You are on page 1of 14

Fluid Phase Equilibria 202 (2002) 263–276

Liquid–liquid equilibria data for systems containing aromatic


+ nonaromatic + sulfolane at 308.15 and 323.15 K
Ricardo Rappel, Luiz Mário Nelson de Góis∗ , Silvana Mattedi
Chemical Engineering Department, Federal University of Bahia, R. Aristides Novis,
2 Federação, 40.210-630 Salvador, Bahia, Brazil
Received 21 August 2000; accepted 1 May 2002

Abstract
Liquid–liquid equilibrium data are presented for the ternary systems: sulfolane + p-xylene + cyclohexane,
sulfolane + p-xylene + n-hexane and sulfolane + toluene + n-hexane at 308.15 and 323.15 K. Tie-line data were
correlated by Hand and Othmer–Tobias methods with correlation coefficients near unity. Parameters for NRTL
and UNIQUAC activity coefficient models were estimated. The NRTL model fits slightly better the liquid–liquid
equilibrium data of the systems studied. The root mean square deviations (RMSDs) obtained comparing calculated
and experimental two-phase compositions are <1.61% for NRTL model and <2.1% for UNIQUAC model.
© 2002 Elsevier Science B.V. All rights reserved.
Keywords: Data; Liquid–liquid equilibria; Sulfolane

1. Introduction

The aromatics hydrocarbon (e.g. benzene, toluene and xylenes (BTX)) extraction process is of great
importance for the petrochemical industry because of the large use of these compounds in several pro-
cesses. The kind of solvent employed to separate aromatics from aliphatic streams plays a fundamental
role in the achievement of high purity levels of the desired products. The quality of the solvent used in this
operation is essential, mainly in what concerns the efficiency of the desired separation. The sulfolane is
one of the most appropriated solvent for this separation due to its physical–chemical properties. Sulfolane
is widely used as a solvent in aromatics extraction from aliphatic streams, liquid–liquid equilibrium data
for systems including aromatics, nonaromatics and sulfolane.
Some liquid–liquid equilibrium data for systems containing sulfolane as a solvent for BTX in systems
containing aliphatic C6–C8 hydrocarbon are presented in literature, including ternary data [1–10] or more
components systems as quaternary or quinary data [9–11]. Some authors [1–5,8–11] used an experimental

Corresponding author. Tel./fax: +55-71-2475123.
E-mail address: 1mario@ufba.br (L.M.N. de Góis).

0378-3812/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 3 8 1 2 ( 0 2 ) 0 0 1 3 8 - 3
264 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

technique consisting of an equilibrium cell in which the heterogeneous ternary mixture is shaken, and after
being decanted, samples of both phases are taken and analyzed (for example via gaseous chromatography).
Other authors used a titration technique: heterogeneous ternary mixtures are titrated with a homogeneous
mixture, when the binodal is reached, the heterogeneous mixture becomes homogeneous and changes
from turbid to clear [5]. In most of the above literature, data were correlated using the activity coefficient
models NRTL [12] and UNIQUAC [13].
In this work, an equilibrium cell was used and liquid–liquid equilibrium data are presented for three
different ternary systems: sulfolane(3) + p-xylene(2) + cyclo-hexane(1) (system a), sulfolane(3) +
p-xylene(2) + n-hexane(1) (system b) and sulfolane(3) + toluene(2) + n-hexane(1) (system c). The
liquid–liquid equilibrium data were obtained at 308.15 and 323.15 K. The experimental data are all
correlated by the Hand and Othmer–Tobias models and the experimental tie lines are showed by the
binodal curves. Parameters for NRTL and UNIQUAC activity coefficient models were estimated.

