Fluid Phase Equilibria 221 (2004) 197205

Prediction of liquidliquid interfacial tension in

multi-component systems
Alireza Bahramian
a
, Ali Danesh
b,
a
Institute of Petroleum Engineering, University of Tehran, Tehran, Iran
b
Institute of Petroleum Engineering, HeriotWatt University, Edinburgh EH14 4AS, UK
Received 17 October 2003; received in revised form 9 April 2004; accepted 23 April 2004
Available online 2 July 2004
Abstract
An equation for prediction of liquidliquid interfacial tension (IFT) in multi-component systems has been developed using the lattice theory
and the regular solution assumptions. The equation does not include any empirical parameter and there is no limitation on the number of
components comprising the system. The reliability of the equation has been evaluated by comparing its prediction with experimental data and
other methods.
2004 Elsevier B.V. All rights reserved.
Keywords: Interfacial tension; Liquidliquid equilibrium; Regular solution; Lattice theory
1. Introduction
The liquidliquid interfacial tension (IFT) plays an im-
portant role in numerous processes such as liquidliquid ex-
traction and improved oil recovery (IOR). Although our un-
derstanding of physics of interfacial phenomena is rapidly
progressing, general theoretical models that predict IFT in
multi-component mixtures frompure compounds data alone
are still lacking.
Most of the available methods [16] for prediction of
liquidliquid IFT require the mutual solubility as input data.
Donahue and Bartell [1] proposed an empirical correlation
that relates IFT of binary systems to a parameter, dened
by the authors as a degree of miscibility. This correlation
has been extended to ternary systems by Treybal [2]. The
Pliskin and Treybal [3] equation is based on the regular so-
lution theory and the concept of diffuse interface developed
by Cahn and Hilliard [7]. Their equation is reliable for bi-
nary aqueous and nonaqueous systems and ternary nonaque-
ous systems. Paul and de Chazal [4] extended the Frankel
equation [6] and the diffuse interface concept to develop a
method for estimating the IFT of binary and ternary aqueous

Corresponding author. Tel.: +44 131 451 3173;

fax: +44 131 451 3127.
E-mail address: ali.danesh@pet.hw.ac.uk (A. Danesh).
systems. Their equation requires some molecular parameters
that make it difcult to use. The simple equation proposed
by Fu et al. [5] is limited to binary and ternary systems.
The aim of this work is to develop a theoretically based
practical equation for prediction of liquidliquid IFT in
multi-component mixtures.
2. Theory
Assuming two liquid phases, and , separated by an
interface, , the chemical potentials
i
of the ith compo-
nent in bulk of (or ) phase and interface () can be
expressed as:

i
=
0,
i
(T, P) +RTln(X

i
) (1)

i
=
0,
i
(T, P) +RTln(X

i
)

a
i
(2)
where
0,
i
and
0,
i
are the standard chemical potential of
pure component i in the bulk () and interface (), R is the
universal gas constant, T is the temperature, X
i
is the mole
fraction,
i
is the activity coefcient, a
i
is the partial molar
surface area and

is the interfacial tension. Employing

the uniformity of chemical potential of each component in
bulk and interface at equilibrium conditions and consider-
ing the equality of the standard (pure compound) chemical
0378-3812/$ see front matter 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.uid.2004.04.012
198 A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205
potentials, Eqs. (1)(2) reduce to the well-known Shain and
Prausnitz [8] equation:

=
RT
a
i
ln
_
X

i
X

i
_
(3)
Solving Eq. (3) for X

i
, using the constraint of

c
i=1
X

i
= 1, we obtain,
c

i=1
_
X

i
exp
_
a
i
RT

_
_
= 1 (4)
where c is the total number of components in the system.
If the activity coefcient and the partial molar surface area
of each component in the interface are known, then Eq. (4)
can be used to predict

