J. Chem.

Thermodynamics 44 (2012) 31–37

Contents lists available at SciVerse ScienceDirect

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

Review

A review of Barker’s activity coefficient method and VLE data reduction

Akand W. Islam a,⇑, Md. Hafizur Rahman b
a
Department of Chemical & Biological Engineering, The University of Alabama, Tuscaloosa, AL 35401, USA
b
Department of Chemical & Biological Engineering, The University of British Columbia, Vancouver, B.C., Canada

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

Article history: The method of Barker is a popular scheme for determination of activity coefficients from total pressure
Received 11 July 2011 measurements. A comprehensive review of this method is presented in this study. While discussing this
Received in revised form 7 August 2011 technique various aspects of (vapor + liquid) equilibrium (VLE) data reduction process including types of
Accepted 12 August 2011
algorithms applied, roles of saturated vapor pressures and equilibrium vapor compositions data, and
Available online 22 August 2011
types of objective functions used are analyzed. Activity coefficient or liquid state models frequently used
in VLE data reduction are shown and their comparisons are investigated. More so, advantages and limi-
Keywords:
tations of Barker’s method are demonstrated.
Barker’s method
Activity coefficient
Ó 2011 Elsevier Ltd. All rights reserved.
Data reduction
Excess Gibbs free energy

Contents

1. Introduction . . . . . . . . . . . . . . . ..................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2. Algorithms, roles of psat i , y, and objective functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3. g Models used. . . . . . . . . . . . . . ..................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4. Applications to higher-ordered systems and limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5. Conclusions. . . . . . . . . . . . . . . . ..................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Acknowledgement . . . . . . . . . . ..................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
References . . . . . . . . . . . . . . . . ..................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

1. Introduction vapor/liquid equilibrium data are sufficiently demonstrated by the

authors in reference [2]. The main advantage of Barker’s method is
Barker’s [1] method is an attractive P–x data reduction proce- that experimental difficulties in measuring y (vapor compositions)
dure and this is a one-step analytical fitting technique that makes can be avoided since all values of y are obtained through data
direct use of the measured data [2]. This uses an arbitrary model regressions. This method is useful in such cases where the gaschro-
for the molar excess Gibbs free energy, GE (here expressed by g) motographic analysis of the vapor phase composition is not accu-
and then obtains the constants of that model from the isothermal rate enough [6]. However, the greatest limitation of Barker’s
P–x data. Alternatively, it can also be applied to isobaric T–x data method is that the g model must be chosen. The quality of the re-
[3]. Primarily it was developed for non-electrolyte hydrocarbon sults will significantly depend on the chosen g model. In this article
mixtures; however, it is now being applied to electrolytes as well. computational techniques applied in Barker’s method will be
This can be carried out without numerical problems for electrolytic investigated and some comparisons of g models used in this meth-
systems and temperatures higher than the critical temperature [4]. od for different (vapor + liquid) equilibrium (VLE) systems will be
Won and Prausnitz [5] extended this method to binary systems discussed.
containing one supercritical component. The general applicability
and advantages of Barker’s method for the reduction of isothermal
2. Algorithms, roles of psat
i , y, and objective functions

Several algorithms are proposed for data reduction using Bar-

⇑ Tel.: +1 2052108153. ker’s method. From thermodynamics, the total pressure P for a bin-
E-mail address: awislam@crimson.ua.edu (A.W. Islam). ary system is given as

0021-9614/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jct.2011.08.013
32 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37

   
=
P ¼ x1 c1 p1 þ x2 c2 p2 :
=
ð1Þ x1 psat
1 dg x2 psat
2 dg
P¼ exp g þ x2 þ exp g  x1 ; ð9Þ
= =
U1 dx1 U2 dx1
The symbols p1 and p2 are the ‘‘corrected’’ vapor pressures of com-
ponent 1 and 2, respectively. These are defined by where
" #
= sat ðV L1  B11 ÞðP  psat
1 Þ Pd12 y22 g  GE =RT ¼ x1 ln c1 þ x2 ln c2 ; ð10Þ
p1 ¼ P1 exp  ; ð2Þ
RT RT y P
" # c1 ¼ 1 sat U1 ð11Þ
x1 p1
ðV L2  B22 ÞðP  psat
2 Þ Pd12 y21
p=2 ¼ Psat
2 exp  :
RT RT and
!
Here, P sat sat ðB11  V L1 ÞðP  psat
1 Þ Py2 d12
1 and P 2 are the vapor pressure of the pure components, x U1 ¼ exp þ 2 : ð12Þ
and y are liquid and vapor molar fractions respectively, c1 and c2 are RT RT
activity coefficients, V L1 and V L2 are liquid molar volumes, B11, B22,
Here equations (9)–(12) are the algebraic manipulations of equa-
and B12 are second virial coefficients in the equations of state of
tions (1), (2), and (8). Equation (12) or the symbol U in equation
the pure and mixed vapors, and d12 = 2B12  B11  B22. R is the uni-
(9) is accounted for vapor pressure corrections. Similar to the previ-
versal gas constant. Barker [1] himself used the following activity
ous case, the calculation starts by a guessed value of y in equation
coefficient functions for equation (1):
(12). Equation (9) is minimized through the regression of a set of
ln c1 ¼ C 0 l1 þ C 1 m1 þ C 2 n1 þ    ; values for the parameters in the correlating expression for g to find
ð3Þ
ln c2 ¼ C 0 l2 þ C 1 m2 þ C 2 n2 þ    ; Pcal. First and second terms of this equation represent Py1 and Py2,
respectively. Therefore new y values can be estimated from corre-
where sponding terms by division of Pcal. This process is continued until
l1 ¼ x22 ; m1 ¼ x22 ð1  4x1 Þ; n1 ¼ x22 ð1  8x1 þ 12x21 Þ; a steady y values are obtained. They [2] provided an alternative pro-
ð4Þ cedure basing the correlation on just P–x data through application
l2 ¼ x21 ; m2 ¼ x21 ð1  4x2 Þ; n2 ¼ x21 ð1  8x2 þ 12x22 Þ: of equation (9). However, single values of psat sat
1 and p2 are required

