to check whether gE -models can be used
for the description of the real behavior of binary systems with ionic liquids and whether binary gE -model
parameters can be applied for the prediction of the VLE behavior of higher systems containing ionic
liquids [6].

In this work, isothermal vapor–liquid equilibrium data of binary mixtures containing ionic liquids
were correlated using three activity coefficient models. Twenty binary systems taken from the literature
were selected for this study. A genetic algorithm was used to determine the interaction parameters for
the three models. The results for the three models were compared, and the best model is recommended
to describe the VLE for mixtures with ILs.
*
E-mail: jlazzus@dfuls.cl

170
 ACTIVITY COEFFICIENT MODELS 171

2. EQUATIONS OF VAPOR–LIQUID EQUILIBRIUM

As known, the phase equilibrium problem to be solved consists in calculating some variables of the
set T − P − x − y (temperature, pressure, liquid–phase concentration and vapor-phase concentration,
respectively), when some of them are known. For a vapor–liquid mixture in thermodynamic equilibrium,
the temperature and the pressure are the same in both phases, and the remaining variables are defined
by the material balance and the “fundamental equation of phase equilibrium.” The application of this
fundamental equation requires the use of thermodynamic models that normally include binary interaction
parameters [7].
Classical thermodynamic models commonly used in the literature to treat these mixtures at a low
pressure required a great amount of binary parameters to be determined from experimental data [8].
These binary parameters must be determined using experimental data for binary systems. Theoretically,
once these binary parameters are known, one could predict the behavior of multicomponent mixtures
using standard thermodynamic relations and thermodynamics models [7].
The fundamental equation of vapor–liquid equilibrium can be expressed as the equality of fugacities
of each component in the mixture in both phases [8]:

f¯iL = f¯iV . (1)

The fugacity of a component in the vapor phase is usually expressed through the fugacity coefficient ϕ̄Vi :

f¯iV = yi ϕ̄Vi P. (2)

The fugacity of a component in the liquid phase is expressed through either the fugacity coefficient ϕ̄L
i
or the activity coefficient γi :

f¯iL = xi ϕ̄L
i P, (3)

f¯iL = xi γi fi0 . (4)

In these equations, yi is the mole fraction of component in the vapor phase, xi is the mole fraction of
component in the liquid phase, and P is the pressure. The fugacity is related to the temperature, the
pressure, the volume, and the concentration though a standard thermodynamic relation [9].
If the fugacity coefficient is used in both phases, the method of solution of the phase equilibrium
problem is known as an equation of state method. Then, equation of state (EoS) and a set of mixing
rules are needed, to express the fugacity coefficient as a function of the temperature, pressure and
concentration [7]. Modern equations of state methods include an excess Gibbs free energy model (gE ) in
the mixing rules of the equation of state, giving origin to the so-called equation of state + gE model [10].
This means that an activity coefficient model (γ) is used to describe the complex liquid phase, and the
fugacity coefficient (ϕ) is calculated using a simple equation of state. If the fugacity coefficient is used for
the vapor phase and the activity coefficient is used for the liquid phase, the equilibrium problem is known
as a gamma–phi method. This method has given acceptable results for some systems [7, 8].
Most models available in literature for the activity coefficient are of the correlating type (Van Laar,
Margules, Redlich-Kister, NRTL, UNIQUAC and Wilson), meaning that experimental data are needed
to calculate certain empirical parameters; although some predictive models are also available (UNIFAC,
ASOG). These predictive models have been used in several applications, and recently including mixtures
with some ILs [5, 6, 11]. But to the best of authors’ knowledge, there is no systematic studies on binary
mixtures, such as the one presented here, which have included several compounds and ILs.
In this study, three activity coefficient models most commonly used (UNIQUAC, Wilson, and
NRTL), are analyzed for binary systems containing ILs. For all the models, the interaction parameters
(Aij for Wilson, Bij and αij for NRTL, and Uij for UNIQUAC) are calculated using VLE data.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