Table 1
Experimental data for the system a: sulfolane + p-xylene + cyclo-hexane
Temperature (K) Top phase (raffinate) Bottom phase (extract)

Sulfolane p-Xylene Cyclo-hexane Sulfolane p-Xylene Cyclo-hexane

308.15 0.0080 0.3185 0.6735 0.9477 0.0352 0.0171


0.0113 0.3986 0.5901 0.9331 0.0509 0.0160
0.0220 0.6045 0.3735 0.8829 0.0948 0.0223
0.0227 0.7105 0.2668 0.8230 0.1542 0.0228
0.0230 0.5808 0.3962 0.8898 0.0917 0.0185
323.15 0.0082 0.3085 0.6833 0.9312 0.0367 0.0321
0.0209 0.4074 0.5717 0.9095 0.0649 0.0256
0.0224 0.5807 0.3969 0.8691 0.1071 0.0238
0.0402 0.6944 0.2654 0.7904 0.1777 0.0319
0.0438 0.5873 0.3689 0.8532 0.1149 0.0319

Table 2
Experimental data for the system b: sulfolane + p-xylene + n-hexane
Temperature (K) Top phase (raffinate) Bottom phase (extract)

Sulfolane p-Xylene n-Hexane Sulfolane p-Xylene n-Hexane

308.15 0.0081 0.4852 0.5067 0.9530 0.0420 0.0050


0.0087 0.7396 0.2517 0.9050 0.0908 0.0042
0.0112 0.5717 0.4171 0.9382 0.0560 0.0058
0.0149 0.7645 0.2206 0.8900 0.1022 0.0078
0.0048 0.4093 0.5859 0.9457 0.0340 0.0203
323.15 0.0066 0.4411 0.5523 0.9526 0.0394 0.0080
0.0083 0.4831 0.5086 0.9476 0.0451 0.0073
0.0094 0.6186 0.3720 0.9174 0.0737 0.0089
0.0151 0.6879 0.2970 0.8954 0.0962 0.0084
0.0161 0.6344 0.3495 0.9074 0.0827 0.0099
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 265

2. Experimental part

2.1. Materials

The p-xylene, the toluene and the sulfolane used had purity superior to 99.5%, while the nonaromatics
cyclo-hexane and n-hexane had a minimum of purity of 99%. All the chemicals were used directly
without additional purification. The sulfolane used was from the Philips Chemical Company, the aliphatic
compounds from Merck Company and the aromatics from Copene Petroquı́mica do Nordeste (Brazil).

2.2. Procedure

The experimental apparatus used includes basically a jacketed glass cell, a magnetic agitator, a thermo-
statically controlled bath and a gas chromatography. The glass equilibrium cell volume is 193 cm3 . The
cell consists of a jacketed container with two samples points and a removable cap in which a thermometer
with precision of ±0.1 ◦ C (0.1 K) was adapted. All the cell’s joints are polished, conferring an adequate
sealing. The sample points are at the top and bottom parts of the cell, so samples of the two phases of

Table 3
Experimental data for the system c: sulfolane + toluene + n-hexane
Temperature Top phase (raffinate) Bottom phase (extract)

(K) Sulfolane Toluene n-Hexane Sulfolane Toluene n-Hexane

308.15 0.0005 0.4548 0.5448 0.8954 0.0970 0.0076


0.0008 0.3624 0.6368 0.9159 0.0755 0.0085
0.0008 0.5892 0.4100 0.8598 0.1305 0.0097
0.0010 0.2881 0.7109 0.9160 0.0495 0.0345
0.0648 0.6167 0.3186 0.8234 0.1656 0.0110
323.15 0.0014 0.2842 0.7144 0.9539 0.0392 0.0069
0.0011 0.4589 0.5400 0.9127 0.0790 0.0083
0.0017 0.6547 0.3435 0.7955 0.1885 0.0160
0.0017 0.6630 0.3353 0.7981 0.1888 0.0131
0.0022 0.5785 0.4193 0.8601 0.1287 0.0112

Table 4
Hand and Othmer–Tobias values of coefficient of correlation (r2 )
Systems Temperature (K) r2 (Hand) r2 (Othmer–Tobias)
a 308.15 0.995 0.993
a 323.15 0.995 0.987
b 308.15 0.997 0.971
b 323.15 0.994 0.989
c 308.15 0.998 0.993
c 323.15 0.991 0.989
266 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