The constants C0, C1, C2 were determined by successive approxima- whereas each data set has its own values for the pure component
tions. To initiate calculations C1 and C2 were set to zero and C0 was vapor pressures. To overcome this, they obtained an additional
given by equation (equation (13)) to correct all reported values of P
  rP rpsat rpsat
2P ¼ y1 1
þ y2 2
: ð13Þ
C 0 ¼ 4 ln : ð5Þ P psat psat
psat
1 þ p2
sat
1 2

P⁄ was the pressure for equimolar mixture estimated graphically. The corrected P values are then used along with the average values
Using this value of C0, approximate vapor concentrations were cal- of psat
1 and psat
2 in equation (9). The parameters determined by
culated and were sufficient enough to be used in the small correc- regression are finalized only when the entire data set (including
Pd12 y22
tion term RT
= =
in equation (2). Subsequently p1 and p2 were psat sat
1 and p2 ) is validated in accord with equation (13).
calculated from equation (2) followed by calculating a new P from Equation (1) can be re-written as follows for component i:
equation (1). The residuals r = Pexp  Pnew, and derivatives of dP/ X
P¼ ðxi ci psat
i =Ui Þ: ð14Þ
dC0, dP/dC1, dP/dC2were predicted by i
= =
dP=dC 0 ¼ l1 c1 p1 þ l2 c2 p2 ; Here psat of the pure components appear explicitly as physical con-
i
dP=dC 1 ¼ c =
m1 1 p1 þ m2 2 p2 ;
=
c ð6Þ stants, presumably known at the temperature at which P–x data are
= = considered. Moreover, psat i are properly measured as part of the P–x
dP=dC 2 ¼ c c
n1 1 p1 þ n2 2 p2 : data set and they should be included in the regression process like
all other data points. As xi approaches 1.0, the psat i values are no
The changes oC0, oC1, oC2 in C0, C1, C2 would most nearly reduce the more significant than other values of P and are equally subject to
pressure residuals to zero by fitting least squares to the equation experimental error and to smoothing during data reduction. Their
ðdP=dC 0 Þ@C 0 þ ðdP=dC 1 Þ@C 1 þ ðdP=dC 2 Þ@C 2 ¼ r: ð7Þ special significance enters only through the right terms in equation
(14) interacting with all data points of the regression set. The
Improved C0, C1, C2 would be obtained upon adding oC0, oC1, oC2 in authors in reference [8] suggested to accommodate these roles of
C0, C1, and C2, respectively. This process would be repeated until C0, psat
i by treating them both as data points and as adjustable param-
C1, C2 did not change significantly. eters; thus measured values of P for xi = 1.0 are included in the data
Prausnitz et al. [7] showed a slightly different approach. In their set, however, are not identified initially with psat
i . Rather, the values

techniques y1 and y2 are set equal to zero in equation (2). Then of psat
i are determined by regression along with the parameters of
experimental P–x data are regressed to estimate the parameters the expression of g. According to them, these values of psat i become
of appropriate activity co-efficient models (expressions of c1 and the smoothed values of P for xi = 1.0 in any comparison of correlated
c2). Once the parameters are obtained, new y1 and y2 can be found results with experimental data. This extended Barker’s method can
by solving the following non-linear equation (8). This process be recommended as a procedure of complete P–x data set reduction,
should be iterative until a consistent y1 and y2 reached however, application of this technique is not guaranteed as a means
 of correcting for the neglect of researchers to provide vigilant mea-
Py1 ðB11  V L1 ÞðP  psat
1 Þ P 1 y22 d12 surements of psat values, taken in the same apparatus and with the
ln c1 ¼ ln sat
 þ ; i
x1 p1 RT RT same materials as all other reported P–x data.
  ð8Þ Kuschel et al. [9] elaborately discussed the role of vapor compo-
Py2 ðB22  V L2 ÞðP  psat
2 Þ P 2 y21 d12
ln c2 ¼ ln sat
 þ : sition (y) in VLE data reduction. The superiority of the g values is
x2 p2 RT RT
most striking in systems having large relative volatility values.
Abbot and Van Ness [2] showed regression of following P–x relation Inclusion of y values of these systems into the data reduction re-
by sults in considerable deterioration of the calculated g values.
 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37 33

Therefore it is recommended to reduce P–x data, not P–x–y data Np

X
simultaneously. On the other hand, it is much better to reduce a OF1 ¼ ðrPÞ2 ; ð18Þ
i¼1
complete P–x–y data than incomplete P–x set. Process of y-mea-
Np
X
surement is usually equal to x-measurement (of course, in depen-
OF2 ¼ ½rðDP=PÞ2 ; ð19Þ
dence on system) and troubles are introduced due to imperfection
i¼1
of a model equation. Moreover, a great experimental problem of Np
X
total pressure measurement consists in a perfect degassing of the OF3 ¼ ðryÞ2 ; ð20Þ
system. This is not easy experimental task nevertheless insufficient i¼1
degassing leads to a great data inconsistence. This problem is fre- Np
X Np
X
1 1
quently neglected so making P–x data doubtful. In data reduction OF4 ¼ ½rðDP=PÞ2 þ ðryÞ2 ; ð21Þ
process measuring y is not trivial also and as a result causes large i¼1
r2DP=P i¼1
r2y
associated uncertainties. It leads to values of y that are less uncer-
where
tain than the corresponding experimental y values and the use of
Barker’s y values in speciation calculations are suggested [10,11]. rP ¼ Pexp  Pcal ;
Barker’s method generally involves following minimization Pexp
formula rðDP=PÞ ¼  1; ð22Þ
Pcal
X
n
ry ¼ yexp  ycal :
S2 ¼ Q 2Pj ; ð15Þ
In equation (21) rDP/P and ry indicate root mean square values of
j¼1
r(DP/P) and ry, respectively. This OF is more appropriate for fit-
where ting process than regressing only P or y values using equations
Z ! (18)–(20). Duce et al. [17] used a new type of OF for Barker’s com-
X
n
  Pj
Q Pj ¼ P j  xi i psat
c u sat
ui exp ðv i =RTÞdP : ð16Þ putations like,
i i =
psat "
i¼1 i
X
n  exp  cal #2
c c1
Shanker et al. [12] exclusively pointed out the facts regarding this OF ¼ ln 1  ln : ð23Þ
c2 k c2 k
objective function. The exponential term results from combining k¼1

