Experimental measurement and modeling of vapor-liquid equilibrium for the ternary system water + ethanol + 1-Butyl-3-methylimidazolium chloride

Article abstract of DOI:10.1021/je900723v

Vapor-liquid equilibrium (VLE) data were measured for the ternary system water (1) + ethanol (2) + 1-butyl-3-methylimidazolium chloride ([bmim]Cl) (3). Complete T, x, and y data were obtained in a relatively wide range of ionic liquid (IL) mass fractions up to 0.8 and in a relatively complete composition range for the volatile binary pair. The nonrandom two-liquid (NRTL) equation was used for correlation and was revealed to be adequate for the ternary system in the experimental composition range. The ternary VLE behavior was also modeled by the correlation of two data sets, in which the ethanol mole fraction on IL-free basis is respectively at 0.1 and 0.98. In this way, the six data sets were reproduced satisfactorily, with root-mean-square deviations of 0.40 K for temperature and 0.0070 for vapor-phase mole fractions. Owing to the regular distribution of the experimental data, a good agreement between the experiment and the calculation was graphically presented. The effect of the IL on the VLE behavior of the volatile components was also illustrated.

Full text of DOI:10.1021/je900723v

J. Chem. Eng. Data 2010, 55, 1679–1683
1679
Experimental Measurement and Modeling of Vapor-Liquid Equilibrium for the
Ternary System Water + Ethanol + 1-Butyl-3-methylimidazolium Chloride
Wei Geng, Lianzhong Zhang,* Dongshun Deng, Yun Ge, and Jianbing Ji
College of Chemical Engineering and Material Science, Zhejiang University of Technology, Hangzhou 310014, China
Vapor-liquid equilibrium (VLE) data were measured for the ternary system water (1) + ethanol (2) +
1-butyl-3-methylimidazolium chloride ([bmim]Cl) (3). Complete T, x, and y data were obtained in a relatively
wide range of ionic liquid (IL) mass fractions up to 0.8 and in a relatively complete composition range for
the volatile binary pair. The nonrandom two-liquid (NRTL) equation was used for correlation and was
revealed to be adequate for the ternary system in the experimental composition range. The ternary VLE
behavior was also modeled by the correlation of two data sets, in which the ethanol mole fraction on IL-
free basis is respectively at 0.1 and 0.98. In this way, the six data sets were reproduced satisfactorily, with
root-mean-square deviations of 0.40 K for temperature and 0.0070 for vapor-phase mole fractions. Owing
to the regular distribution of the experimental data, a good agreement between the experiment and the
calculation was graphically presented. The effect of the IL on the VLE behavior of the volatile components
was also illustrated.
measurements. Therefore, only x and y data were reported. Zhao
et al.,5 Calvar et al.,7 and Ge et al.11 have reported T, x, and y
data for the system. However, the measurements were generally
limited to a relatively small range of IL mass fraction, especially
in the ethanol-rich region.
Introduction
In recent years, ionic liquids (ILs) have received considerable
attention for their use in the chemical industry and are
considered to be potential solvents for extractive distillation.1,2
With the addition of an IL, the activity coefficients of the
components to be separated change their values to a different
extent. It is desirable that the changes in activity coefficients
result in an increase of relative volatility. Therefore, the
performance of an IL depends mainly on the composition
dependence of the activity coefficients in the IL-containing
mixture. At the present stage, such information is mainly
obtained from experimental measurement and by correlation
using a model of excess Gibbs energy.3-8
Experimental Section
Materials. Water was double-distilled. Ethanol was an
analytical grade reagent and was used without further purifica-
tion. The purity of ethanol was checked by gas chromatography
(GC), and the result was 99.9 %, while Karl Fischer analysis
indicated a water mass fraction of 4.3 ·10-4. [bmim]Cl was
synthesized by the reaction of 1-chlorobutane with 1-meth-
ylimidazole under a nitrogen atmosphere. The purification of
[bmim][Cl] was by recrystallization in a mixture of ethyl acetate
+ acetonitrile. Before use, the IL was dried for 24 h under
vacuum at 353 K to remove volatile impurities. Karl Fischer
analysis showed typically a water mass fraction of 1.2 ·10-3 in
the IL. The IL was also checked by electrospray ionization mass
spectrometry (ESI-MS), showing a single positive ion, m/z )
139 ([bmim]).