. For evaluation of the activity

coefcient in the interface, rst we employ the lattice the-
ory and the regular solution assumptions to nd the excess
Helmholtz free energy of a single bulk liquid phase and also
a system comprising of two liquid phases, and , separated
by an interface . Then by calculating the activity coef-
cients from the excess Helmholtz free energies we will nd
an expression for the activity coefcient of each component
in the interface that relates

i
to

i
and

i
.
Applying the lattice theory, we imagine that each
molecule in the liquid interacts with Z nearest neigh-
bours. Then the internal congurational energy of a
multi-component system is calculated as,
U
conf
=
c

i=1
(N
ii

ii
) +
1
2
c

i=1
c

j=i
(N
ij

ij
) (5)
where
ij
is the potential energy of interaction between a
molecule of type i with a molecule of type j and N
ij
is the
number of nearest neighbour contacts of type ij. N
ii
and N
ij
are calculated by:
N
ii
=
1
2
(N
0
nX
i
)(ZX
i
) =
1
2
N
0
nZ(X
i
)
2
(6)
N
ij
= (N
0
nX
i
)(ZX
j
) = N
0
nZX
i
X
j
(7)
where N
0
is the Avogadros number and n is the total number
of moles.
Combining Eqs. (5)(7) with some algebraic simplica-
tion (Appendix A), we obtain,
U
conf
=
1
2
N
0
nZ
c

i=1
(X
i

ii
) +
1
2
n
c

i=1
c

j=i
(X
i
X
j

ij
) (8)
where,

ij
N
0
Z(
ij

1
2
(
ii
+
jj
)) (9)
The rst term in the right hand side of Eq. (8) is the
internal congurational energy of an ideal solution, hence,
the next term is the energy of mixing or the excess internal
congurational energy, U
EX
U
EX
=
1
2
n
c

i=1
c

j=i
(X
i
X
j

ij
) (10)
By assuming the excess volume and the excess entropy to
be zero (regular solution assumptions) we have equality of
excess internal congurational energy with excess Gibbs and
Helmholtz free energies. Therefore, the activity coefcient
in a homogeneous phase is:
RTln (
i
) =

n
i
_
_
1
2
n
c

i=1
c

j=i
(X
i
X
j

ij
)
_
_
T,P,n
j
=
c

j=i
(X
j

ij
)
1
2
c

i=1
c

j=i
(X
i
X
j

ij
) (11)
Similarly, we can employ the lattice theory to evaluate
the excess internal congurational energy of a system com-
prising two homogeneous liquid phases ( and ) meeting
each other at an interface (). Fig. 1 depicts the concept,
where molecules are shown resided in planes parallel to the
interface. Each molecule has Z neighbours such that lZ of
them are in the same plane and mZ in each of the adjacent
plains, hence,
l +2m = 1 (12)
We assume the interface as a monomolecular layer and
also we consider the liquid phases ( and ) as composed
of series of monomolecular layers (K

plains in phase and

K

in phase), where each layer contains, like interface,

N
0
n

molecules. The monomolecular model of the inter-

face, as pointed out by Defay et al. [9] for liquidvapour and
liquidliquid interfaces, is incompatible with the Gibbs ad-
sorption model, and can result to a nonzero IFT gradient at
the consolute point. This discrepancy, however, can be elim-
inated by using multi-molecular layers but the obtained IFT
value, according to Defay et al. [9], will be practically the
same as that of a monomolecular model. We now evaluate
the internal congurational energy of the interface plane.
In the interface plane we acknowledge that the interaction
potential between two different specious (i and j) is different
A A
Z Z
phase, K plains
phase, K plains
Fig. 1. Conceptual residency of molecules in the two homogeneous bulk
phases and the interface.
A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205 199
to that in bulk phases. We denote the potential energy of in-
teraction between a type i molecule with a molecule of type j
in the interface plane by

ij
with

ij
=

ji
. We consider differ-
ent cross interaction potentials at the interface and in the bulk
phases. However, the interaction of a molecule of type i at
the interface with a molecule of type j in the bulk is assumed
to be the same for both and phases. The system, as we
have supposed, is homogeneous with regard to the number
density (the number of particles per unit volume) but hetero-
geneous with regard to the mass density and composition.
The congurational energy in the interface plane (U
1
) due
to interaction of molecules situated in it is:
U
1
=
1
2
c

