Fluid Phase Equilibria 315 (2012) 21–28

Contents lists available at SciVerse ScienceDirect

Fluid Phase Equilibria

journal homepage: www.elsevier.com/locate/fluid

Extraction of propylbenzene from its mixtures with heptadecane using

4-methyl-N-butylpyridinium tetrafluoroborate
Mohammad S. Al-Tuwaim, Khaled H.A.E. Alkhaldi, Mohamed S. Fandary, Adel S. Al-Jimaz ∗
Chemical Engineering Department, College of Technological Studies, PAAET, P.O. Box 42325, Shuwaikh 70654, Kuwait

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

Article history: The separation of aromatic and aliphatic compounds from their mixtures is an important goal in chemi-
Received 17 August 2011 cal operations that produce both types of compounds. In petroleum industries, the properties of middle
Received in revised form 20 October 2011 distillate fuels can be improved using dearomatization processes. The ionic liquid (IL), 4-methyl-N-
Accepted 10 November 2011
butylpyridinium tetrafluoroborate [(mebupy)(BF4 )], was evaluated as a solvent for the extraction of
Available online 18 November 2011
propylbenzene from its mixtures with high molar mass n-alkane. Liquid–liquid equilibria (LLE) for a
ternary system heptadecane + propylbenzene + [(mebupy)(BF4 )] was measured over a temperature range
Keywords:
(313–333) K and at different solvent to feed ratios. The reliability of the experimental data was tested
LLE
Ionic solvent
using the Othmer–Tobias correlation. The effect of temperature and solvent to feed ratio upon solubility,
Alkane selectivity, and distribution coefficient were investigated experimentally. In addition, the UNIQUAC and
Aromatic the NRTL models were verified to accurately correlate the experimental data. A proposal of extraction
UNIQUAC process with this IL as a solvent was simulated by conventional software.
NRTL © 2011 Elsevier B.V. All rights reserved.

1. Introduction with better selectivity and capacity. From activity coefficients at

infinite dilution measurements, it was shown that a large number
Dearomatization process of the middle distillate fuels by sol- of ionic liquids have better selectivity and capacity in extraction of
vent extraction is an effective operation which is used to produce aromatics from aromatic/aliphatic mixtures than typical solvents
kerosene with a better smoke point, diesel oil with a higher cetane such as sulfolane and NMP [14].
number or a mineral oil with a better viscosity index. In the process In the past few years, ionic liquids (ILs) are considered as new
of manufacturing lubricating oil base stocks, removal of aromatic promising solvents to substitute organic solvents in industrial pro-
hydrocarbons from vacuum distillates is required to improve sev- cesses. Ionic liquids are liquid salts consisting of large, mostly
eral lubricating properties [1,2]. Due to environmental legislation, organic, cations and a great variety of anions. The positive prop-
the demand of ‘clean’ fuels is increasing and most likely will erties of the ionic solvents are negligible vapour pressure, stability
increase even more towards fuels with almost zero content of at high temperatures and the ability of tailoring. The latter pro-
certain aromatics, e.g., benzene and toluene. In particular, the con- vides the possibility to develop a special solvent for a specific
centration of benzene has to be reduced to ≤0.1 wt.% in carburant application, e.g., separation of aromatic and aliphatic hydrocar-
fuels. bons with less damage to the environment [15–20]. Meindersma
The developed extraction technologies can be radically reduc- et al. [15,16] focused on the selection of ionic liquids for the
ing the cost of diesel dearomatization and desulfurization since extraction of aromatic hydrocarbons from aromatic/aliphatic mix-
they do not demand hydrogen and are carried out at atmospheric tures and they also compared the suitability of sulfolane and four
pressure and comparatively low temperatures [3–5]. A large num- ionic liquids; 1,3-dimethyl-imidazolium methylsulfate, 1-ethyl-3-
ber of solvents, {dimethylsulfoxide (DMSO) [6,7], furfural [8,9], methyl-imidazolium ethylsulfate, 1-butyl-3-methyl-imidazolium
N-methylpyrrolidone (NMP) [10,11], dimethylformamide (DMF) methylsulfate and 4-methyl-N-butylpyridinium tetrafluoroborate
[3,4,12], and sulfolane [13]}, have been used for extraction of aro- as solvents in liquid–liquid extraction at 313.2 and 348.2 K. The
matic and sulfur hydrocarbons from middle distillate fractions. best solubility of toluene was observed in [(mebupy)(BF4 )] in
Although the processes based on these solvents are well known, comparison with the three imidazolium salts: [mmim][CH3 SO4 ],
there is still a need to find better environmental friendly solvents [emim][EtSO4 ], and [bmim][CH3 SO4 ] and they also stated that
from these four ILs, [(mebupy)(BF4 )] was the best solvent for
the aromatic/aliphatic separation and all ILs were better than
∗ Corresponding author. sulfolane [15,16]. Domańska et al. [17] studied the separation
E-mail address: a jimaz@yahoo.com (A.S. Al-Jimaz). of aromatic hydrocarbons from alkanes using ammonium ionic