Table 5
Estimated NRTL Parameters for the systems a–c and their respective RSMD values
System Temperature (K) i j Aij × 102 Aj i × 102 α ij RMSD
a 308.15 1 2 −5.279 −5.647 0.2 1.005
1 3 13.28 4.274 0.2
2 3 7.666 −0.4420 0.2
a 323.15 1 2 −10.10 1.116 0.2 1.091
1 3 12.42 3.533 0.2
2 3 7.242 −1.619 0.2
b 308.15 1 2 −11.88 −2.938 0.2 1.603
1 3 9.961 4.172 0.2
2 3 7.726 −0.5691 0.2
b 323.15 1 2 −10.68 −2.938 0.2 1.010
1 3 9.832 4.246 0.2
2 3 7.419 −0.6225 0.2
c 308.15 1 2 −4.093 −2.565 0.2 1.585
1 3 13.29 4.246 0.2
2 3 7.601 −0.7047 0.2
c 323.15 1 2 −4.803 −2.557 0.2 1.452
1 3 11.38 5.558 0.2
2 3 8.556 −1.147 0.2

Table 6
Estimated UNIQUAC parameters for the systems a–c and their respective RSMD values
System Temperature (K) I j Aij × 102 Aij × 102 i ri qi RMSD
a 308.15 1 2 −1.021 −2.421 1 4.0464 3.2400 1.063
1 3 3.859 0.6257 2 4.6578 3.5360
2 3 2.844 −0.5207 3 4.0358 3.2000
a 323.15 1 2 −1.377 −0.1501 1 4.0464 3.2400 1.564
1 3 4.010 1.192 2 4.6578 3.5360
2 3 2.359 0.1561 3 4.0358 3.2000
b 308.15 1 2 −3.898 −0.8380 1 4.4998 3.8560 1.219
1 3 5.012 −0.4624 2 4.6578 3.5360
2 3 2.505 −0.3068 3 4.0358 3.2000
b 323.15 1 2 −3.610 −0.6266 1 4.4998 3.8560 1.747
1 3 2.658 1.211 2 4.6578 3.5360
2 3 2.327 −0.1068 3 4.0358 3.2000
c 308.15 1 2 −1.538 −0.3167 1 4.4998 3.8560 2.037
1 3 4.816 1.853 2 3.9228 2.9680
2 3 1.336 −0.0128 3 4.0358 3.2000
c 323.15 1 2 −0.4484 −0.2908 1 4.4998 3.8560 1.131
1 3 10.02 7.488 2 3.9228 2.9680
2 3 1.490 0.3184 3 4.0358 3.2000
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 267

the mixture can be collected, the phase with higher density and the phase with lower density. In order to
maintain a constant temperature, the cell is immersed in a thermostatically controlled bath.
In the experiments the mixture was added in order to fill the maximum possible volume of the cell
considering included the shaker bar and the thermometer, leaving a minimum possible vapor space. It
was agitated for 2.5 h. Afterwards, it was decanted for another 2 h. The agitating and decanting times
were obtained through experimental tests done in each one of the systems.
After decanting, samples from the two phases were taken for analysis. The samples were diluted in
pure acetone and analyzed by a gas chromatography equipped with a ionization flame detector. A good
separation of the components was attained using a DB-1 capillary column (15 mm × 0.53 mm) with
a polydimethylsiloxane of 1.0 ␮m film. The injector and detector were maintained at the temperatures
of 493.15 and 523.15 K, respectively, while the column temperature was programmed for an initial
temperature of 328.15 K and a final of 413.15 K.

Fig. 1. Experimental tie lines, equilibrium curve and plait-point calculated by NRTL for system a (sulfolane + p-xylene +
cyclo-hexane) at 308.15 K.
268 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

3. Experimental results

Tables 1–3 present the equilibrium compositions of each phase (extract and raffinate) at 308.15 and
323.15 K for the three studied systems. The data are shown as mass fractions.
In order to determine the standard deviation for the chromatograph, there was prepared a mixture
of sulfolane + toluene + n-hexane and 12 analyses for both phases were done. The obtained standard
deviation for each component in both phase were always <0.005, that is a satisfactory value according
to Prauznitz et al. [14]. Experimental runs with measured composition for any component with standard
deviation >0.005 were neglected.
Hand [15] and Othmer–Tobias [16] correlations were used to ensure the quality of the obtained exper-
imental data. These correlations relate the tie lines concentration of the top phase with the bottom phase