i=1
[(N
0
n

i
)(lZX

i
)
ii
]
+N
0
n

lZ
c

i=1
c

j=i
(X

i
X

ij
) (13)
Using the same algebraic simplication described in
Appendix A, Eq. (13) reduces to:
U
1
=
1
2
N
0
n

lZ
c

i=1
(X

i

ii
) +
1
2
n

l
c

i=1
c

j=i
(X

i
X

ij
)
(14)
where,

ij
N
0
Z(2

ij

1
2
(
ii
+
jj
)) (15)
The internal congurational energy between the interface
and the two adjacent plain, U
2
, is:
U
2
=
c

i=1
[(N
0
n

i
)(mZX

i
)
ii
]
+
c

i=1
[(N
0
n

i
)(mZX

i
)
ii
]
+
c

i=1
c

j=i
[(N
0
n

i
)(mZX

j
)
ij
]
+
c

i=1
c

j=i
[(N
0
n

i
)(mZX

j
)
ij
] (16)
By simplication described in Appendix B, Eq. (16) reduces
to:
U
2
=N
0
n

mZ
c

i=1
(X

i

ii
) +n

m
c

i=1
c

j=i
(X

i
X

ij
)
+n

m
c

i=1
c

j=i
(X

i
X

ij
) +
1
2
N
0
n

mZ
c

i=1
(X

i

ii
)
+
1
2
N
0
n

mZ
c

i=1
(X

i

ii
) (17)
The internal congurational energy of each plain in two
liquid phases, U
3
and U
5
, and between each plain and the
adjacent plain, U
4
and U
6
, can be calculated by the same
procedure as for U
1
and U
2
.
U
3
=
1
2
N
0
n

lZ
c

i=1
(X

i

ii
) +
1
2
n

l
c

i=1
c

j=1
(X

i
X

ij
)
(18)
U
4
= N
0
n

mZ
c

i=1
(X

i

ii
) +n

m
c

i=1
c

j=1
(X

i
X

ij
)
(19)
U
5
=
1
2
N
0
n

lZ
c

i=1
(X

i

ii
) +
1
2
n

l
c

i=1
c

j=1
(X

i
X

ij
)
(20)
U
6
= N
0
n

mZ
c

i=1
(X

i

ii
) +n

m
c

i=1
c

j=1
(X

i
X

ij
)
(21)
The total internal congurational energy of system is the
summation of all internal congurational energies in and
between plains,
U
total
=U
1
+U
2
+K

U
3
+(K

1
2
)U
4
+K

U
5
+(K

1
2
)U
6
(22)
K

and K

are the number of plains in phases and ,

respectively. The 1/2 in Eq. (22) is for the farthest plains
fromthe interface (plains AA and ZZ in Fig. 1), where only
one half of the interaction is to be attributed to the system.
The contribution of the other half is due to that of outside the
system. Substituting the calculated energies in Eq. (22), and
remembering that K

= n

and K

= n

we obtain,
U
total
=
1
2
N
0
n

Z
c

i=1
(X

i

ii
) +
1
2
n

i=1
c

j=i
(X

i
X

ij
)
+
1
2
N
0
n

Z
c

i=1
(X

i

ii
) +
1
2
n

i=1
c

j=i
(X

i
X

ij
)
+
1
2
N
0
n

Z
c

i=1
(X

i

ii
) +
1
2
n

l
c

i=1
c

j=i
(X

i
X

ij
) +n

m
c

i=1
c

j=i
(X

i
X

ij
)

1
2
n

m
c

i=1
c

j=i
(X

i
X

ij
) +n

m
c

i=1
c

j=i
(X

i
X

ij
)
1
2
n

m
c

i=1
c

j=i
(X

i
X

ij
) (23)
According to Eq. (8), the rst two terms in Eq. (23) are
the internal congurational energies of bulk phase and
200 A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205
the next two terms are the same of bulk phase . The fth
term is the internal congurational energy of interface as an
ideal solution; therefore, the remaining terms are the excess
internal congurational energy attributed to the interface,
U
EX,
=
1
2
n