172 LAZZÚS, MARÍN

3. EQUATION OF STATE METHOD

From the relation between the fugacity, the Gibbs free energy, and an EoS, the fugacity coefficient
can be calculated as [8]:
V    
1 RT ∂P
ln ϕ̄i = − dV − ln Z. (5)
RT V ∂Ni T,V,Nj=i
∞

Thus, the experimental vapor–liquid phase data can be correlated with the Peng–Robinson (PR) EoS,
and the Wong and Sandler (WS) mixing rule [7].
The PR EoS can be expressed as follows:
RT a
P = + , (6)
V − b V (V + b) + b(V − b)

R2 Tc2
a = 0.457235 α(Tr ), (7)
Pc

RTc
b = 0.077796 , (8)
Pc
  2
α(Tr ) = 1 + κ(1 − Tr ) , (9)

κ = 0.37646 + 1.54226ω − 0.26992ω 2 ; (10)

and for mixtures:

RT am
P = + . (11)
V − bm V (V + bm ) + bm (V − bm )

The WS mixing rules used in this EoS can be summarized as follows [7]:

N

N  a 
xi xj b −
i j RT ij
bm = , (12)

xi ai
N AE
∞ (x)
1− −
i bi RT ΩRT
 √
a  1 ai aj
b− = (bi + bj ) − (1 − kij ), (13)
RT ij 2 RT
N 
 xi ai AE (x)
am = bm + ∞ . (14)
bi Ω
i

In these equations am and bm are the EoS constants, kij is the adjustable parameter with Ω = 0.34657,
and AE ∞ (x), the excess Helmholtz free energy at the limit of infinite pressure is calculated using an
activity coefficient model and assuming that AE∞ ≈ A0 ≈ g0 [10].
E E

4. ACTIVITY COEFFIENT MODELS

The main equations for the three models studied (UNIQUAC, Wilson, and NRTL) and the meaning
of the different variables and parameters are described in what follows:

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 173

4.1. UNIQUAC Model

The universal-quasi-chemical theory [12], from which the UNIQUAC model is derived, can be
expressed in terms of the gE as:
giE = gi,C
E E
+ gi,R (15)
E and residual part g E given by:
with the combinatorial part gC R
E
gC  Φi z  θi
= xi ln + qi xi ln , (16)
RT xi 2 Φi
i i

⎛ ⎞
E
gR  
=− qi xi ln ⎝ θj τji ⎠. (17)
RT
i j

The activity coefficient expression from this model is [7]:

ϕi z θi ϕi 
ln γiC = ln + qi ln + li − xj lj , (18)
xi 2 ϕi xi
j

⎡ ⎛ ⎞ ⎤
  θi τij
ln γiR = qi ⎣1 − ln ⎝ θj τji ⎠ −
⎦. (19)
θk τkj
j j
k

The volume (ϕi ) and surface (θi ) fractions are detailed as:
ri xi
ϕi =
, (20)
rj xj
j

qi xi
θi =
. (21)
rj xj
j

The required ri and qi parameters for UNIQUAC are calculated according to Bondi [13], and the
equation below must be satisfied with the coordination number (z) equal to 10:
z
li = (ri − qi ) − (ri − 1). (22)
2
The adjustable parameter τji is detailed as:
 
Aij − (Uij − Uji )
τji = exp (23)
RT

with
bij (Uij − Uji )
aij + = Aij + , aij = Aij , (24)
T RT

(Uij − Uji )
bij = − , (25)
R
where Uij and Uji represent the interaction energy difference between i − j and j − j pair molecules and
between j − i and i − i pair molecules, respectively, R is the universal constant of gases, and T is the
temperature.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