the corrections due to pressure for the activity coefficient and pure
component fugacity and considering the v i (partial molar volume),
is well represented by the pure component molar volume vi. S2 is
minimized with respect to experimental P by finding ‘‘suitable’’ cor-
relating parameters for activity coefficient model ci, and calculating
the fugacity coefficient from an appropriate equation of state. How-
ever, the experimental data itself are subject to error. The principle
of maximum likelihood is used to account for errors in the experi-
mental data. If the distributions of error are normal, the method
employing the principle of maximum likelihood reduces to the
method of least squares. Moreover, it is probable that if the error
distributions are symmetrical, the maximum likelihood and least
squares estimates will be nearly equal. Since the experimental er-
rors should be at least symmetrically, if not normally, distributed,
the analysis should use the following definition of S2
n 
X     
S2 ¼ Q 2Pj =r2Pj þ Q 2x= =r2x= þ Q 2T = =r2T = ; ð17Þ
j j
j¼1
where Q Pj is defined by equation (16), Q x= ¼ X j  xj and Q T = ¼ T j  t j .
j j
Xj and Tj are observed values where xj and tj are estimates of the
experimental values. The r2 values refer to the variance of the
experimental measurements which are determined by replication.
Equation (17) is based on the principle of maximum likelihood by
noting that, for a given experiment the probability density function
(PDF) of the measurements is described by a multivariate normal
distribution around the experimental values. Since each experiment
is independent, the likelihood function is then a product of the PDF
values of the experiments being considered. If the covariances be-
tween measurements are zero, minimizing equation (17) is equiva-
lent to maximizing the likelihood function. Maximum likelihood
allows simultaneous determination of estimates of the parameters
and of the experimental uncertainties if the latter are unknown.
Since there are always experimental uncertainties determined from
replicate experiments, this more general method should not be
used. Equation (17) is recommended to minimize the same iterative
technique to improve the fugacity coefficients while minimizing S2
in equation (15) [12–15]. Bertucco et al. [16] summarized following
objective functions to use in Barker’ method for data reduction:
This OF will make data reductions faster with more accuracy be-
cause parameters of activity coefficient model will be predicted di-
rectly from experimental c’s. Islam et al. [31] showed an easy way to
find experimental c’s provided P and y are available.
3. g Models used
As mentioned earlier, the quality of the results from Barker’s
data reduction is largely dependent on the g model used. Detailed
discussions on the g models (or liquid state models) are beyond the
scope of this article, therefore only the types of g models frequently
used in Barker’s calculations will be illustrated.
Regenstein [18] successfully determined activity coefficients of
CT (Charge Transfer) complexes exhibiting termolecular complex
formation due to donor excess. He used some polynomial function
for g expression rather than following any well known model like
Wilson, Van Laar, UNIQUAC, UNIFAC, NRTL, Redlich–Kister and so
forth. Jain and Yadav [19], Jain and Wadi [20] used the following
j j
correlation similar to Redlich–Kister equation
n o
g ¼ x1 x2 C 0 þ C 1 ðx1  x2 Þ þ C 2 ðx1  x2 Þ2 : ð24Þ
Weiguo et al. [21] applied following for ternary systems
g ¼ D13 þ D23 þ D12 þ D123 ; ð25Þ
where binary terms D , D , D 13 23 12
are expressed as (in terms of i and
j)
X
m
Dij ¼ ðxi þ xj Þxi xj Af ð1  2xi Þf ð26Þ
f ¼0
and
xi i=ðxi þ xj Þ: ð27Þ
Here A1, A2, A3, etc., represent constants. The ternary term D123 is gi-
ven by
D123 ¼ x1 ð1  x1  x2 ÞfC 0 þ C 1 ð3x1  1Þ þ C 2 ð3x2  1Þg: ð28Þ
For associated type solutions (like hexanol + hexane), satisfactory
results can be obtained using [22,23]
34 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37