Apparatus and Procedure. VLE data were measured by an
ebulliometer, which was described in detail in previous work.12,13
In the measurements, liquid-phase circulation was enhanced by
a pump-like stirrer and had a value of approximately 200
cm3 ·min-1. Vapor-phase circulation was maintained at ap-
proximately 1 cm3 ·min-1, while the vapor-to-liquid circulation
ratio calculated on mass basis had a value of 0.005. Vapor
condensate was cooled to 275 K. The volume of vapor material
is estimated to be 1.7 cm3, which is very small compared to the
total sample volume of 270 cm3 in the ebulliometer. For more
details, please refer to our previous work.12,13
In our previous work,9 we presented an procedure for the
experimental measurement and modeling of vapor-liquid
equilibrium of ternary systems containing ionic liquid, taking
the system water + ethanol + 1-hexyl-3-methylimidazolium
chloride as a case study. Six sets of complete T, x, and y data
were measured, in which two data sets were used for correlation
of the nonrandom two-liquid (NRTL) model10 and the other
four data sets were used for checking the model. On the basis
of detailed discussion, the modeling of vapor-liquid equilibrium
(VLE) behavior of a ternary system containing ILs has been
generally recommended by the correlation of two ternary data
sets, which are measured at a relatively wide range of IL mass
fraction, while the mole fractions of the volatile quasi-binary
pair are distributed respectively in the two diluted ends.
In this work, we have extended the experimental measurement
and modeling to the system water (1) + ethanol (2) + 1-butyl-
3-methylimidazolium chloride ([bmim]Cl) (3). The modeling
was based on T, x, and y data in a relatively wide range of IL
mass fraction up to 0.8 and in a relatively complete composition
range for the volatile binary pair. This system has been studied
by Jork et al.8 The headspace technique was used in the
Pressure was measured by a precision pressure gauge with
an uncertainty of ( 0.04 kPa. A 20 dm3 glass container was
used as a pressure buffer connected to the ebulliometer.
Temperature measurement was by a standard platinum ther-
mometer and a 6.5 digit multimeter. The uncertainty of the
* Corresponding author. Phone: +86 571 88320892. E-mail: zhanglz@
zjut.edu.cn.
10.1021/je900723v CCC: $40.75  2010 American Chemical Society
Published on Web 10/23/2009

1680 Journal of Chemical & Engineering Data, Vol. 55, No. 4, 2010
Table 1. VLE Data for the Ternary System Water (1) + Ethanol
(2) + [bmim]Cl (3)
resistance measurement was ( 8 mΩ, which is equivalent to
( 0.08 K for the temperature measurement.
p
T
The VLE measurements were performed in a way in which
the IL mass fraction, w3, changed from high to low, while x′2,
which is the mole fraction of ethanol on IL-free basis, remained
almost unchanged. At high IL mass fractions, namely, at w3 )
0.6, 0.7, and 0.8, the measurements were performed at subat-
mospheric pressures. As has been commonly recognized, the
activity coefficients in the liquid mixture depend strongly on
composition. Often, they depend only weakly on temperature
and very weakly on pressure. Therefore, the present measure-
ments will provide mainly the composition dependence of the
activity coefficients.