174 LAZZÚS, MARÍN

4.2. Wilson Model

Wilson [14] presented the following expression for the gE :

gE
= −xi ln(xi + Aij xj ) − xj ln(xj + Aji xi ). (26)
RT

Based on this equation, we obtained the expression for the liquid phase activity coefficients as [7]:
⎛ ⎞
N 
N
xk Aki
ln γi = − ln ⎝ xj Aij ⎠ + 1 − . (27)

N
j=1 k xj Akj
j=1

In this expression,
   
Vj (λij − λii )
Aij = exp − , (28)
Vi RT

where Vi is the liquid molar volume of the component i, λij − λii is the empirical energy term, and xi
is the mole fraction of the component i. The parameters: Aij and Aji , represent the interaction energy
difference between i − j and j − j pair molecules and between j − i and i − i pair molecules, respectively.

4.3. NRTL Model

In the non-random-two-liquid model (NRTL), the Gibbs free energy for a binary mixture is given
as [7]:
⎛ N ⎞

⎜ Gji τji xj ⎟
gE N
⎜ j ⎟
= xi ⎜ln N ⎟; (29)
RT ⎝
⎠
i Gki xk
k

and the expression of activity coefficient for this model has the following form [15]:
⎡ ⎤

N
τji Gji xi xj Gji ⎢ τki Gkj xk ⎥
j 
N
j ⎢ k ⎥
ln γi = N + ⎢τij − N ⎥ (30)

N ⎣
⎦
Gki xk j Gkj xk Gkj xk
k k k

with
τji = Bji /T, (31)

Gji = exp(−αji τji ), (32)

αji = αij , (33)

where Bij and Bji represent the interaction energy difference between i − j and j − j pair molecules
and between j − i and i − i pair molecules, respectively; R is the universal constant of gases, and T
is the temperature. The terms Bij , Bij , and αij are the adjustable parameters usually calculated from
experimental VLE data.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 175

Table 1. Nomenclature and structure of the ionic liquids involved in this study

Ionic liquid Abbrevation Chemical structure

1,3-Dimethylimidazolium [dmim] [DMPO4]

Dimethylphosphate

1-Butyl-4-methylpyridinium [bmpy] [TOS]

Tosylate

1-Ethyl-3-methylimidazolium [emim] [Tf2N]

Bis(trifluoromethylsulfonyl)imide

1-Butyl-3-methylimidazolium [bmim] [Tf2N]

Bis(trifluoromethylsulfonyl)imide

1-Hexyl-3-methylimidazolium [hmim] [Tf2N]

Bis(trifluoromethylsulfonyl)imide

1-Octyl-3-methylimidazolium [omim] [Tf2N]

Bis(trifluoromethylsulfonyl)imide

5. BINARY SYSTEMS USED

Table 1 shows the ILs involved in this study. Note, that these ILs have been an object of increas-
ing attention due to their unique physico-chemical properties such as high thermal stability, large
liquidus range, high ionic conductivity, high solvating capacity, negligible vapor pressure, and non-
flammability [16].

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

176 LAZZÚS, MARÍN

Table 2. Properties of the substances used in this work

Substance M Tc (K) Pc (bar) Vc (cm3 mol−1 ) ω ri qi

Water 18.0 647.1 220.6 56.0 0.345 0.920 1.400
Ethanol 46.1 514.0 61.4 168.0 0.644 2.105 1.972
Acetone 58.1 508.2 47.0 209.0 0.307 2.574 2.336
1-Propanol 60.1 536.8 51.7 218.0 0.620 2.780 2.513
2-Propanol 60.1 508.3 47.6 222.0 0.667 2.779 2.508
Tetrahydrofuran 72.1 540.2 51.9 224.0 0.225 2.941 2.720
1-Butanol 74.1 563.0 44.1 274.0 0.589 3.454 3.052
Hexane 86.2 507.6 30.3 371.0 0.301 4.500 3.856
Toluene 92.1 591.8 41.1 316.0 0.264 3.923 2.968
Octane 114.2 568.7 24.9 486.0 0.400 5.848 4.936
Nonane 128.3 594.6 22.9 544.0 0.443 6.523 5.476
[dmim] [DMPO4] 222.0 880.4 28.6 598.4 0.507 7.162 5.844
[bmpy] [TOS] 321.4 1212.5 25.1 965.7 0.599 8.580 7.741
[emim] [Tf2N] 391.3 1244.9 32.6 892.9 0.182 9.890 8.780
[bmim] [Tf2N] 419.4 1265.0 27.6 1007.1 0.266 11.160 10.200
[hmim] [Tf2N] 447.4 1287.3 23.9 1121.3 0.354 11.565 10.399
[omim] [Tf2N] 475.4 1311.9 21.0 1235.6 0.445 12.229 10.990