g ¼ g ass þ g res ; ð29Þ methanol + o-xylene, ethanol + o-xylene). For ternary mixtures like
(methanol + methyl acetate + o-xylene) and (ethanol + ethyl ace-
where
tate + o-xylene), Wilson is the best g model in reproducing exper-
g ass =RT ¼ xfð1 þ KÞ lnð1 þ KÞ  ð1 þ K uÞ½lnð1 þ K uÞg=K; imental data [143]. For a (hydrocarbon or CCl4 + a cyclic ketone)
ð30Þ
g res =RT ¼ xð1  xÞC 0 C 1 =fC 0 x þ C 1 ð1  xÞg system, Wilson rather than the Redlich–Kister model generates
more accurate results [43]. This expression represents binary and
and ternary mixtures containing di-isopropyl ether, 2-propanol, and
u ¼ x=fx þ R21 ð1  xÞg: ð31Þ benzene better than Wohl Expansion, Margules, UNIQUAC and
NRTL models [116]. Accuracy of Wilson is higher than Redlich–Kis-
Here gass represents the contribution of association which may be ter for (butanone + hexanol or octanol) systems [82]. The Wilson
desired by chemical equilibria and gres the contribution of residual equation does as well if not better for cryogenic mixtures of (nitro-
interactions. The parameter K represents the association constant gen + argon), (nitrogen + carbon monoxide), (argon + methane),
and R21 is the ratio of molar volumes between components 2 and 1. and (carbon monoxide + methane) than the two-term Redlich–Kis-
Among the multi-parameter expression the Marsh equation is ter equation, and for the (nitrogen + methane) system, the use of
flexible enough to describe binary mixtures of methanol, ethanol, the three-term Redlich–Kister equation a significant improvement
propanol, and butanol with benzene, toluene, and p-xylene [24– [109].
26] The results of the three binary systems (pentane + methanol,
" #" #1 methanol + acetone, and acetone + pentane) shows that the four-
Xm1
i
X
m2
j
g ¼ x1 ð1  x1 Þ Ai ð2x1  1Þ 1 þ Aj ð2x1  1Þ ; suffix Margules equation, the two- and three-parameter forms of
i¼0 j¼1
ð32Þ the Wilson equation and the NRTL equation yield essentially iden-
m1 þ m2 6 6: tical results and represent the best models [148]. The UNIQUAC
equation does not do as well and models the (pentane + methanol)
Pade approximant is as follows: poorly. As a consequence, the latter model also performs
P significantly worse than previous three models for ternary system
C 0 þ Nn¼1 C n ðx1  x2 Þn
g ¼ x1 x2 PM : ð33Þ consisting of these binaries (pentane + methanol + acetone). Camp-
1 þ m¼1 Am ðx1  x2 Þm bell et al. [134] found that four suffix Margules equation is far bet-
This equation reduces to Redlich–Kister expansion when M = 0. In ter than three suffix Margules equation, however the five suffix
data regression this is also used frequently [26–33]. Margules equation is not significantly better than the four suffix
Legendre polynomial [34] used for g expression is given as: Margules equation. Nevertheless more parameters do not improve
significantly this correlation. Five parameters of this equation are
X
n
required for successful data reduction of the (ethanol + heptanes)
g ¼ x1 x2 Ak C k ; ð34Þ
k¼0
system, whereas for (heptane + isobutanol) and (ethanol + isobuta-
nol) systems, the four- and three-suffix equations are more prefer-
where able [139].
C 0 ¼ 1; Dragoescu et al. [47] applied Redlich–Kister, Wilson, UNIQUAC,
NRTL for (cyclohexanone + dicholoro-alkane) binary mixtures,
C 1 ¼ ð2x1  1Þ;
ð35Þ however, observed that for some reason UNIQUAC and NRTL are
1 unable to represent meaningful data reduction. On the other hand
C 2 ¼ ½3ð2x1  1ÞC 1  1
2 Wilson, Margules and NRTL showed reasonable outputs for (tetra-
for 2 6 k, hydrofuran + propanol), and 2-propanol systems [115].
Wang et al. [146] analyzed their experimental data of binary
1
Ck ¼ ½ð2k  1Þð2x  1ÞC k1  ðk  1ÞC k2 : ð36Þ (a-pinene + alkanol) systems using the UNIQUAC equation with
k the linear temperature-dependent binary parameters in Barker’s
This polynomial has been used by many investigators [35–42] for calculations and obtained satisfactory results. Smith and Visco
determining activity coefficients. The confidence limits of the [149] also used Margules models with exponential temperature
parameters of the Legendre polynomial increases and the absolute dependent parameters.
values of the off-diagonal elements become closer to unity. In addi- Garriga et al. [26] obtained the best standard deviation for the
tion the confidence limit of the fifth parameter includes the param- pressure with a Pade 3/1 coefficients for (butanol + 2-methylpen-
eter value zero. It turns out that the four parameters are sufficient tane, and +3-methylpentane). Pando et al. [150] developed g model
to represent the data [37]. appropriate to (alcohol + alcohol) mixtures. They showed their
From literature it is observed that researchers mostly use model performed significantly better compared to UNIQUAC and
Redlich–Kister [9,24,26,27,29,43–113], Wilson [12,24, 47,69, 82, NRTL frameworks. In some cases, molecular model based on
104,107–109,114–127], Margules [115,116,126,128–141], (3- Lennard-Jones theory is more applicable than Redlich–Kister for
Suffix [130,131,139,140], 4-Suffix [132,133,135–137,139,141], (xenon + methyl chloride) system [59].
5-Suffix [116,139], 6-Suffix [140]), NRTL [24,34,35 ,47,69 ,104, From the above observations, it can be said that rather than try-
107,115,126,142–145], UNIQUAC [6,24 ,34,35 ,47,116, 142,146], ing with any particular model, it is very important to check the
Van Laar [24,29,143], UNIFAC [36,115,143,147] models in Barker’s results using different g models in order to obtain reliable y com-
method for their data regression. Redlich–Kister and Margules are positions out of Barker’s method. This is due not only the
basically identical having only a different form; very trivial rela- limitation of Barker’s method itself but also indicates the short-
tionships exist between parameters of both the equations. UNIFAC comings of liquid state models used. To avoid the use of g models,
is only the predictive method; therefore, it cannot be used for cor- several authors have recommended procedures for numerical cal-
relation. Some comparative observations of all above mentioned culations [3]. These procedures fall into two categories: indirect
schemes on various binary or ternary VLE systems for data reduc- models yielding the g expression and direct methods integrating
tion are described below. the Gibbs–Duhem equation in order to obtain y. However, these
Wilson correlations are better for those systems containing model-free algorithms are necessarily applicable to binary mix-
compounds with very different boiling temperatures (for instance, tures. Other than Hu et al. [3], some investigators [151,152] tried
 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37 35

to apply to ternary systems. However, they came up with ineffi-
cient calculations.
In many cases, VLE measurements are carried out under iso-
baric conditions. This is due to the fact that isobaric data are di-
rectly applicable in the synthesis and design of separation
technologies, particularly in distillation. Moreover, pressure is eas-
ier variable to control in dynamic equilibrium stills. However, from
a theoretical point of view, the treatment of isobaric VLE data is re-
stricted to the availability of excess enthalpy data for the liquid
phase. For every data reduction method, this fact presents a serious
limitation because excess enthalpy data are very scarce, generally
limited to narrow temperature ranges and, in addition, difficult
to predict. Model-free approaches do not constitute an exception
to this limitation. Lam et al. [153] showed application of the
model-free technique systematically to the treatment of isobaric
data. Hu et al. [3] treated binary, ternary, and quaternary data fol-
lowing Barker’s method.
reproduce from correlations developed from only P–x data than
from correlations based on the full set of P–x–y data [141]. It is dif-
ficult to take into account the unusual behavior of total pressure at
the terminal points of the mole fraction space in the model using
Barker’s method, since the weight of the end points is small com-
pared to the whole range. Therefore, the method of Barker is not
adequate for end points of composition range [37].
5. Conclusions
Despite some limitations, there is no doubt that Barker’s meth-
od is used very frequently for isothermal or isobaric VLE data
reduction. This is a helpful tool both for VLE experimentalists
and data analyzers to validate or to estimate activity coefficients
as well as vapor compositions data. To obtain reliable outcomes
from this technique, the choice of algorithm, objective function,
and excess Gibbs free energy model are extremely important.