x′2
w3
y2
kPa
K
γ1
γ2
R2,1
x′2 ) 0.1
30.00
30.04
0.1031
0.0935
0.0951
0.0961
0.0970
0.0975
0.0980
0.0988
0.8005
0.6994
0.5999
0.5003
0.4002
0.3002
0.2002
0.1000
0.4061
0.3855
0.3858
0.3939
0.3977
0.4089
0.4202
0.4343
360.15
349.17
348.14
367.56
364.42
362.42
360.95
359.69
0.47
0.65
0.83
0.91
0.98
1.01
1.03
1.04
1.25
1.73
2.19
2.50
2.70
2.89
3.06
3.24
5.95
6.08
5.98
6.11
6.15
6.40
6.67
7.00
40.00
100.00
100.00
100.00
100.00
100.00
x′2 ) 0.2
30.00
30.00
0.2020
0.1927
0.1993
0.2000
0.2000
0.1997
0.1991
0.1990
0.8002
0.6992
0.6003
0.5004
0.4000
0.3002
0.2000
0.1003
0.5880
0.5637
0.5602
0.5394
0.5418
0.5318
0.5334
0.5393
358.93
346.23
344.85
363.67
360.51
358.49
357.19
356.32
0.40
0.60
0.79
0.91
0.98
1.06
1.08
1.08
1.01
1.43
1.77
1.92
2.08
2.14
2.21
2.27
5.64
5.41
5.12
4.68
4.73
4.55
4.60
4.71
40.00
At the beginning of the measurement, samples of water,
ethanol, and the IL were introduced into the ebulliometer. The
water contents of ethanol and the IL were determined by Karl
Fischer analysis. Every sample added in or taken out of the
ebulliometer was weighed by an electronic balance (Mettler-
Toledo AL204) with an uncertainty of ( 0.0002 g. Masses of
water and the other components added in the ebulliometer were
calculated. Therefore, we have the overall synthetic masses for
the first measurement. When equilibrium was established, the
vapor condensate was sampled and analyzed. As the IL is
nonvolatile, the vapor phase is composed of water and ethanol.
Vapor-phase composition was determined by analyzing the
water content using the Karl Fischer method. When the water
mole fraction is greater than 0.1, the sample was first mixed
with ethanol quantitatively, and the water content of the mixture
was analyzed. Vapor-phase composition was then calculated by
the ratio of mixing and the result of the water content. The
uncertainty of the vapor-phase composition was estimated to
be 0.0001 in water mole fraction or relatively 1 %, whichever
is the greater. Liquid-phase compositions were calculated on
the basis of mass balances, using a procedure presented in
previous work.12,14 The next measurement was carried out by
the replacement of a certain amount of the mixture in the boiler
with IL-free mixture of water and ethanol to keep x′2 almost
unchanged. The measurement was repeated until w3 became
close to 0.1.
100.00
100.00
100.00
100.00
100.00
x′2 ) 0.4
29.99
30.00
0.3998
0.3996
0.3997
0.3996
0.3995
0.3997
0.3997
0.4006
0.8000
0.6997
0.5997
0.5000
0.3996
0.2999
0.2017
0.1000
0.7746
0.7381
0.7025
0.6798
0.6591
0.6513
0.6397
0.6343
359.58
345.28
343.32
361.54
358.00
355.85
354.55
353.77
0.31
0.53
0.79
0.94
1.10
1.18
1.25
1.28
0.71
0.99
1.23
1.35
1.42
1.47
1.48
1.47
5.16
4.23
3.55
3.19
2.91
2.81
2.67
2.60
40.00
100.00
100.00
100.00
100.00
100.00
x′2 ) 0.6
30.00
30.01
0.5979
0.5988
0.6010
0.6011
0.6002
0.6000
0.6002
0.6000
0.7997
0.6998
0.5999
0.5003
0.4001
0.2999
0.1999
0.0999
0.8790
0.8472
0.8220
0.7928
0.7769
0.7587
0.7289
0.7074
362.02
347.01
344.37
361.81
357.61
355.04
353.42
352.52
0.24
0.46
0.71
0.94
1.12
1.28
1.49
1.62
0.53
0.75
0.95
1.06
1.15
1.19
1.18
1.16
4.88
3.71
3.07
2.54
2.32
2.10
1.79
1.61
40.00
100.00
100.00
100.00
100.00
100.00
x′2 ) 0.8
29.98
30.04
0.8012
0.8003
0.8001
0.7999
0.8000
0.7995
0.7998
0.7994
0.7996
0.6999
0.6000
0.4999
0.4007
0.2996
0.2000
0.0999
0.9514
0.9365
0.9161
0.8977
0.8833
0.8686
0.8540
0.8411
364.73
349.89
346.56
363.09
358.19
355.02
353.07
351.86
0.19
0.36
0.63
0.90
1.16
1.41
1.63
1.81
0.41
0.58
0.75
0.88
0.98
1.04
1.06
1.06
4.85
3.68
2.73
2.19
1.89
1.66
1.46
1.33
40.00
100.00
100.00
100.00
100.00
100.00
x′2 ) 0.98
30.00
30.00
0.9778
0.9790
0.9794
0.9797
0.9799
0.9797
0.9796
0.9799
0.8002
0.7001