Table 2 shows the thermodynamic properties for the ILs [17,18] and for the other compounds [19]
used. In the table, M is the molecular weight, Tc is the critical temperature, Pc is the critical pressure,
Vc is the critical volume, and ω is the acentric factor. The required Van der Waals parameters (ri and qi )
for the UNIQUAC model were estimated according to Bondi [13].
Twenty isothermal (P − x) VLE systems containing ILs were considered in this study. Phase
equilibrium data of these mixtures are necessary for further development of some physico-chemical
processes involving ILs. These systems are of interest for the conception of industrial operations as
environmental, chemical industries, among others. The experimental VLE data taken from the literature
for the ionic liquid systems [5, 6, 11] are described in Table 3. As seen in this table, the temperature and
pressure ranges are narrow and go from 353 to 373 K and from 1 to 227 kPa, respectively.

6. OPTIMIZATION METHOD AND PARAMETERS ESTIMATION

Among the global stochastic optimization techniques, the evolutionary algorithms known as genetic
algorithms (GAs) have found many applications in several fields in science and engineering [20]. The
genetic algorithm was first developed by Holland [21] and was based on the mechanics of natural
selection in biological systems. It uses a structure to utilize genetic information in finding new directions
of search. The major genetic operators that reflect nature’s evolutionary process are reproduction,
crossover, and mutation [22].
The GA maintains a population of individuals, whose characteristics are encoded in a fixed-length bit
string, modelling the biological genotype. It is at the discretion of the programmer the way in which these
bits represent the phenotype (ontogeny). As a parallel to nature, genetic material is swapped between
the individuals and mutated to produce offspring, with corresponding changes in their phenotypic
performance. The crossover operator is an analog of the recombination of genetic material, as observed in
the reproduction. Crossover involves splitting the genomic bit-strings of two parents at a given number
of locations and then splicing together complementary sections of each parent’s bit-string to form the

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 177

Table 3. Details of the vapor–liquid equilibrium data considered in this study

System (1) + (2) VLE data

Ref.
Comp. (1) Comp. (2) ND T (K) ΔP (kPa) Δx1
Acetone [emim] [Tf2N] 29 353.15 1–215 0–1 [5]
2-Propanol [emim] [Tf2N] 34 353.15 1–92 0–1 [5]
Tetrahydrofuran [emim] [Tf2N] 37 353.15 2–155 0–1 [5]
Water [emim] [Tf2N] 26 353.15 3–47 0–1 [5]
Acetone [bmim] [Tf2N] 31 353.15 4–215 0–1 [5]
2-Propanol [bmim] [Tf2N] 32 353.15 4–92 0–1 [5]
Water [bmim] [Tf2N] 20 353.15 5–47 0–1 [5]
Acetone [dmim] [DMPO4] 32 353.15 10–217 0–1 [5]
Ethanol [dmim] [DMPO4] 33 353.15 1–109 0–1 [5]
Tetrahydrofuran [dmim] [DMPO4] 33 353.15 9–158 0–1 [5]
Water [dmim] [DMPO4] 29 353.15 1–48 0–1 [5]
Octane [bmim] [Tf2N] 6 353.15 3–23 0–1 [6]
Nonane [bmim] [Tf2N] 5 353.15 1–10 0–1 [6]
Toluene [bmim] [Tf2N] 14 363.15 1–50 0–0.6 [6]
Hexane [hmim] [Tf2N] 7 353.15 9–133 0–0.2 [6]
Water [hmim] [Tf2N] 5 353.15 12–43 0–0.4 [6]
Hexane [omim] [Tf2N] 8 353.15 9–140 0–0.3 [6]
1-Butanol [bmpy] [TOS] 10 373.15 35–52 0–1 [11]
Ethanol [bmpy] [TOS] 11 373.15 49–227 0–1 [11]
1-Propanol [bmpy] [TOS] 11 373.15 53–113 0–1 [11]