Acknowledgement
4. Applications to higher-ordered systems and limitations
The authors are very grateful to the anonymous reviewer whose
Generally Barker’s method is widely applied to binary systems.
insightful suggestions helped a lot to finalize this article.
However, if computational complexities can be handled, then there
is no reason why Barker’s method cannot be used for ternary and
higher ordered systems. Of the methods reported in literature for References
P–x data reduction, Barker’s technique is easily generalized to ter- [1]
nary and higher ordered systems [151]. Fonseca and Lobo [154] [2]
outlined an extension of Barker’s method to ternary systems and [3]
[4]
also performed error analysis. They showed data regression apply-
[5]
ing to their experimentally measured (CH3F + N2O + Xe) ternary [6]
data. Oh and Campbell [155] showed the same for a (CuCl2 + - [7]
CH3OH + C2H5OH) mixture. For systems of salt and two solvents,
[8]
there is a good chance of widely differing temperatures for isobaric [9]
data and therefore data reduction would be complicated, unless it [10]
is assumed that the activity coefficients of the species in solution [11]
[12]
are independent of temperature or unless explicit temperature
dependence is incorporated in the solution model used. This kind [13]
of complexity does not exist for isothermal data, for which it is [14]
[15]
an excellent assumption to neglect the pressure dependence of
the activity coefficients. Barker’s method is a brilliant approach [16]
for regression of such types of data. The authors [141] also reduced [17]
data of three (salt + solvent + solvent) systems. To calculate equi- [18]
librium composition of the (NH3 + H2O + KOH) and (NH3 + H2O + - [19]
NaOH) ternary systems, Salavera et al. [156] extended Barker’s [20]
[21]
method [154] for their data reductions. Lam et al. [153] showed a
model-free approach for ternary data treatment based on Barker’s [22]
equation. Model free techniques are potentially useful for treating [23]
complex ternary systems. [24]
[25]
There are many other literature references [3,21,63, [26]
116,126,140,145,148,157–163] which show the application of Bar-
ker’s method to various ternary systems. Hu et al. [3] examined [27]
two quaternary systems such as (cyclohexane + benzene + 2-pro- [28]
panol + butanone) and (ethanol + chloroform + acetone + hexane)
at atmospheric pressure. Costa-Lopez et al. [143] applied this to [29]
multicomponent mixtures consisting (methanol + ethanol + [30]
methyl acetate + ethyl acetate + o-xylene).
Barker’s method makes use of only the P–x or T–x data. If the [31]
[32]
data are considered as more reliable than y values, this method [33]
generates the best fit of P values consistent with the correlating g
equation applied. The main challenge of Barker’s method is to [34]
choose an appropriate expression of g for a meaningful data reduc-
[35]
tion. Accurate measurement of P–x data assures reliable values of y,
provided the measured data are reproduced within the limits of [36]
experimental precision [2]. Data reduction procedures based on
[37]
only P–x data are superior to those which include measured y in
the data reduction. Nevertheless, measured y values are easier to
J.A. Barker, Aust. J. Chem. 6 (1953) 207–210.
M.M. Abbott, H.C. Van Ness, AIChE J. 21 (1975) 62–71.
Y. Hu, H. Liu, J.M. Prausnitz, Fluid Phase Equilib. 95 (1994) 73–92.
G. Wozny, H. Cremer, Fluid Phase Equilib. 6 (1981) 149–168.
K.W. Won, J.M. Prausnitz, Ind. Eng. Chem. Fundam. 12 (1973) 459–463.
M. Keller, S. Schnabel, A. Heintz, Fluid Phase Equilib. 110 (1995) 231–265.
J.M. Prausnitz, R.N. Lichtenthaler, E.G. Azevedo, Molecular Thermodynamics
of Fluid Phase Equilibria, 3rd ed., Prentice Hall, New York, 1999.
M.M. Abbott, H.C. Van Ness, Fluid Phase Equilib. 1 (1977) 3–11.
F. Kuschel, H. Kraetsch, H. Kehlen, H. Sackmann, AIChE J. 23 (1977) 205–207.
J.F. Smith, Fluid Phase Equilib. 78 (1992) 219–227.
H.C. Van Ness, F. Pederson, P. Rasmussen, AIChE J. 24 (1978) 1055–1063.
G. Shanker, C.S. Howat, R.R. Torres, G.W. Swift, Fluid Phase Equilib. 5 (1980/
1981) 305–321.
T.F. Anderson, D.S. Abrams, E.A. Grens II, AIChE J. 24 (1978) 20–29.
J.F. Fabries, H. Renon, AIChE J. 21 (1975) 735–743.
P.G. Guest, Numerical Methods of Curve Fitting, Cambridge University Press,