0378-3812/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.fluid.2011.11.008
22 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28

Table 1
Details of the chemicals; purities, UNIQUAC structural parameters, and refractive indices (nD ).

Compound Supplier Purity (mole%) UNIQUAC structural parameter nD 20

r q exp Lit

[(mebupy)(BF4 )] Merck >98 8.4337 6.5290 1.4509a 1.4517a,b

Propylbenzene Fluka >99 5.4983 4.3560 1.4915 1.4920c
Heptadecane Sigma >99 11.9182 9.7959 1.4362 1.4369c
a
nD at 25 ◦ C.
b
Ref. [49].
c
Ref. [50].

liquid [C2 NTf2 ] at T = 298.15 K. García et al. [18] reported the sep- The 4-methyl-N-butylpyridinium tetrafluoroborate
aration of toluene and heptane by liquid–liquid extraction using [(mebupy)(BF4 )] is one of the few ILs that shows both higher
z-methyl-N-butylpyridinium tetrafluoroborate isomers (z = 2, 3, or selectivity and extractive capacity than those of sulfolane for the
4) at T = 313.2 K. Arce et al. [19,20] investigated the separation extraction of aromatic hydrocarbons [15,16]. Nevertheless, LLE
of aromatic hydrocarbons from alkanes using the ionic liquid 1- data are still quite scarce for systems containing ionic liquids. Cur-
ethyl-3-methylimidazolium bis (trifluoromethyl) sulfonyl amide rently, there are only a few published experimental liquid–liquid
and the effect of the alkyl-substituent length upon the separa- equilibrium data for [(mebupy)(BF4 )] + aromatic + n-aliphatic
tion of benzene and hexane using 1-alkyl-3-methylimidazolium (C5 –C9 ) and hardly any for systems containing carbon number
bis{(trifluoromethyl)sulfonyl}amide. greater than nine for the aliphatic and/or aromatics [15–17,32–37].
The liquid–liquid equilibria experimental measurements for Abu-Eishah and Dowaidar [34] studied the liquid–liquid equi-
the ternary systems (aliphatic hydrocarbons + aromatic hydro- librium of ternary systems of cyclohexane + (benzene, +toluene,
carbons + ILs) is necessary to evaluate the feasibility of using +ethylbenzene, or +o-xylene) + 4-methyl-N-butyl pyridinium
ILs as extractive solvents in the separation of aromatic and tetrafluoroborate at 303.15 K. Meindersma et al. [35] reported
aliphatic hydrocarbons [21–33]. Arce et al. [21] investigated 1- the study of the liquid–liquid equilibrium for 4-methyl-N-
ethyl-3-methylimidazolium bis{(trifluoromethyl)sulfonyl}amide butylpyridinium tetrafluoroborate + (benzene or ethylbenzene
as solvent for the separation of aromatic and aliphatic or m-xylene) + (N-hexane or N-octane) mixtures at T (313.2 and
hydrocarbons by liquid extraction-extension to C7- and C8- 348.2) K and T (313.2 and 333.2) K. Alkhaldi et al. [36] investigated
fractions. Wang et al. [22] studied the liquid–liquid equilibria the separation of propylbenzene and tetradecane or hexadecane
for benzene + cyclohexane + 1-methyl-3-methylimidazolium from their mixtures using 4-methyl-N-butylpyridinium tetrafluo-
dimethylphosphate or +1-ethyl-3-methylimidazolium roborate at several temperatures. Al-Tuwaim et al. [37] reported
diethylphosphate and ternary liquid–liquid equilibria mea- the extraction of propylbenzene or butylbenzene from dodecane
surement for benzene + cyclohexane + N-methylimidazole, or using 4-methyl-N-butylpyridinium tetrafluoroborate at different
N-ethylimidazole, or N-methylimidazolium dibutylphosphate at temperatures.
298.2 K and atmospheric pressure [23]. Maduro and Aznar [24] This paper is a continuation of our study on the liquid–liquid
reported the liquid–liquid equilibrium of ternary systems 1-butyl- phase equilibria for dearomatization of Kuwait middle distilled