genotype of the new individual. Crossover occurs with a random probability. The mutation operator
simulates natural mutation of DNA. This simply involves flipping bits in the string in a stochastic
manner. Mutation should be fairly infrequent and should be applied following crossover [20].
The most significant differences between GA and more traditional search and optimization methods,
are: (i) GAs search a population of points in parallel, not a single point: (ii) GAs do not require derivative
information or other auxiliary knowledge; only the objective function and the corresponding fitness levels
influence the directions of search; (iii) GAs use probabilistic transition rules, not deterministic ones; and
(iv) GA work on an encoding environment of the parameter set rather than the parameter set itself [22].
The interaction parameters: Aij for Wilson, Bij and αij for NRTL, and Uij for UNIQUAC, were
determined using a GA optimization programmed in MATLAB version 6.5.0 [23]. The procedure was
based on the minimization of the overall objective function (OF):
ND  
100   Pcalc − Pexp 
  ,
OF = |%ΔP | =   (34)
ND Pexp i
i=1

where ND is the number of data point, P is the system pressure, and the subscripts exp and calc refer to
the experimental and calculated values.
An exhaustive trial-and-error procedure was applied for tuning the GA parameters. Table 4 shows
the selected parameters for the GA optimization. Figure 1 shows a flow diagram of the GA optimization
used for the vapor–liquid equilibrium modelling.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

178 LAZZÚS, MARÍN

Fig. 1. Flow diagram of the GA used for the vapor–liquid equilibrium modelling.

7. RESULTS AND DISCUSSION

In order to provide a substantial margin of safety, the range for the interaction parameters: Aij
(Wilson), Bij (NRTL), and Uij (UNIQUAC), was defined as [−1000, 2500] cal·mol−1 . This wide range
was based on physical considerations [24], and is extremely likely that it will contain the optimal
parameter values. The range for αij with theoretical bases [25] was defined as [0.2, 0.4].
Figure 2 shows the interaction parameters: Aij for Wilson, Bij for NRTL and Uij for UNIQUAC,
determined with the GA optimization and based on the minimization of Eq. (34).
Table 5 shows the calculated parameter for the activity coefficient models considered (UNIQUAC,
Wilson, and NRTL). This table shows the |%ΔP | deviations obtained for Eq. (34). The end of the table
shows the average deviation obtained by the three models. The results found using the UNIQUAC model

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 179

Table 4. Parameters used in the genetic algorithm

GA Parameters Value
Number of generations (Gen) 50
Number of individuals (Ni ) 25
Length of chromosome (L) 20
Length of individuals (Li ) 80
Crossover probability (Cros) 0.8
Mutation probability (Mut) 0.1
Crossover operator Multipoint
Mutation operator Binary
Objective function Eq. (34)

Fig. 2. Average deviations obtained for the interaction parameters: Uij for UNIQUAC (), Aij for Wilson (Δ), and
Bij for NRTL (), determined using GA optimization.