Cambridge, 1961.
A. Bertucco, M. Barolo, N. Elvassore, AIChE J. 43 (1997) 547–554.
C. Duce, M.R. Tine, L. Lepori, E. Matteoli, Fluid Phase Equilib. 199 (2002) 197–
212.
W. Regenstein, Talanta 40 (1993) 1019–1022.
D.V.S. Jain, O.P. Yadav, J. Chem. Thermodyn. 5 (1973) 541–544.
D.V.S. Jain, R.K. Wadi, J. Chem. Thermodyn. 8 (1976) 493–497.
S. Weiguo, A.X. Qin, P.J. McElroy, A.G. Williamson, J. Chem. Thermodyn. 22
(1990) 915–921.
S.A. Wieczorek, J. Stecki, J. Chem. Thermodyn. 10 (1978) 177–186.
S.A. Wieczorek, J. Chem. Thermodyn. 11 (1979) 239–245.
P. Oracz, G. Kolasinska, Fluid Phase Equilib. 35 (1987) 253–278.
H.T. French, J. Chem. Thermodyn. 19 (1987) 1151–1161.
R. Garriga, F. Sanchez, P. Perez, J. Valero, M. Gracia, Fluid Phase Equilib. 101
(1994) 227–236.
J.W. Kang, V. Diky, R.D. Chirico, J.W. Magee, C.D. Muzny, I. Abdulagatov, A.F.
Kazakov, M. Frenkel, J. Chem. Eng. Data 55 (2010) 3631–3640.
A. del Rio, B. Coto, C. Pando, J.A.R. Renuncio, Ind. Eng. Chem. Res. 40 (2001)
689–695.
M. Gracia, F. Sanchez, P. Perez, J. Valero, C.G. Losa, J. Chem. Thermodyn. 24
(1992) 463–471.
M. Gracia, I. Banos, P. Perez, J. Valero, C.G. Losa, J. Chem. Thermodyn. 24
(1992) 1237–1242.
A.W. Islam, A. Javvadi, V.N. Kabadi, Ind. Eng. Chem. Res. 50 (2011) 1034–1045.
R.G. Rubio, J.A.R. Renuncio, M.D. Pena, Thermochim. Acta 65 (1983) 69–79.
B. Coto, A. Cabanas, C. Pando, C. Menduina, R.G. Rubio, R.G. Juan, J. Chem. Soc.,
Faraday Trans. 91 (1995) 2779–2804.
P. Uusi-Kyyny, J.-P. Pokki, M. Laakkonen, J. Aittamaa, S. Liukkonen, Fluid
Phase Equilib. 201 (2002) 343–358.
T. Ouni, P. Uusi-Kyyny, J.-K. Pokki, J. Aittamaa, J. Chem. Eng. Data 49 (2004)
787–794.
J. Roininen, P. Uusi-Kyyny, J.-P. Pokki, M. Pakkanen, V. Alopaeus, J. Chem. Eng.
Data 53 (2008) 607–612.
J.-P. Pokki, Development of Vapor Liquid equilibrium Calculation Methods for
Chemical Engineering Design, PhD Thesis, Helsinki University of Technology,
2004.
36 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37
[38] P. Haimi, P. Uusi-Kyyni, J.-P. Pokki, V. Alopse, Fluid Phase Equilib. 299 (2010)
148–160.
[39] H. Holldorff, H. Knapp, Fluid Phase Equilib. 44 (1988) 195–209.
[40] A. Penttila, P. Uusi-Kyyny, J.-P. Pokki, M. Pakkanen, V. Alopse, J. Chem. Eng.
Data 55 (2010) 291–296.
[41] G. Wierink, J. Roininen, P. Uusi-Kyyny, J.-P. Pokki, M. Pakkanen, V. Alopse, J.
Chem. Eng. Data 53 (2008) 1539–1544.
[42] A. Sundberg, P. Uusi-Kyyny, M. Pakkanen, J.-P. Pokki, V. Alopse, J. Chem. Eng.
Data 54 (2009) 1311–1317.
[43] E. Matteoli, L. Lepori, J. Chem. Eng. Data 33 (1988) 247–250.
[44] K.C. Kalra, K.C. Sing, M. Soni, D.C. Spah, J. Chem. Thermodyn. 22 (1990) 771–
775.
[45] P. Kumar, K.C. Kalra, K.C. Sing, Thermochim. Acta 188 (1991) 151–156.
[46] J.D. Gomez-Ibanez, J.J.C. Shieh, E.M. Thorsteinson, J. Phys. Chem. 70 (1966)
1998–2002.
[47] D. Dragoescu, A. Barhala, M. Teodorescu, D. Chiscan, J. Serb. Chem. Soc. 76
(2011) 305–315.
[48] D. Dragoescu, A. Barhala, R. Vilcu, 2002. <http://gw-chimie.math.unibuc.ro/
anunivch/2002-1/AUBCh2002XI13744.pdf>.
[49] M. Gracia, P. Perez, J. Valero, Fluid Phase Equilib. 97 (1994) 147–153.
[50] Y. Azpiazu, F. Royo, C.G. Losa, J. Chem. Thermodyn. 16 (1984) 561–565.
[51] K.L. Young, R.A. Mentzer, R.A. Greenkorn, J. Chem. Thermodyn. 9 (1977) 979–
985.
[52] R. Garriga, P. Perez, M. Gracia, Fluid Phase Equilib. 227 (2005) 71–78.
[53] R. Garriga, P. Perez, M. Gracia, Fluid Phase Equilib. 227 (2005) 79–86.
[54] R.C. Pemberton, C.J. Mash, J. Chem. Thermodyn. 10 (1978) 867–888.
[55] K.N. Marsh, H.T. French, H.P. Rogers, J. Chem. Thermodyn. 11 (1979) 897–903.
[56] K. Aim, Fluid Phase Equilib. 2 (1978) 119–142.
[57] D.V.S. Jain, S.B. Saini, R.S. Sidhu, J. Chem. Thermodyn. 14 (1982) 689–693.
[58] T. Kasprzycka-Guttman, I. Chojnacka, J. Chem. Thermodyn. 21 (1989) 721–
725.
[59] J.C.G. Calado, E.J.S.G.D. Azevedo, K. Lucas, K.P. Shukla, Fluid Phase Equilib. 16
(1984) 171–183.
[60] S. Villa, R. Garriga, P. Perez, M. Gracia, J.A. Gonzalez, I.G. de la Fuente, J.C.
Cobbos, Fluid Phase Equilib. 231 (2005) 211–220.
[61] F. Amireche-Ziar, G. Boukais-Belaribi, A. Jakob, I. Mokbel, F.B. Belaribi, Fluid
Phase Equilib. 268 (2008) 39–44.
[62] R. Garriga, F. Sanchez, P. Perez, M. Gracia, J. Chem. Thermodyn. 29 (1997)
649–659.
[63] J. Zielkiewicz, J. Chem. Thermodyn. 38 (2006) 701–706.