3-methylimidazolium hexafluorophosphate + aromatic + aliphatic. fraction [36–39]. We are interested in providing experimen-
González et al. [25] studied the liquid–liquid equilibria for ternary tal LLE data containing IL. The purpose of this work is to
mixtures of (alkane + benzene + [EMpy] [ESO4 ]) at several tem- study LLE of a ternary system; {heptadecane (1) + propylbenzene
peratures and atmospheric pressure. Pereiro and Rodriguez (2) + [(mebupy)(BF4 )] (3)}. The LLE data for the studied ternary sys-
[26] investigated the application of the ionic liquid Ammoeng tem was measured at 313 K, 323 K, and 333 K and atmospheric
102 for aromatic/aliphatic hydrocarbon separation. Arce et al. pressure. The reliability of the experimentally measured tie-line
[27] studied bis{(trifluoromethyl)sulfonyl}amide ionic liquids data was ascertained by the Othmer–Tobias correlation [40], and
as solvents for the extraction of aromatic hydrocarbons from the extracting distribution coefficient as well as the selectivity were
their mixtures with alkanes: effect of the nature of the cation. calculated. The experimental data were correlated by the UNI-
García et al. [28] investigated the liquid–liquid equilibria for QUAC and the NRTL models [41,42]. Finally proposed single stage
{hexane + benzene + 1-ethyl-3-methylimidazoliumethylsulfate} extraction process was simulated by HYSYS software to investi-
at (298.2, 313.2 and 328.2) K and the ternary liquid–liquid gate economic feasibility at different mole fraction of aromatic in
equilibria measurement for hexane and benzene with 1-butyl-3- the feed stream [43–45].
methylimidazolium methylsulfate at T = (298.2, 313.2, and 328.2) K
[29]. Han et al. [30] studied the liquid–liquid equilibria of ionic liq-
uid 1-butyl-3-methylimidazolium tetrafluoroborate and sodium
citrate/tartrate/acetate aqueous two-phase systems at 298.15 K. 2. Experimental
Gonzále, et al. [31] reported the effect of the size of the aliphatic
hydrocarbons upon the separation of benzene from alkanes using 2.1. Chemicals
1-ethyl-3-methylpyridinium ethylsulfate ionic liquid at several
temperatures and atmospheric pressure. Heintz et al. [32] studied The [(mebupy)(BF4 )] and propylbenzene were stored under
the activity coefficients at infinite dilution of alkanes, alkenes, 4 nm molecular sieve. The water content was determined using
and alkylbenzenes in 4-methyl-N-butylpyridinium tetrafluo- Karl Fischer titration method showing the mass fraction of water
roborate using gas-liquid chromatography. García et al. [33] was less than 6 × 10−4 , the water in IL was regularly mea-
investigated the isomeric effect of the cation of N-butylpyridinium sured using Karl-Fischer method, no increase of water content
bis-(trifluoromethylsulfonyl)imide for the liquid–liquid extraction has been observed. The purities of heptadecane and propylben-
of aromatics from their mixtures with alkanes. Most of the avail- zene were determined by gas chromatography. All chemicals
able literature is focused on the application of imidazolium-based were used without further purification. The purities and refrac-
ionic liquids for organic reactions, while very limited information tive indices of all chemicals used in this study are presented
is available on the application of pyridinium-based ionic liquids. in Table 1.
 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28 23

Table 2
Experimental data for the ternary system {heptadecane (1) + propylbenzene (2) + [(mebupy)(BF4 )] (3)} at T = (313–333) K and P = 101.3 kPa.