l
c

i=1
c

j=i
(X

i
X

ij
) +n

m
c

i=1
c

j=i
(X

i
X

ij
)
1
2
n

m
c

i=1
c

j=i
(X

i
X

ij
)
+n

m
c

i=1
c

j=i
(X

i
X

ij
)
1
2
n

m
c

i=1
c

j=i
(X

i
X

ij
) (24)
We can now evaluate the activity coefcient in the inter-
face similarly to that for the bulk solution,
RTln(

i
) =l

i
_
_
1
2
n

i=1
c

j=i
(X

i
X

ij
)
_
_
T,P,n

j
,

+m
c

j=i
(X

ij
) m
1
2
c

i=1
c

j=i
(X

i
X

ij
)
+m
c

j=i
(X

ij
) m
1
2
c

i=1
c

j=i
(X

i
X

ij
)
(25)
Analogous to Eq. (11) we suppose that:
0
5
10
15
20
25
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
Concentration of Acetone in Water Rich Phase / Mole Fraction
I
F
T

/
m
N
m
-
1
exp, IFT
This Work
Fu et al. [5]
Treybal [2]
Fig. 2. Experimental [12] and predicted liquidliquid interfacial tension of watertrichloromethaneacetone mixtures as a function of acetone mole fraction
in the water-rich phase at 25

C.

n
i

_
_
1
2
n

i=1
c

j=i
(X

i
X

ij
)
_
_
T,P,n

j
,

= RTln(

i
) (26)
Combining Eqs. (11), (25) and (26) yields,
RTln(

i
) = lRTln(

i
) +mRTln(

i
) +mRTln(

i
)
(27)
hence,

i
= (

i
)
m/(1l)
(28)
Employing Eq. (12) in the above equation gives the nal
expression for the activity coefcient of each component in
the interface,

i
= (

i
)
1/2
(29)
By substituting

i
from Eq. (29) in Eq. (4) and remem-
bering that at equilibrium conditions

i
/

i
= X

i
/X

i
, we
obtain,
c

i=1
_
(X

i
X

i
)
0.5
exp
_
a
i
RT

__
= 1 (30)
Eq. (30) is the nal equation, which can be used for pre-
diction of liquidliquid IFT in multi-component systems.
The application of Eq. (30) requires the value of partial
molar surface area of each component, a
i
. The partial molar
surface area of component i is dened as [9]:
a
i

_
A
n

i
_
T,P,

,n

j
(31)
A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205 201
Table 1
Experimental [5] and predicted interfacial tension values of binary mixtures
System