0.6000
0.5003
0.4001
0.2999
0.1999
0.1000
0.9953
0.9943
0.9919
0.9889
0.9874
0.9854
0.9834
0.9813
367.97
352.31
348.71
364.55
359.34
355.48
353.25
351.93
0.15
0.29
0.55
0.93
1.21
1.54
1.82
2.12
0.33
0.47
0.63
0.77
0.87
0.95
1.00
1.01
4.80
3.73
2.58
1.86
1.61
1.40
1.23
1.08
Results and Discussion
40.00
100.00
100.00
100.00
100.04
100.00
The experimental VLE data for the ternary system water (1)
+ ethanol (2) + [bmim]Cl (3) are listed in Table 1. The
measurements were performed in a way in which w3 varied from
0.8 to 0.1, while x′2 remained almost unchanged. There are six
data sets in Table 1, at respectively x′2 ) 0.1, 0.2, 0.4, 0.6, 0.8,
and 0.98. The six data sets of forty-eight data points were
regularly distributed at eight w3 and six x′2. In the measurements
the pressure was kept, respectively, at 30 kPa for w3 ) 0.8 and
0.7, at 40 kPa for w3 ) 0.6, and at 100 kPa for w3 ) 0.5 down
to 0.1. Therefore, the pressure was practically the same for a
given w3 at different x′2. Complete T, x, and y data were
obtained. The liquid-phase compositions were reported in x′2
and w3. Activity coefficients of water (γ1) and ethanol (γ2), as
well as the relative volatility of ethanol to water (R2,1), were
also reported. In the calculation of the activity coefficients the
vapor phase was regarded as an ideal gas, and the vapor
pressures were calculated by parameters in the literature.15
Modeling of the ternary VLE was first performed by the
correlation of the six data sets, using the NRTL equation.10 In
the correlation, the binary parameters for water + ethanol were
taken from literature.16 The suitability of the binary parameters
was discussed in previous work.9 While the values of nonran-
domness factors R13 and R23 were set as 0.3, binary parameters
for water + [bmim]Cl and ethanol + [bmim]Cl were obtained
by the minimization of the following objective function:
N
F )
(γ1,cal/γ1,exp - 1)2/N +
∑
ꢀ
n)1
N
In the measurements, the temperature stability was better than
0.04 K. As discussed in our previous work,9,13 the uncertainty
for the reported w3 was estimated to be ( 0.003. The uncertainty
of the molar composition of the volatile binary pair, x′2 or x′1,
was estimated to be less than relatively 1 %.
(γ2,cal/γ2,exp - 1)2/N (1)
∑
ꢀ
n)1
in which N is the number of data points. The obtained
parameters were used for the calculation of the ternary VLE

Journal of Chemical & Engineering Data, Vol. 55, No. 4, 2010 1681
Table 2. Root-Mean-Square Deviations, δT and δy, in the
Calculation of VLE Data of Water (1) + Ethanol (2) + [bmim]Cl
(3), Based on Correlations by the NRTL Equation
root mean square deviationsa
data sets used
in correlation
δT/Kb
δyc
six data sets in Table 1
x′2 ) 0.1 and 0.98
0.40
0.40
0.0062
0.0070
a Deviations are calculated for the six data sets. b δT/K ) (∑(Tcal/K -
Texp/K)2/N)1/2 c δy ) (∑(y2,cal - y2,exp)2/N)1/2
.
.
Table 3. Energy Parameters, ∆gij and ∆gji, and Nonrandomness
Factors, rij, for the NRTL Model Obtained from the Correlation of
Ternary VLE Data of Water (1) + Ethanol (2) + [bmim]Cl (3),
Using Data Sets at x′2 ) 0.1 and x′2 ) 0.98
binary parametersa
∆gij
∆gji
component i
component j
J ·mol-1
J·mol-1
Rij
water
ethanol
[bmim]Cl
[bmim]Cl
–5270.6
–2449.9
–7372.2
–6457.3
0.3
0.3
a Binary parameters for water (1) + ethanol (2) were taken from ref
16 and were fixed as ∆g12/J·mol-1 ) 4458.8 + 8.4420(T/K), ∆g21/
J·mol-1 ) -3791.4 + 4.1451(T/K), and R12 ) 0.1448. The objective
function used in the optimization was shown in eq 1.
data in comparison with experimental values. Results are given
in Table 2, in which δT and δy are respectively the root-mean-
square deviations of temperature and vapor-phase mole fractions.