with the GA optimization show that the VLE of binary systems containing ILs can be correlated with
low deviations between experimental and calculated values. These results indicate that the UNIQUAC
model is reliable enough to estimate VLE for the twenty binary systems containing ILs used in this work
(see Table 3), with |%ΔP | deviations a bit higher than 5%. Although one model gives better results for
some particular cases, it is not possible to generalize the results and select one of the models used as the
best one for this type of complex mixtures containing ILs. If better results are desired, further studies are
needed with data restricted to narrow ranges of temperature, pressure, and concentration.
Figure 3 shows an example of the accuracy of the three models to describe VLE of binary systems
containing ILs. The figure shows experimental and correlated VLE of acetone (1), 2-propanol (1),
tetrahydrofuran (1), and water (1) in [emim] [Tf2N] (2). As is observed in the figure, the best method to

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

180 LAZZÚS, MARÍN

Table 5. Parameters (Uij , Aij , Bij , and αij ) for the models UNIQUAC, Wilson and NRTL calculated with GA
optimization from experimental VLE data

UNIQUAC WILSON NRTL

Binary system
U12 U21 |%ΔP | A12 A21 |%ΔP | B12 B21 α12 |%ΔP |

[emim] [Tf2N] (2) +

Acetone (1) 207.54 −257.92 3.6 −641.29 1911.42 5.4 601.29 −768.79 0.205 5.3

2-Propanol (1) 179.31 102.37 4.9 227.04 2460.10 6.6 911.51 −203.39 0.398 5.6

Tetrahydrofuran (1) 403.34 −216.26 4.4 −177.76 1649.33 5.3 −48.60 36.60 0.368 4.7

Water (1) −135.95 887.57 3.9 1661.02 2409.81 5.1 1436.82 −185.16 0.273 5.9

[bmim] [Tf2N] (2) +

Acetone (1) 288.98 −314.52 2.9 −409.59 −208.34 4.5 422.33 −621.61 0.397 5.5

2-Propanol (1) 280.44 −293.90 3.8 −27.85 2403.20 4.2 1314.90 −617.66 0.204 4.6

Water (1) −142.93 1004.70 6.3 1609.80 2494.52 6.6 1508.50 −233.77 0.251 5.1

Octane (1) 95.43 353.02 5.2 1658.50 2343.80 7.4 1561.71 438.42 0.204 5.9

Nonane (1) 924.53 −75.32 4.1 1905.31 2498.71 6.9 674.48 788.17 0.218 5.2

Toluene (1) 927.19 −417.51 5.6 −74.49 2488.43 6.8 1567.82 −557.60 0.211 5.1

[dmim] [DMPO4] (2) +

Acetone (1) 1548.33 −504.98 6.9 304.00 2494.34 6.7 2145.32 −684.02 0.202 5.8

Ethanol (1) −104.54 −508.85 5.2 −960.02 2485.00 6.7 316.38 −986.76 0.395 4.5

Tetrahydrofuran (1) 538.41 −45.77 5.8 971.42 2495.41 6.1 1568.01 −267.10 0.204 5.6

Water (1) −621.56 −799.31 6.8 −740.11 −999.27 5.9 −311.95 −998.89 0.396 5.4

[hmim] [Tf2N] (2) +

Hexane (1) 235.60 187.34 5.7 932.61 2490.72 7.2 2247.00 10.19 0.201 6.4

Water (1) 56.49 815.46 6.8 1772.33 2383.91 7.5 −196.01 769.39 0.293 6.1

[omim] [Tf2N] (2) +

Hexane (1) 325.89 68.99 4.9 694.65 2498.62 7.1 2075.03 91.64 0.239 6.6

[bmpy] [TOS] (2) +

1-Butanol (1) 1201.01 −672.67 4.9 428.14 −947.57 6.9 −636.55 676.05 0.374 5.8

Ethanol (1) −673.68 863.52 5.8 −896.82 2481.83 6.7 713.24 −846.97 0.358 6.6

1-Propanol (1) −579.99 616.78 5.5 −382.80 −975.75 6.9 −870.88 −379.23 0.393 6.2

Average deviation 5.2 6.3 5.6

estimate VLE of the systems used is the UNIQUAC model. This figure ratifies the discussion presented
above.