T/K Heptadecane rich phase Solvent rich phase K S

x1 x2 x1 x2

313 0.8535 0.1465 0.0016 0.0334 0.23 119.30

0.7424 0.2576 0.0017 0.0614 0.24 102.34
0.6552 0.3448 0.0018 0.0847 0.25 89.41
0.5855 0.4145 0.0018 0.1049 0.25 80.55
0.5318 0.4682 0.0020 0.1313 0.28 74.58
0.4854 0.5146 0.0021 0.1511 0.29 67.85
0.4466 0.5534 0.0022 0.1704 0.31 63.98
0.4123 0.5877 0.0022 0.1842 0.31 58.74
0.3568 0.6432 0.0023 0.2076 0.32 50.06
0.3167 0.6833 0.0024 0.2426 0.36 46.86

323 0.8566 0.1434 0.0020 0.0373 0.26 114.32

0.7459 0.2541 0.0020 0.0670 0.26 98.30
0.6593 0.3407 0.0021 0.0926 0.27 85.33
0.5894 0.4106 0.0022 0.1140 0.28 74.36
0.5350 0.4650 0.0023 0.1398 0.30 69.94
0.4885 0.5115 0.0025 0.1605 0.31 62.07
0.4499 0.5501 0.0025 0.1818 0.33 58.76
0.4155 0.5845 0.0026 0.1967 0.34 53.78
0.3598 0.6402 0.0027 0.2223 0.35 46.27
0.3189 0.6811 0.0028 0.2554 0.37 42.70

333 0.8591 0.1409 0.0023 0.0405 0.29 107.26

0.7500 0.2500 0.0024 0.0733 0.29 91.67
0.6639 0.3361 0.0025 0.1012 0.30 79.97
0.5947 0.4053 0.0026 0.1257 0.31 70.96
0.5385 0.4615 0.0027 0.1488 0.32 64.33
0.4921 0.5079 0.0028 0.1711 0.34 59.20
0.4536 0.5464 0.0029 0.1939 0.35 55.50
0.4190 0.5810 0.0030 0.2097 0.36 50.40
0.3632 0.6368 0.0031 0.2380 0.37 43.79

Composition in mole fractions with an uncertainty of ±0.0005.

Temperature controlled with an uncertainty of ±0.2 K.

2.2. Apparatus and procedure 5 min. The heating rate was 35 K/min, and the carrier gas (Helium)
flow rate was maintained at 3 × 10−5 m3 /min. The injection tem-
The experimental apparatus used for extraction consists of a perature was 523 K and the detector temperature was 573 K. The
60 mL glass cells with a water jacket in order to maintain a con- temperature was controlled with a precision of ±0.03 K. Each mole
stant temperature. The temperature controlled with an uncertainty fraction was measured repeatedly for four times to reduce experi-
of ±0.2 K. The cells connected to a Haake K15 water bath fitted mental uncertainty associated with random errors and the average
with a Haake DC1 thermostat. Mixtures, comprising of 20 g of value was recorded. The compositions in mole fractions were mea-
[(mebupy)(BF4 )], 20 g of heptadecane, and different amounts of sured with an experimental uncertainty of ±0.0005.
propylbenzene were placed in the extraction vessels. The mixtures
were vigorously stirred for 1 h, and then left to settle for 4 h. A series 3. Results and discussion
of LLE measurements over a temperature range of 313–333 K were
performed. 3.1. Experimental data