The NRTL equation, with the nonrandomness factors R13 and
R23 having the most common value of 0.3, appeared suitable
for the ternary system in the experimental composition range,
with δT ) 0.40 K and δy ) 0.0062.
Following a procedure recommended in previous work9 for
modeling of ternary VLE behavior, two data sets at x′2 ) 0.1
and x′2 ) 0.98 were correlated. The optimized binary parameters
are given in Table 3. The parameters obtained were used for
calculation in comparison with the six sets of experimental data.
The results were quite good, with δT ) 0.40 K and δy ) 0.0070.
These deviations are very close to those by direct correlation
of the six data sets. Therefore, the two data sets at x′2 ) 0.1
and x′2 ) 0.98 appear to be adequate for modeling the VLE
behavior in the experimental composition range.
Figure 1. Experimental and calculated activity coefficients of (a) water,
γ1, and (b) ethanol, γ2, in relation with ethanol mole fraction on IL-free
basis, x′2, for the saturated mixture water (1) + ethanol (2) + [bmim]Cl
(3): O, w3 ) 0.1, p ) 100 kPa; 0, w3 ) 0.3, p ) 100 kPa; ∆, w3 ) 0.5, p
) 100 kPa; b, w3 ) 0.6, p ) 40 kPa; 9, w3 ) 0.7, p ) 30 kPa; 2, w3 )
0.8, p ) 30 kPa. Lines were calculated by NRTL parameters in Table 3:
solid lines, calculated values at w3 ) 0.1, 0.3, 0.5, 0.6, 0.7, and 0.8,
respectively, and at relevant pressures; dashed line, calculated values for
the system water (1) + ethanol (2) at p ) 100 kPa.
VLE results calculated by using parameters in Table 3 are
also shown in Figures 1, 2, and 3, in comparison with the
experimental values. In Figure 1, γ1 and γ2 are shown in relation
with x′2. As the experimental data were measured regularly at
w3 ) 0.8 to 0.1, in an interval of 0.1, the calculated activity
coefficients, in relation with x′2, can be compared with the
experimental values at several fixed IL mass fractions. Good
agreement can be observed. For the best illustration, typical
results at w3 ) 0 (no IL), 0.1, 0.3, 0.5, 0.6, 0.7, and 0.8 are
presented. While γ2 decreases with increasing x′2 at all given
w3, γ1 varies with x′2 in a more complicated manner. When the
IL mass fraction is high, namely, at w3 ) 0.6, 0.7, and 0.8, γ1
decreases rapidly with increasing x′2. At the same time, γ1
increases with increasing x′2 at w3 ) 0.1 and 0.3, while having
a maximum with changing x′2 for w3 ) 0.5. These trends are
similar to those presented for the system water (1) + ethanol
(2) + [hmim]Cl.9 On the other hand, composition dependence
of the activity coefficients on the IL mass fraction can also be
observed in Figure 1. Generally, both γ1 and γ2 decrease with
increasing w3. While the decrease of γ1 is beneficial for the
enhancement of the relative volatility, the decrease of γ2 is not
desirable. The effect of the IL on the relative volatility of ethanol
to water, R2,1, is shown in Figure 2. While the addition of the
IL results in an increase of relative volatility in the ethanol-
rich region, showing a salting-out effect, there is a salting-in
effect in the water-rich region. Moreover, T, x′2, and y2 relations
at typically w3 ) 0.7 and 0.8 are shown in Figure 3. Such an
illustration is realized because we have set the pressure at
practically the same value for a given w3 at different x′2 in the
measurements. While the boiling temperature of water is higher
than ethanol, the boiling temperature of water + [bmim]Cl
becomes lower than that of ethanol + [bmim]Cl at w3 ) 0.8.
Although the azeotrope of water + ethanol has been removed
by addition of the IL, there is a minimum for respectively the
bubble line and the dew line at a given w3.
VLE data for the same ternary system at p ) 101.3 kPa have
been reported by Calvar et al.7 By using the parameters proposed
in Table 3, the literature ternary data were calculated, showing
root-mean-square deviations of δT ) 2.97 K and δy ) 0.0410.