8. CONCLUSIONS
In this work, three activity coefficient models (UNIQUAC, Wilson, and NRTL) were used to describe
the isothermal vapor–liquid equilibrium of twenty binary mixtures containing ionic liquids. Based on the
results and discussion presented in this study, the following main conclusions were obtained:
(i) the UNIQUAC model is appropriate to modeling VLE for binary systems that contain the ionic
liquids presented above;
(ii) although one model gives better results for some particular cases, it is not possible to generalize
the results and select one of the models used as the best one for this type of complex mixtures
containing ILs.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 181

Fig. 3. Experimental vapor–liquid equilibrium of the systems [emim] [Tf2N] with: acetone (—), 2-propanol (– –),
tetrahydrofuran (- - -), water (· – ·), and correlated values of: () UNIQUAC, (Δ) Wilson, and () NRTL.

ACKNOWLEDGMENTS
The first author thanks the Direction of Research of the University of La Serena and the Department
of Physics of the University of La Serena for the special support that made possible the preparation of
this paper.

NOTATIONS
Symbols
Aij —Wilson parameter
Bij —NRTL parameter
M —molecular weight
ND —number of points in a data set
P —pressure
Pc —critical pressure
qi —Van der Waals area parameter
R—universal gas constant
ri —Van der Waals volume parameter
T —system temperature
Tc —critical temperature
Vc —critical volume
xi —mole fraction in the liquid phase
yi —mole fraction in the vapor phase
Abbreviations
GA—genetic algorithm
LLE—liquid-liquid equilibrium
OF—objective function
VLE—vapor-liquid equilibrium

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

182 LAZZÚS, MARÍN

Greek letters
α—NRTL parameter
ϕ—fugacity coefficient
γ—activity coefficient
ω—acentric factor

Super/subscripts
calc—calculated
exp—experimental
i, j—component
L—liquid
V —vapor