[64] I. Mokbel, A. Blondel-Telouk, D. Vellut, J. Jose, Fluid Phase Equilib. 149 (1998)
287–308.
[65] A. Dahmani, A.A. Kaci, J. Jose, Fluid Phase Equilib. 130 (1997) 271–279.
[66] U. Domanska, J. Lachwa, J.A. Gonzalez, Fluid Phase Equilib. 235 (2005) 182–
190.
[67] S. Weiguo, A.X. Qin, P.J. McElroy, A.G. Williamson, J. Chem. Thermodyn. 22
(1990) 905–914.
[68] A. Sporzynski, M. Szurgocinska, U. Domanska, J.A. Gonzalez, Ind. Eng. Chem.
Res. 42 (2003) 4382–4388.
[69] J. Zielkiewicz, J. Chem. Thermodyn. 24 (1992) 1319–1322.
[70] M. Rogalski, K. Chylinski, B. Wisniewska, J. Chem. Thermodyn. 19 (1987) 57–
60.
[71] V. Gupta, S. Maken, K.C. Kalra, K.C. Singh, Fluid Phase Equilib. 120 (1996) 195–
203.
[72] R.K. Nigam, P.P. Singh, K.C. Singh, Thermochim. Acta 35 (1980) 1–10.
[73] B.R. Sharma, P.P. Singh, Thermochim. Acta 32 (1979) 213–223.
[74] J. Herraiz, S. Shen, A. Coronas, J. Chem. Eng. Data 43 (1998) 191–195.
[75] M.E. Alvarez, M. Bourouis, X. Esteve, J. Chem. Eng. Data 56 (2011) 491–496.
[76] D.F. Gray, N.F. Pasco, G. Williamson, J. Chem. Eng. Data 33 (1988) 335–337.
[77] R. Garriga, F. Sanchez, P. Perez, M. Gracia, J. Chem. Eng. Data 41 (1996) 451–
454.
[78] C. Yokoyama, L.-S. Lee, W. Morkoff, H.-M. Lin, K.-C. Chao, Ind. Eng. Chem. Res.
26 (1987) 1695–1698.
[79] R. Garriga, S. Martinez, P. Perez, M. Gracia, J. Chem. Eng. Data 47 (2002) 322–
328.
[80] R. Garriga, A. Cristina, P. Perez, M. Gracia, J. Chem. Eng. Data 44 (1999) 296–
302.
[81] S.G. d’Avila, R.S.F. Silva, J. Chem. Eng. Data 15 (1970) 421–424.
[82] R. Garriga, F. Sanchez, P. Perez, M. Gracia, J. Chem. Eng. Data 41 (1996) 1091–
1096.
[83] G. Luengo, J. Aracil, R.G. Rubio, M.D. Pena, J. Phys. Chem. 92 (1988) 228–234.
[84] U. Domanska, J. Lachwa, Ind. Eng. Chem. Res. 44 (2005) 5795–5804.
[85] J.D. Gomez-Ibanez, J.J.C. Shieh, J. Phys. Chem. 69 (1965) 1660–1666.
[86] R. Garriga, F. Sanchez, P. Perez, M. Gracia, J. Chem. Thermodyn. 28 (1996)
567–576.
[87] R. Garriga, J. Ilarraza, P. Perez, M. Gracia, J. Chem. Thermodyn. 28 (1996) 233–
243.
[88] R. Garriga, P. Perez, M. Gracia, J. Chem. Thermodyn. 27 (1995) 1057–1066.
[89] S. Maken, N. Verma, A. Gaur, H.-J. Song, K.C. Singh, J.-W. Park, Korean J. Chem.
Eng. 25 (2008) 273–278.
[90] P.P. Singh, B.R. Sharma, K.S. Sidhu, Aust. J. Chem. 31 (1977) 1419–1423.
[91] J.D. Olson, Int. J. Thermophys. 16 (1995) 215–226.
[92] R. Garriga, S. Martinez, P. Perez, M. Gracia, Fluid Phase Equilib. 147 (1998)
195–206.
[93] O.P. Yadav, J. Chem. Thermodyn. 21 (1989) 985–987.
[94] M. Gracia, F. Sanchez, P. Perez, J. Valero, C.G. Losa, J. Chem. Thermodyn. 24
(1992) 843–849.
[95] R. Gracia, P. Perez, M. Gracia, Fluid Phase Equilib. 227 (2005) 79–86.
[96] P. Perez, J. Valero, M. Gracia, C.G. Losa, J. Chem. Thermodyn. 21 (1989) 259–
264.
[97] J.A. Lopez, P. Perez, M. Gracia, C.G. Losa, J. Chem. Thermodyn. 20 (1988) 447–
451.
[98] R. Garriga, F. Sanchez, P. Perez, M. Gracia, Fluid Phase Equilib. 130 (1997)
195–206.
[99] R. Garriga, S. Martinez, P. Perez, M. Gracia, Fluid Phase Equilib. 149 (1998)
111–120.
[100] M.D. Pena, A. Compostizo, A.C. Colin, I. Escudero, J. Chem. Thermodyn. 12
(1980) 1051–1055.
[101] M.E. Fernandez, J. Noles, J.A. Zollweg, W.B. Streett, Fluid Phase Equilib. 36
(1987) 91–97.
[102] G.I. Kaminishi, C. Yokoyama, S. Takahashi, Fluid Phase Equilib. 34 (1987) 83–
99.
[103] B. Semeniuk, T. Kasprzycka-Guttman, J. Chem. Thermodyn. 20 (1988) 1427–
1431.
[104] P. Perez, J. Valero, M. Gracia, J. Chem. Eng. Data 39 (1994) 789–792.
[105] E. Chang, J.C.G. Calado, W.B. Streett, J. Chem. Eng. Data 27 (1982) 293–298.
[106] B.R. Sharma, P.P. Singh, J. Chem. Eng. Data 20 (1975) 360–363.
[107] Y. Sassa, R. Konishi, T. Katayama, J. Chem. Eng. Data 19 (1974) 44–48.
[108] M. Goral, A. Maczynski, G. Schmidt, H. Wenzel, Ind. Eng. Chem. Fundam. 20
(1981) 267–277.
[109] F.B. Sprow, J.M. Prausnitz, AIChE J. 12 (1966) 781–784.
[110] J. Nath, A.P. Dixit, Fluid Phase Equilib. 60 (1990) 205–212.
[111] J.R. Nath, R. Saini, Fluid Phase Equilib. 68 (1991) 281–290.
[112] J. Pardo, P. Perez, F.M. Royo, M. Gracia, C.G. Losa, J. Chem. Thermodyn. 19
(1987) 521–526.
[113] R. Garriga, S. Martinez, P. Perez, M. Gracia, Fluid Phase Equilib. 181 (2001)
203–214.
[114] A. Dahmani, I. Mokbel, J. Jose, Fluid Phase Equilib. 203 (2002) 193–204.
[115] J.A. Salas, E.L. Arancibia, M. Katz, Can. J. Chem. 75 (1997) 207–211.