Fig. 2. Experimental tie lines, equilibrium curve and plait-point calculated by NRTL for system b (sulfolane + p-xylene +
n-hexane) at 308.15 K.
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 269

to obtain a linear function. In Table 4 are shown the values of the correlation coefficients (r2 ) for each
system studied. The values of the obtained correlation coefficients were all very close to unity, revealing
a good data adjust. Observing the data of Table 4, Hand’s correlation makes a better description of these
systems as noted in r2 values, which are closer to unity than Othmer–Tobias values.
Hand’s correlation in logarithmic scale presents a linear dependency, being written as follows:
   
XAR XAE
log = a log +b (1)
XnAR XSE
where XAR is the mass fraction of aromatics at the light phase (raffinate), XnAR the mass fraction of
nonaromatics at the light phase (raffinate), XAE the mass fraction at the heavy phase (extract) and XSE is
the mass fraction of sulfolane at the heavy phase (extract).

Fig. 3. Experimental tie lines, equilibrium curve and plait-point calculated by NRTL for system c (sulfolane+toluene+n-hexane)
at 308.15 K.
270 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

The Othmer–Tobias correlation in logarithmic scale also presents a linear dependency, being written as:
   
1 − XnAR 1 − XSE
log = a log +b (2)
XnAR XSE
where XnAR is the mass fraction of nonaromatic at the light phase (raffinate) and XSE is the mass fraction
of sulfolane at the heavy phase (extract).
As shown in Eq. (2), Othmer–Tobias correlation does not include the distributed component, the
aromatic in this case, which could have led to inferior results to those obtained through Hand’s correlation.

4. Data correlation

Binary interaction parameters for NRTL and UNIQUAC activity coefficient models were fitted to the
experimental data. Following the work proposed by Sorensen and Arlt [17], two objective functions

Fig. 4. Experimental tie lines, equilibrium curve and plait-point calculated by UNIQUAC for system a (sulfolane + p-xylene +
cyclo-hexane) at 323.15 K.
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 271

were used to minimize the difference between the experimental and the calculated points. The functions
are:
n 
 3
Func = [aijE − aijR ]2 (3)
i=1 j =1

n
 
3 
2
expt
Func = min [xijL − xijL
calc 2
] (4)
i=1 j =1 L=1

where aij R and aij E represent the activities of the components in the light and heavy phases obtained
expt
through the experimental point, xijL is the experimental molar fraction, xijL
calc
the calculated molar fraction,
i the number of tie lines and j the number of components and L the number of phases.

Fig. 5. Experimental tie lines, equilibrium curve and plait-point calculated by UNIQUAC for system b (sulfolane + p-xylene +
n-hexane) at 323 K.
272 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

These objective functions were optimized with Direction Set (Powell’s) Methods in Multidimentions
published by Press et al. [18] in order to find the interaction parameters (Aij ) for NRTL and UNIQUAC
models.
Tables 5 and 6 show the parameters obtained for each system and the root mean square deviation
(RMSD) in molar fraction obtained from experimental data and calculated results.
Although, some binaries presents are the same for different systems, their parameters were not fixed for
all studied systems as proposed by Chen et al. [9]. In this work, it was preferred to obtain a set of indepen-
dent parameters for each system as there were done by Mondragón-Garduño et al. [7] and Letcher et al. [8].
The RSMD values were calculated through the following equation:
 1/2
 n 3  expt
2
[xijL − xijL
calc
]
RMSD = 100   (5)
i j L
6n