REFERENCES
1. Lazzús, J.A., Estimation of Density as a Function of Temperature and Pressure for Imidazolium-Based Ionic
Liquids Using a Multilayer Net with Particle Swarm Optimization, Int. J. Therm., 2009, vol. 30, pp. 883–909.
2. Yang, J., Peng, C., Liu, H., and Hu, Y., Calculation of Vapor–Liquid and Liquid–Liquid Phase Equilibria for
Systems Containing Ionic Liquids Using a Lattice Model, Ind. Eng. Chem. Res., 2006, vol. 45, pp. 6811–
6817.
3. Heintz, A., Recent Developments in Thermodynamics and Thermophysics of Non-Aqueous Mixtures Con-
taining Ionic Liquids: A Review, J. Chem. Thermodyn., 2005, vol. 35, pp. 525–535.
4. Wang, T., Peng, C., Liu, H., Hu, Y., and Jiang, J., Equation of State for the Vapor–Liquid Equilibria of Binary
Systems Containing Imidazolium-Based Ionic Liquids, Ind. Eng. Chem. Res., 2007, vol. 46, pp. 4323–4329.
5. Kato, R. and Gmehling, J., Measurement and Correlation of Vapor–Liquid Equilibria of Binary Systems
Containing the Ionic Liquids [EMIM] [(CF3 SO2 )2 N], [BMIM] [(CF3 SO2 )2 N], [MMIM] [(CH3 )2 PO4 ] and
Oxygenated Organic Compounds Respectively Water, Fluid Phase Equilib., 2005, vol. 231, pp. 38–43.
6. Nebig, S., Bölts, R., and Gmehling, J., Measurement of Vapor–Liquid Equilibria (VLE) and Excess
Enthalpies (H E ) of Binary Systems with 1-Alkyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide and
Prediction of These Properties and γ ∞ Using Modified UNIFAC (Dortmund), Fluid Phase Equilib., 2007,
vol. 258, pp. 178–188.
7. Orbey, H. and Sandler, S.I., Modeling Vapor–Liquid Equilibria. Cubic Equations of State and Their
Mixing Rules, USA: Cambridge Univ. Press, 1998.
8. Walas, S.M., Phase Equilibria in Chemical Engineering, Storeham: Butterworth Publ., 1985.
9. Gmehling, J., Onken, U., and Arlt, W., Vapor–Liquid Equilibrium Data Collection. DECHEMA, Frank-
furt: Verlag+Druckerei Friedrich Bischoff, 1982.
10. Lazzús, J.A., Prediction of Sublimation Pressures from SCO2 +Hydrocarbon Systems Using a Particle
Swarm Optimization, J. Eng. Therm., 2009, vol. 18, pp. 306–314.
11. Domańska, U., Królikowski, M., and Paduszyński, K., Phase Equilibria Study of the Binary Systems (N-
Butyl-3-Methylpyridinium Tosylate Ionic Liquid+an Alcohol), J. Chem. Thermodyn., 2009, vol. 41, pp. 932–
938.
12. Abrams, D.S. and Prausnitz, J.M., Statical Thermodynamics of Liquid Mixtures: A New Expression for the
Excess Gibbs Energy of Partly or Completely Miscible Systems, AIChE J., 1975, vol. 21, pp. 116–128.
13. Bondi, A., Physical Properties of Molecular Crystals, Liquids and Glasses, New York: Wiley, 1968.
14. Wilson, G.M., Vapour–Liquid Equilibrium. XI. A New Expression for the Excess Free Energy of Mixing, J.
Am. Chem. Soc., 1964, vol. 86, pp. 127–130.
15. Prausnitz, J.M., Lichtenthaler, R.N., and Gomes de Azevedo, E., Molecular Thermodynamics of Fluid-
Phase Equilibria, New Jersey: Prentice Hall Int. Ser., 1999.
16. Lazzús, J.A., ρ − T − P Prediction for Ionic Liquids Using Neural Networks, J. Taiwan Inst. Chem. Eng.,
2009, vol. 40, pp. 213–232.
17. Valderrama, J.O. and Robles, P.A., Critical Properties, Normal Boiling Temperatures, and Acentric Factors
of Fifty Ionic Liquids, Ind. Eng. Chem. Res., 2007, vol. 46, pp. 1338–1344.
18. Valderrama, J.O., Sanga, W.W., and Lazzús, J.A., Critical Properties, Normal Boiling Temperatures, and
Acentric Factor of Another 200 Ionic Liquids, Ind. Eng. Chem. Res., 2008, vol. 47, pp. 1318–1330.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010

 ACTIVITY COEFFICIENT MODELS 183

19. Daubert, T.E., Danner, R.P., Sibul, H.M., and Stebbins, C.C., Physical and Thermodynamic Properties of
Pure Chemicals. Data Compilation, London: Taylor & Francis, 2000.
20. Davis, L., Handbook of Genetic Algorithms, New York: Van Nostrand, 1991.
21. Holland, J., Adaptation in Natural and Artificial Systems, USA: Univ. of Michigan Press, 1975.
22. Kim, K.W., Yun, Y., Yoon, J., Gen, M., and Yamazaki, G., Hybrid Genetic Algorithm with Adaptative Abilities
for Resource-Constrained Multiple Project Scheduling, Comput. Ind., 2005, vol. 56, pp. 143–160.
23. Gopi, E.S., Algorithm Collections for Digital Signal Processing Applications Using Matlab, The
Netherlands: Springer, 2007.
24. Chemstations, ChemCAD: Process Flowsheet Simulator. Operating Manual, Houston: Chemstations,
2001.
25. Tester, J. and Modell, M., Thermodynamics and Its Applications, New York: Prentice-Hall, 2000.

JOURNAL OF ENGINEERING THERMOPHYSICS Vol. 19 No. 3 2010