[116] R.M. Villamanan, M.C. Martin, M.A. Villamanan, C.R. Chamorro, J.J. Segovia, J.
Chem. Eng. Data 55 (2010) 2741–2745.
[117] R. Garriga, S. Martinez, P. Perez, M. Gracia, J. Chem. Eng. Data 49 (2004) 224–
230.
[118] C.S. Howat, G.W. Swift, Fluid Phase Equilib. 21 (1985) 113–134.
[119] R. Garriga, F. Sanchez, P. Perez, M. Gracia, Fluid Phase Equilib. 138 (1997)
131–144.
[120] R. Garriga, F. Sanchez, P. Perez, M. Gracia, Fluid Phase Equilib. 124 (1996)
123–134.
[121] R. Garriga, S. Martinez, P. Perez, M. Gracia, J. Chem. Thermodyn. 36 (2004)
59–69.
[122] R. Garriga, S. Martinez, P. Perez, M. Gracia, Fluid Phase Equilib. 207 (2003)
97–109.
[123] S. Uno, K. Kurihara, K. Ochi, K. Kojima, Fluid Phase Equilib. 257 (2007) 139–
146.
[124] M.N.M. Al-Hayan, J.A.M. Al-Kandary, Fluid Phase Equilib. 239 (2006) 133–
137.
[125] B. Li, Z. Zhu, Y. Luo, Fluid Phase Equilib. 27 (1986) 491–500.
[126] D. Gao, D. Zhu, L. Zhang, H. Guan, H. Sun, H. Chen, J. Si, J. Chem. Eng. Data 55
(2010) 5887–5895.
[127] M. Kato, H. Tanaka, K. Aizawa, Fluid Phase Equilib. 58 (1990) 181–189.
[128] M.A. Villamanan, C. Gonzalez, H.C. Van Ness, J. Chem. Eng. Data 29 (1984)
427–429.
[129] S.W. Campbell, R.A. Wilsak, G. Thodos, J. Chem. Eng. Data 31 (1986) 424–430.
[130] J.R. Loehe, H.C. Van Ness, M.M. Abbott, J. Chem. Eng. Data 28 (1983) 405–407.
[131] K. Mara, V.R. Bhethanabotla, S.W. Campbell, Fluid Phase Equilib. 127 (1997)
147–153.
[132] P.C. Proust, A.L. Ramirez, J.B. Yianatos, J. Chem. Eng. Data 25 (1980) 329–331.
[133] D.P. Barton, V.R. Bhethanabotla, S.W. Campbell, J. Chem. Eng. Data 41 (1996)
1138–1140.
[134] S.W. Campbell, R.A. Wilsak, G. Thodos, J. Chem. Eng. Data 32 (1987) 357–362.
[135] J.M. Rhodes, V.R. Bhethanabotla, S.W. Campbell, J. Chem. Eng. Data 47 (1997)
731–734.
[136] F.J. Carmona, J.A. Gonzalez, I.G. Fuente, J.C. Cobos, V.R. Bhethanabotla, S.W.
Campbell, J. Chem. Eng. Data 45 (2000) 699–703.
[137] J.E. Myers, H.C. Hershey, W.B. Kay, J. Chem. Thermodyn. 11 (1979) 1019–
1028.
[138] M.A. Villamanan, A.J. Allwal, H.C. Van Ness, J. Chem. Eng. Data 29 (1984) 431–
435.
[139] V.R. Bhethanabotla, S.W. Campbell, Fluid Phase Equilib. 62 (1991) 239–258.
[140] L.M. Lozano, E.A. Montero, M.C. Martin, M.A. Villamanan, Fluid Phase Equilib.
110 (1995) 219–230.
[141] S.K. Oh, S.W. Campbell, Fluid Phase Equilib. 76 (1992) 237–278.
[142] M. Mukhopadhyay, K. Sahasranaman, Ind. Eng. Chem. Process Des. Dev. 21
(1982) 632–640.
[143] J. Costa-Lopez, A. Garvin, F.J. Espana, J. Chem. Eng. Data 40 (1995) 1067–1071.
[144] E.B. Munday, J.C. Mullins, D.D. Edie, J. Chem. Eng. Data 25 (1980) 191–194.
[145] L. Karrer, J. Gaube, Fluid Phase Equilib. 42 (1988) 195–207.
[146] C. Wang, H. Li, S. Han, Fluid Phase Equilib. 209 (2003) 147–154.
[147] M. Mukhopadhyay, K.R. Dongaonkar, Ind. Eng. Chem. Process Des. Dev. 22
(1983) 521–532.
 A.W. Islam, M.H. Rahman / J. Chem. Thermodynamics 44 (2012) 31–37
[148] R.A. Wilsak, S.W. Campbell, G. Thodos, Fluid Phase Equilib. 33 (1987) 173–
190.
[149] C.M. Smith, J.D.P. Visco, J. Chem. Eng. Data 49 (2004) 306–310.
[150] C. Pando, J.A.R. Renuncio, R.W. Hanks, J.J. Christensen, Ind. Eng. Chem. Process
Des. Dev. 23 (1984) 67–72.
[151] F.O. Mixon, B. Gumowski, B.H. Carpenter, Ind. Eng. Chem. Fundam. 4 (1965)
455–459.
[152] Y. Hu, X. Ying, J. Chem. Ind. Eng. China 31 (1980) 27–36.
[153] E. Lam, A. Mejia, H. Segura, J. Wisniak, S. Loras, Ind. Eng. Chem. Res. 40 (2001)
2149–2159.
[154] I.M.A. Fonseca, L.Q. Lobo, Fluid Phase Equilib. 154 (1999) 205–211.
[155] S.K. Oh, S.W. Campbell, J. Chem. Eng. Data 40 (1995) 504–508.
[156] D.S. Salavera, S.K. Chaudhuri, X. Exteve, A. Coronas, J. Chem. Eng. Data 50
(2005) 471–476.
37
[157] J.R. Loehe, H.C. Van Ness, M.M. Abbot, J. Chem. Eng. Data 26 (1981) 178–181.
[158] S.K. Oh, J. Chem. Eng. Data 42 (1997) 1082–1084.
[159] J.L. Relmers, V.R. Bhethanabolta, S.W. Campbell, J. Chem. Eng. Data 37 (1992)
127–130.
[160] C. Alonso, E.A. Montero, C.R. Chamorro, J.J. Segovia, M.C. Martin, M.A.
Villamanan, Fluid Phase Equilib. 182 (2001) 241–245.
[161] M.M. Abbott, J.K. Floess, G.E. Walsh Jr., H.C.V. Ness, AIChE J. 21 (1975) 72–76.
[162] E. Ruckenstein, I. Shulgin, Fluid Phase Equilib. 180 (2001) 281–297.
[163] C.A. Tristan, J.J. Segovia, C.R. Chamorro, M.A. Villamanan, Fluid Phase Equilib.
245 (2006) 57–62.
JCT -11-293