Fig. 6. Experimental tie lines, equilibrium curve and plait-point calculated by UNIQUAC for system c (sulfolane + toluene +
n-hexane) at 323.15 K.
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 273

where n is the total number of tie lines and the other indices are the same as described in Eq. (4). For NRTL,
the nonrandomness parameter (α ij ) was set as 0.2. For UNIQUAC model, the geometric parameters (ri
and qi ) were taken from Sorensen and Arlt [17].
It can be seen in Tables 5 and 6, the two models provide a very good representation of the systems
studied. The NRTL model fits slightly better the liquid–liquid equilibrium data of the systems studied. The
square mean root deviations obtained comparing calculated and experimental two-phase compositions
are <1.61% for NRTL model and <2.1% for UNIQUAC model.
With the adjusted parameters, presented in Tables 5 and 6, the binodal curves of each system were
plotted. Figs. 1–3 show the experimental tie lines, the equilibrium curve obtained by NRTL and the
plait-point calculated by NRTL for the systems a–c at 308.15 K. While Figs. 4–6 show the experimental
tie lines, the equilibrium curve obtained by UNIQUAC and the plait-point calculated by UNIQUAC for
the systems a–c at 323.15 K. These figures show that experimental data are very well correlated by both
models.

Fig. 7. Comparison of experimental data at different temperatures (308.15 and 323.15 K obtained in this work and 298.15 K
published by Chen et al. [9]). Observe only the first and the last tie line for each set are indicated in the figure.
274 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

Experimental data for system c are compared at different temperatures (Fig. 7): 308.15 and 323.15 K
data sets obtained in this work and 298.15 K data set published by Chen et al. [9]. Fig. 7 show that the
two-phase region is enlarged as the temperature is increased. It indicates that at a higher temperature,
sulfolane is a more selective solvent for aromatic compounds. The slopes of the tie lines from this
work differs from the ones published by Chen et al. [9]. The main reason for this is the difference in
the behavior of sulfolane concentrations in the light phase, between the two works. As stated before,
the standard deviation for the gas chromatographic analyzes was <0.005 and analyses with standard
deviation >0.005 were neglected.

5. Conclusions

Liquid–liquid equilibrium data were presented for the ternary systems: sulfolane + p-xylene + cyclo-
hexane, sulfolane+p-xylene+n-hexane and sulfolane+toluene+n-hexane at 308.15 and 323.15 K. Data
were obtained using a jacketed glass cell with 193 cm3 . The cells were immersed in a thermostatically
controlled bath. The initial mixture fills completely the cell and after agitating and decanting samples
from the two phase were taken and analyzed through gaseous chromatography. Hand and Othmer–Tobias
methods were used to correlate the tie lines, correlation factor for both methods were very close to the
unity ascertaining the good data quality.
Binary interaction parameters for UNIQUAC and NRTL activity coefficient models were fitted to exper-
imental data in order to minimize square deviations from phase concentrations and activity as suggested
by Sorensen and Arlt [17]. The optimization method used were the Direction Set (Powell’s) Methods
in Multidimentions presented by Press et al. [18]. Both models presented a very good correspondence
with the experimental tie line data, although the NRTL model fits slightly better the liquid–liquid equi-
librium data of the systems considered. The square mean root deviations obtained comparing calculated
and experimental two-phase compositions are <1.61% for NRTL model and <2.1% for UNIQUAC
model.
Experimental data for system c are compared at different temperatures: 308.15 and 323.15 K data
sets obtained in this work and 298.15 K data set published by Chen et al. [9]. The result shows that
the two-phase region is enlarged as the temperature is increased indicating that at a higher temperature,
sulfolane is a more selective solvent for aromatic compounds.

List of symbols
aij components activities
a, b coefficients for Hand and Othmer–Tobias methods
Aij , Aj i binary interaction parameters for NRTL model.
Func objective function for minimization
n total number of tie lines
qi area parameter for UNIQUAC model
2
r coefficient of correlation
ri volume parameter for UNIQUAC model
RMSD root mean square deviation
x molar fraction
X mass fraction
R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276 275

Greek symbol
α ij nonrandomness parameter in NRTL model

Subscripts
A aromatics
E heavy phase (extract)
i number of the tie lines
j number of the components
L number of the phase
nA nonaromatics
R light phase (raffinate)
S sulfolane