, experimental (mNm
1
) This work Fu et al. [5] DonahueBartell [1]
(1) Pentanewater 49 47.7 49.8 48.9
(2) Hexanewater 49.7 46.4 48.9 46.9
(3) Heptanewater 50.2 46 48.3 45.9
(4) Octanewater 50.2 45.4 47.4 44.9
(5) Cyclohexanewater 50.2 47.8 48.2 47.2
(6) Benzenewater 34.1 34.4 34.9 35.2
(7) Toluenewater 36.1 36.6 38.2 37.4
(8) Xylenewater 37.2 36.6 38.7 36.9
(9) Ethyl benzenewater 38.4 36.9 39.2 37.2
(10) n-Butanolwater 1.8 2.8 4 1.0
(11) i-Butanolwater 2 3.4 4.7 1.8
(12) n-Pentanolwater 4.4 5.5 6.6 3.6
(13) i-Pentanolwater 4.8 5.7 6.8 3.8
(14) n-Hexanolwater 6.8 6.3 8.2 3.9
(15) n-Heptanolwater 7.7 7.3 8.5 5.6
(16) Cyclohexanolwater 3.9 4.3 5.3 2.3
(17) Diethyl etherwater 11 15 14.2 14.9
(18) Diisopropyl etherwater 17.9 16.6 18.3 16.7
(19) 2-Pentanonewater 6.3 9.4 11.1 9.2
(20) 2-Hexanonewater 9.6 12.1 13.7 11.3
(21) 2-Heptanonewater 12.4 14.7 16.2 13.5
(22) Ethyl acetatewater 6.8 9.4 11.1 9.4
(23) n-Butyl acetatewater 14.5 13.5 15.3 12.2
(24) Furfuralwater 4.7 7.4 9.3 6.7
(25) Dichloromethanewater 28 26.9 25.7 26.7
(26) Chloroformwater 31.6 29.7 29.4 30.3
(27) Tetrachloromethanewater 43.7 42.8 42.1 44.2
(28) 1,1,2-Trichloroethanewater 29.6 28.3 29.2 28.6
(29) Bromoethanewater 31.3 30.5 29.6 31.0
(30) Bromobenzenewater 38.1 36.1 38.1 36.2
(31) Nitromethanewater 9.5 12.2 13.8 11.8
(32) Carbon disuldewater 48.1 46.5 43 44.5
(33) Methanoltetradecane 4.6 9.5 12.6 15.3
(34) Perurohexanetetradecane 6.1 10.2 4.5 18.9
(35) i-Pentanewater 50.1 47.3 49.4 48.7
(36) 2,2-Dimethyl butanewater 49.7 48.1 50.9 49.6
(37) 2,3-Dimethyl butanewater 49.8 46.1 48.8 47
(38) i-Hexanewater 48.9 48.2 51 49.2
(39) 3-Methyl pentanewater 49.9 47.8 50.6 48.9
(40) 2,2,4-Trimethyl pentanewater 50 44.7 47.1 44.6
(41) 3-Methyl hexanewater 50.4 48.9 51.7 49.5
(42) 2,4-Dimethyl pentanewater 50 48.2 51.1 48.9
(43) n-Buthyl benzenewater 41.4 36.9 38.9 36.3
(44) Methyl isobutyl ketonewater 10.1 12.3 14 11.8
(45) n-Decanewater 52 48.1 49.8 47.3
(46) n-Dodecanewater 52.8 47.6 49.3 46.8
(47) n-Hexadecanewater 53.3 46.9 48.6 46.2
(48) Carbon disuldeperurohexane 9.6 11.3 14.7 15.2
(49) Benzeneperurohexane 5.8 10.4 7.1 10.7
(50) n-Octanolwater 8.5 10.3 10.7 7.8
(51) m-Xylenewater 37.9 36.8 39 37.1
AAD% 18.1 24.8 23.5
Table 2
Deviations of predicted interfacial tension values for ternary systems
System Average absolute deviation (%)
This work Fu et al. [5] Treybal [2] No. of data points
Water (1)trichloromethane (2)acetone (3) 45 61 44 5
Water (1)1,1,2-trichloroethane (2)acetone (3) 53 59 46 7
Water (1)4-methyl-2-pentanone (2)acetone (3) 73 13 14 5
Water (1)acetic acid isopropylester (2)acetic acid (3) 79 82 39 5
Overall AAD% 62 54 36
202 A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205
0
5
10
15
20
25
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
Concentration of Acetone in Water Rich Phase / Mole Fraction
I
F
T

/

m
N
m
-
1
exp, IFT
This Work
Fu et al. [5]
Treybal [2]
Fig. 3. Experimental [12] and predicted liquidliquid interfacial tension of water1,1,2-trichloroethaneacetone mixtures as a function of acetone mole
fraction in the water-rich phase at 25

C.
We are not aware of any rigorous method to calculate the
partial molar surface area in a multi-component system. In
general, it must depend on the temperature, pressure and
composition of both phases. Since our model relies on the
regular solution assumptions, it is reasonable to assume that
the excess surface area in the interface is zero. Therefore, we
propose using the molecular surface area of a pure compound
instead of its partial molar value.
Sprowand Prausnitz [10] proposed the following equation
that relates a
i
to the liquid molar volume, V
i
, of a pure
0
2
4
6
8
10
12
14
0.00 0.02 0.04 0.06 0.08 0.10
Concentration of Acetone in Water Rich Phase / Mole Fraction
I
F
T



/

m
N
m
-
1
exp, IFT
This Work
Fu et al. [5]
Treybal [2]
Fig. 4. Experimental [12] and predicted liquidliquid interfacial tension of water4-methyl-2-pentanoneacetone mixtures as a function of acetone mole
fraction in the water-rich phase at 25