The calculated temperatures are generally lower than the
reported values. In the original literature, all of the nine binary
parameters for the ternary mixture were optimized. The reported
deviations were δT ) 1.18 K and δy ) 0.06. The reported
deviations in vapor mole fractions are larger than those we

1682 Journal of Chemical & Engineering Data, Vol. 55, No. 4, 2010
the parameters in Table 3, calculations of the VLE data show
root-mean-square deviations of δT ) 0.55 K and δy ) 0.0013.
The same ternary system has also been studied by Jork et al.8
using the headspace technique. Vapor compositions were
reported at 363.15 K and at three IL mole fractions, namely, at
x3 ) 0.1, 0.3, and 0.5. The reported uncertainty in vapor mole
fractions was 0.015. At low IL mole fractions, namely, at x3 )
0.1, calculations using the parameters in Table 3 show good
agreement, with δy ) 0.0082. However, relatively large
deviations can be observed at high IL concentration, with δy )
0.0290 for all of the three IL mole fractions. The calculated
vapor mole fractions of ethanol are smaller than the reported
values.
Conclusions
T, x, and y data were measured for the ternary system water
(1) + ethanol (2) + [bmim]Cl (3). Six data sets of 48 data
points were obtained, which were regularly distributed at w3
) 0.1 to 0.8, in an interval of 0.1, and at x′2 ) 0.1, 0.2, 0.4,
0.6, 0.8, and 0.98. The pressure was kept, respectively, at
30 kPa for w3 ) 0.8 and 0.7, at 40 kPa for w3 ) 0.6, and at
100 kPa for w3 ) 0.5 down to 0.1. The NRTL equation was
used for correlation and was revealed to be adequate for the
ternary system in the experimental composition range. By
correlating the two data sets respectively at x′2 ) 0.1 and
0.98, all of the six data sets were reproduced satisfactorily,
with δT ) 0.40 K and δy ) 0.0070. Owing to the regular
distribution of the experimental data, a good agreement
between the experiment and the calculation was graphically
presented. The effect of the IL on the VLE behavior of the
volatile components was also illustrated.
Figure 2. Experimental and calculated relative volatility of ethanol to water,
2,1, in relation with ethanol mole fraction on IL-free basis, x′2, for the
R
saturated mixture water (1) + ethanol (2) + [bmim]Cl (3): O, w3 ) 0.1, p
) 100 kPa; 0, w3 ) 0.3, p ) 100 kPa; ∆, w3 ) 0.5, p ) 100 kPa; b, w3
) 0.6, p ) 40 kPa; 9, w3 ) 0.7, p ) 30 kPa; 2, w3 ) 0.8, p ) 30 kPa.
Lines were calculated by NRTL parameters in Table 3: solid lines, calculated
values at w3 ) 0.1, 0.3, 0.5, 0.6, 0.7, and 0.8, respectively, and at relevant
pressures; dashed line, calculated values for the system water (1) + ethanol
(2) at p ) 100 kPa.
Literature Cited
(1) Seiler, M.; Jork, C.; Karvanou, A.; Arlt, W.; Hirsch, R. Separation of
azeotropic mixtures using hyperbranched polymers or ionic liquids.
AIChE J. 2004, 50, 2439–2454.
(2) Marsh, K. N.; Boxall, J. A.; Lichtenthaler, R. Room temperature ionic
liquids and their mixtures-a review. Fluid Phase Equilib. 2004, 219,
93–98.
(3) Lei, Z.; Arlt, W.; Wasserscheid, P. Separation of 1-hexene and
n-hexane with ionic liquids. Fluid Phase Equilib. 2006, 241, 290–
299.
(4) Doeker, M.; Gmehling, J. Measurement and prediction of vapor-liquid
equilibria of ternary systems containing ionic liquids. Fluid Phase
Equilib. 2005, 227, 255–266.
(5) Zhao, J.; Dong, C.; Li, C.; Meng, H.; Wang, Z. Isobaric vapor-liquid
equilibria for ethanol-water system containing different ionic liquids
at atmospheric pressure. Fluid Phase Equilib. 2006, 242, 147–153.
(6) Orchilles, A. V.; Miguel, P. J.; Vercher, E.; Martinez-Andreu, A.
Isobaric vapor-liquid equilibria for ethyl acetate + ethanol + 1-ethyl-
3-methylimidazolium trifluoromethanesulfonate at 100 kPa. J. Chem.