Superscripts
expt experimental
calc calculated

References

[1] G.C. Cassel, N. Dural, A.L. Hines, Liquid–liquid equilibrium of sulfolane–benzene–pentane and sulfolane–toluene–pentane
systems, Ind. Eng. Chem. Res. 28 (1989) 1369–1374.
[2] G.C. Cassel, N. Hasan, A.L. Hines, Correlation of the phase equilibrium data for the heptane–toluene–sulfolane and
heptane–xylene–sulfolane system, J. Chem. Eng. Data. 34 (1989) 434–438.
[3] G.C. Cassel, N. Dural, A.L. Hines, Phase equilibria of the cyclo-hexane–toluene–sulfolane and hexane–toluene–sulfolane
ternary systems, Chem. Eng. Comm. 85 (1989) 233–243.
[4] S. Lee, H. Kim, Liquid–liquid equilibria for the ternary systems sulfolane + octane + benzene, sulfolane + octane +
toluene, sulfolane + octane + p-xylene, J. Chem. Eng. Data 40 (2) (1995) 499–503.
[5] S. Lee, H. Kim, Liquid–liquid equilibria for the ternary systems sulfolane + octane + benzene, sulfolane + octane +
toluene, sulfolane + octane + p-xylene at elevated temperatures, J. Chem. Eng. Data 43 (3) (1998) 358–361.
[6] R.M. De Fré, L.A. Verhoeye, Phase equilibria in systems composed of an aliphatic and an aromatic hydrocarbon and
sulfolane, J. Appl. Chem. Biotech. 26 (1976) 469–487.
[7] M. Mondragón-Garduño, A. Romero-Martı́nez, A. Trejo, Liquid–liquid equilibria for ternary systems. Part I. C6-isomers +
sulfolane + toluene at 298.15 K, Fluid Phase Equilib. 64 (1991) 291–303.
[8] T.M. Letcher, G.G. Redhi, S.E. Radloff, U. Domañska, Liquid–liquid equilibria of the ternary mixtures, J. Chem. Eng. Data
41 (1996) 634–638.
[9] J. Chen, L.P. Duan, J.G. Mi, W.Y. Fei, Z.C. Li, Liquid–liquid equilibria of multicomponent systems including n-hexane,
n-octane, benzene, toluene, xylene and sulfolane at 298.15 K and atmospheric pressure, Fluid Phase Equilib. 173 (2000)
109–119.
[10] J. Chen, L.P. Duan, Liquid–liquid equilibria of ternary and quaternary systems including cyclohexane, 1-heptane, benzene,
toluene and sulfolane at 298.15 K, J. Chem. Eng. Data 45 (2000) 689–692.
[11] J. Chen, W.Y. Fei, Z.C. Li, Liquid–liquid equilibria of quaternary and quinary data including sulfolane at 298.15 K, J.
Chem. Eng. Data 46 (2001) 169–171.
[12] H. Renon, J.M. Prausnitz, Local compositions in the thermodynamics excess functions for liquid mixtures, AIChE J. 14
(1968) 135.
[13] D.S. Abrams, J.M. Prausnitz, Statistical thermodynamics of liquid mixtures: a new expression for the excess Gibbs energy
of partly or completely miscible systems, AIChE J. 21 (1975) 116–128.
[14] J.M. Prauznitz, T.F. Anderson, E.A. Grens, C.A. Eckert, R. Hsieh, J.P. O’Connel, Computer Calculations for Multicomponent
Vapor–Liquid and Liquid–Liquid Equilibria, McGraw-Hill, New York, 1986.
276 R. Rappel et al. / Fluid Phase Equilibria 202 (2002) 263–276

[15] D.B. Hand, Dineric distribution, 1930, J. Phys. Chem. (1961) 34.
[16] D.F. Othmer, P.E. Tobias, Tie-line correlation, Ind. Eng. Chem. 34 (1942) 693–700.
[17] J.M. Sorensen, W. Arlt, Liquid–Liquid Equilibrium, Data Collection, DECHEMA, Frankfurt, Vol. V, part 2, pp. XV and
XVI.
[18] W.H. Press, B.P. Flannery, S.A. Teutolsky, et al. Numerical Recipes. The Art of Scientific Computing (Fortran Version),
Cambrigde University Press, New York, 1989.

You might also like