C.
compound,
a
i
= V
2/3
i
N
1/3
0
(32)
3. Results
Table 1 shows the predicted interfacial tensions for 47
aqueous and four nonaqueous binary systems. The exper-
imental data in Table 1 are from reference [5]. It should
A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205 203
-2
0
2
4
6
8
10
0.00 0.02 0.04 0.06 0.08 0.10 0.12
Concentration of Acetic Acid in Water Rich Phase / Mole Fraction
I
F
T


/

m
N
m
-
1
exp, IFT
This Work
Fu et al. [5]
Treybal [2]
Fig. 5. Experimental [12] and predicted liquidliquid interfacial tension of wateracetic acid isopropyl esteracetic acid mixtures as a function of acetic
acid mole fraction in the water-rich phase at 25

C.
be noted that we used a value of 32
2
for water molec-
ular surface area, based on the monomolecular character
of an immiscible lm measured by the Langmuir method
[11], instead of the generalised equation, Eq. (32). The pre-
dicted results by Fu et al. and DonahueBartell equations
are also included in Table 1. Note that the predicted values
of the new model are comparable with the results obtained
by the Fu et al. and DonahueBartell methods, with a sig-
nicantly lower average absolute deviation from the exper-
imental data. It should be pointed out that the new equation
has no empirical parameter, whereas the Fu et al. equation
has one and the DonahueBartell equation has two empirical
parameters.
Table 2 shows the average absolute deviations of the
predicted interfacial tension by the new, Fu et al. and the
modied DonahueBartell equations [2] for four aque-
ous ternary systems [12]. The predicted values have also
been plotted in Figs. 25. The results of the new equa-
tion are comparable with the other two methods, although
this new model uses no empirical parameter. Since the
new model is based on the regular solution assumptions,
higher deviations from experimental data on systems de-
viating far from the regular solution are expected. The
higher deviation of predicted IFT values in the ternary
systems is partly due to the assumption of regular so-
lution but more due to the estimated molecular area. It
should be noted that the proposed model has the advan-
tage of being employed simply in multi-component sys-
tems while the other two methods are limited to ternary
systems.
4. Conclusion
An equation for prediction of liquidliquid IFT in
multi-component systems has been developed on the con-
cepts of lattice theory and regular solution assumptions. The
IFT values predicted by this model were compared with
two semi-empirical equations and evaluated against mea-
sured data for a large number of systems. The model, which
does not use any empirical data, compares favourably with
added advantages of being applicable to multi-component
systems with no limitation on the number of components.
List of symbols
A total surface area
AAD average absolute deviation (%)
a partial molar surface area (m
2
mol
1
)
c number of components
K number of plains in the bulk phase
l lattice parameter
m lattice parameter
n total number of moles
N
0
Avogadros number
N
ij
number of nearest neighbour contacts of type ij
P pressure
R universal ideal gas constant
T temperature (K)
U internal energy
U
conf
internal congurational energy
U
EX
excess internal congurational energy
U
total
total internal congurational energy
V molar liquid volume (m
3
mol
1
)
204 A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205
X mole fraction
Z number of nearest neighbours
Greek letters

ij
energy parameter, Eq. (9)

ij

energy parameter, Eq. (15)

ij
interaction potential energy

ij

interaction potential energy at the interface

activity coefcient
chemical potential
interfacial tension (mNm
1
)
Subscripts
i ith compound
j jth compound
Superscripts
attributed to the bulk phase
attributed to the bulk phase
attributed to the interface
0 at standard conditions
Appendix A
Eqs. (5)(7) yield:
U
conf
=
1
2
N
0
nZ
c

i=1
(X
2
i

ii
) +
1
2
N
0
nZ
c

i=1
c

j=i
(X
i
X
j

ij
)
(A.1)
By using X
i
= 1

c
j=i
X
j
the rst term of the above
equation reduces to:
1
2
N
0
nZ
c

i=1
(X
2
i

ii
) =
1
2
N
0
nZ
c

i=1
(X
i

ii
)