Eng. Data 2007, 52, 2325–2330.
Figure 3. Experimental and calculated relations of the T, x′2, and y2 diagram
for the system water (1) + ethanol (2) + [bmim]Cl (3) at p ) 30.0 kPa: O,
b, w3 ) 0.7; 0, 9, w3 ) 0.8. Lines were calculated by NRTL parameters
in Table 3 at p ) 30 kPa and at w3 ) 0.7 and 0.8, respectively: Solid
points and solid lines represent the T, x′2 relation; hollow points and dashed
lines represent the T and y2 relation.
(7) Calvar, N.; Gonzalez, B.; Gomez, E.; Dominguez, A. Vapor-liquid
equilibria for the ternary system ethanol + water + 1-butyl-3-
methylimidazolium chloride and the corresponding binary systems at
101.3 kPa. J. Chem. Eng. Data 2006, 51, 2178–2181.
(8) Jork, C.; Seiler, M.; Beste, Y. A.; Arlt, W. Influence of ionic liquids
on the phase behavior of aqueous azeotropic systems. J. Chem. Eng.
Data 2004, 49, 852–857.
(9) Zhang, L.; Ge, Y.; Ji, D.; Ji, J. Experimental measurement and
modeling of vapor-liquid equilibrium for ternary systems containing
ionic liquids: a case study for the system water + ethanol + 1-hexyl-
3-methylimidazolium chloride. J. Chem. Eng. Data 2009, 54, 2322–
2329.
(10) Renon, H.; Prausnitz, J. M. Local compositions in thermodynamic
excess functions for liquid mixtures. AIChE J. 1968, 14, 135–144.
(11) Ge, Y.; Zhang, L.; Yuan, X.; Geng, W.; Ji, J. Selection of ionic liquids
as entrainers for separation of (water + ethanol). J. Chem. Thermodyn.
2008, 40, 1248–1252.
(12) Zhang, L.; Deng, D.; Han, J.; Ji, D.; Ji, J. Isobaric vapor-liquid
equilibria for water + 2-propanol + 1-butyl-3-methylimidazolium
tetrafluoroborate. J. Chem. Eng. Data 2007, 52, 199–205.
calculated. When, on the other hand, the reported parameters
are used for the calculation of the six data sets in this work, the
deviations are δT ) 3.87 K and δy ) 0.0175, which appear to
be much larger than those we presented by using parameters in
Table 3. It should be also noted that the ternary VLE data
reported in this work, as compared with the literature data, are
in a relatively wide range of IL mass fraction, especially in the
ethanol-rich region. The same ternary system has been studied
by Zhao et al.,5 reporting data at p ) 101.32 kPa and at w3 ≈
0.2 and 0.3. By using parameters proposed in Table 3, the
literature data were calculated, showing root-mean-square
deviations of δT ) 0.54 K and δy ) 0.0460. The deviations of
vapor mole fractions are relatively large in the water-rich region.
In our previous work, we have measured VLE data for the same
system at x′2 ) 0.95 and IL mass fractions up to 0.59. By using

Journal of Chemical & Engineering Data, Vol. 55, No. 4, 2010 1683
(13) Zhang, L.; Han, J.; Deng, D.; Ji, J. Selection of ionic liquids as
entrainers for separation of water and 2-propanol. Fluid Phase Equilib.
2007, 255, 179–185.
(16) Gmehling, J.; Onken, U.; Rearey-Nies, J. R. Vapor-Liquid Equilibrium
Data Collection; DECHEMA: Frankfurt, 1988; Vol. I, Part Ib.
(14) Zhang, L.; Yuan, X.; Qiao, B.; Qi, R.; Ji, J. Isobaric vapor-liquid
equilibria for water + ethanol + ethyl acetate + 1-butyl-3-methylimi-
dazolium acetate at low water mole fractions. J. Chem. Eng. Data
2008, 53, 1595–1601.
(15) Reid, R.; Prausnitz, J.; Poling, B. The Properties of Gases and Liquids;
McGraw-Hill: New York, 1987; Appendix A.
Received for review September 3, 2009. Accepted October 12, 2009.
The authors gratefully appreciate the financial support by the National
Natural Science Foundation of China (20776132).
JE900723V