1
2
N
0
nZ
c

i=1
c

j=i
(X
i
X
j

ii
) (A.2)
The second term of Eq. (A.2) can be written as follows:
1
2
N
0
nZ
c

i=1
c

j=i
(X
i
X
j

ii
) =
1
2
N
0
nZ
c

i=1
c

j=i
_
X
i
X
j

ii
2
_
+
1
2
N
0
nZ
c

i=1
c

j=i
_
X
j
X
i

jj
2
_
(A.3)
Combination of the above three equations yields Eq. (8).
Appendix B
By substituting X

i
= 1

c
j=i
X

j
and X

i
= 1

c
j=i
X

j
in the rst term and the second term of Eq. (16), respectively,
Eq. (16) simplies to:
U
2
=2N
0
n

Zm
c

i=1
(X

i

ii
)
+N
0
n

Zm
c

i=1
c

j=i
[X

i
X

j
(
ij

ii
)]
+N
0
n

Zm
c

i=1
c

j=i
[X

i
X

j
(
ij

ii
)] (B.1)
We can write the second term of Eq. (B.1) as follows:
N
0
n

Zm
c

i=1
c

j=i
[X

i
X

j
(
ij

ii
)]
= N
0
n

Zm
c

i=1
c

j=i
_
X

i
X

j
_

ij

ii
2
__
N
0
n

Zm
c

i=1
c

j=i
_
X

i
X

ii
2
_
(B.2)
Substituting

c
j=i
X

j
= 1 X

i
in the above, the second
part of Eq. (B.2) reduces to:
N
0
n

Zm
c

i=1
c

j=i
_
X

i
X

ii
2
_
= N
0
n

Zm
c

i=1
_
X

ii
2
_
N
0
n

Zm
c

i=1
_
X

i
X

ii
2
_
(B.3)
Substituting X

i
= 1

c
j=i
X

j
in the second term of
the above equation, we obtain:
N
0
n

Zm
c

i=1
_
X

i
X

ii
2
_
= N
0
n

Zm
c

i=1
_
X

ii
2
_
N
0
n

Zm
c

i=1
c

j=i
_
X

j
X

ii
2
_
(B.4)
Remembering that

c
i=1

c
j=i
_
X

j
X

ii
2
_
=

c
i=1

c
j=i

_
X

i
X

jj
2
_
and combining Eqs. (B.2)(B.4) reduce the
A. Bahramian, A. Danesh / Fluid Phase Equilibria 221 (2004) 197205 205
second term of Eq. (B.1) as follows:
N
0
n

Zm
c

i=1
c

j=i
[X

i
X

j
(
ij

ii
)]
= mn

i=1
c

j=i
(X

i
X

ij
) +
1
2
N
0
n

Zm
c

i=1
(X

i

ii
)

1
2
N
0
n

Zm
c

i=1
(X

i

ii
) (B.5)
The last term of Eq. (B.1) is treated exactly the same as the
second term, shown above, except changing the superscript
to . Substituting the above terms in Eq. (B.1) we obtain
Eq. (17).
References
[1] D.J. Donahue, F.E. Bartell, J. Phys. Chem. 56 (1952) 480484.
[2] R.E. Treybal, Liquid Extraction, second ed., McGraw-Hill, New
York, 1963.
[3] I. Pliskin, R.E. Treybal, AIChE J. 12 (1966) 795801.
[4] G.W. Paul, L.E.M. de Chazal, Ind. Eng. Chem. Fundam. 8 (1969)
104108.
[5] J. Fu, B. Li, Z. Wang, Chem. Eng. Sci. 41 (1986) 26732679.
[6] J. Frankel, Kinetic Theory of Liquid, Dover, New York, 1955.
[7] J.W. Cahn, J.E. Hilliard, J. Chem. Phys. 28 (1958) 258267.
[8] S.A. Shain, J.M. Prausnitz, AIChE J. 10 (1964) 766773.
[9] R. Defay, I. Prigogine, A. Bellemans, D.H. Everett, Surface Tension
and Adsorption, Longmans, London, 1966.
[10] F.B. Sprow, J.M. Prausnitz, Can. J. Chem. Eng. 45 (1967) 2528.
[11] E.K. Rideal, An Introduction to Surface Chemistry, Cambridge, Lon-
don, 1930.
[12] H.M. Backes, J.J. Ma, E. Bender, G. Maurer, Chem. Eng. Sci. 45
(1990) 275286.