2.3. Measurements of phase compositions The measured equilibrium mole fractions for the ternary system
are reported in Table 2. As shown in this table, the temperature has
Samples were carefully taken by a syringe from the lower and no effect on the solubility of [(mebupy)(BF4 )] in heptadecane rich
upper layers. Each sample was dissolved in 0.5 mL 1-butanol, to phase, while it has a little effect upon the solubility of heptadecane
avoid splitting and maintain a homogeneous mixture, and ana- in ionic solvent rich phase. On the other hand, the concentration of
lyzed using a Varian 450 gas chromatography equipped with an propylbenzene has no effect on the solubility of [(mebupy)(BF4 )] in
autosampler (Varian CP-8400), an on-column injecter, flame ion- heptadecane rich phase, while it has a little effect upon the solu-
ization detector (FID), and a data processing system. Varian VF-5 ms bility of heptadecane in ionic solvent rich phase. The experimental
CP8944 (30 m length and 0.25 mm I.D., 0.25 ␮m film thickness) col- and the predicted tie lines for the ternary system at 313 K, 323 K and
umn was used. 333 K are shown in Figs. 1–3 respectively. No detectable concentra-
The ionic solvent, [(mebupy)(BF4 )], has negligible vapor pres- tions of [(mebupy)(BF4 )] were found in the heptadecane rich phase.
sure it cannot be analyzed by GC. In the ternary mixture, only two The complete absence of [(mebupy)(BF4 )] in the heptadecane rich
components need to be analyzed; the third one, the ionic liquid, phase is desirable, since it permits to avoid the need of a unit for
determined by a mass balance of the measured mass fractions of recovering the solvent from the raffinate in a continuous extraction
heptadecane and propylbenzene in the two phases. In order to process.
avoid inaccuracy of the analysis caused by fouling of the GC by The reliability of the experimentally measured tie-line data
the ionic liquid a pre-column was used to protect the column and ascertained by applying the Othmer–Tobias correlation [40]:
collect the ionic solvent in order not to disrupt the analysis.    
1 − w3II 1 − w1I
The GC column temperature was programmed for an initial tem- ln = a + b ln (1)
perature of 363 K for 2 min, and a final temperature of 673 K for w3II w1I
24 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28

Fig. 1. Experimental and predicted LLE data for the ternary system: heptadecane (1) + propylbenzene (2) + [(mebupy)(BF4 )] (3) at 313 K: 䊉, experimental; dashed line, NRTL;
solid line, UNIQUAC.

Fig. 2. Experimental and predicted LLE data for the ternary system: heptadecane (1) + propylbenzene (2) + [(mebupy)(BF4 )] (3) at 323 K: 䊉, experimental; dashed line, NRTL;
solid line, UNIQUAC.

Fig. 3. Experimental and predicted LLE data for the ternary system: heptadecane (1) + propylbenzene (2) + [(mebupy)(BF4 )] (3) at 333 K: 䊉, experimental; dashed line, NRTL;
solid line, UNIQUAC.
 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28 25

Table 3
Constants of the Othmer–Tobias correlation, correlation factor (R2 ) and standard
deviations () at T = (313–333) K and P = 101.3 kPa.

T/K A b R2 

313 −1.7853 0.8393 0.9996 0.0393

323 −1.7670 0.8361 0.9992 0.0416
333 −1.6771 0.8238 0.9995 0.0447

Fig. 5. Measured selectivity (S) against solvent to feed ratio (˛stf ) at: 䊉, 313; , 323;
, 333 K.

where the subscript 1 represents heptadecane and 2 represents

propylbenzene.
From the selectivity data represented in Fig. 5, for the ternary
system {heptradecane (1) + propylbenzene (2) + [(mebupy)(BF4 )]
(3)} at temperature range of (313–333) K, the higher the tem-
Fig. 4. Measured distribution coefficient (K) against solvent to feed ratio (˛stf ) at: 䊉, perature, the lower the selectivity. This trend is the same that
313; , 323; , 333 K. followed by the size of the immiscibility region, i.e., the higher
in the immiscibility zone, the higher the selectivity values. As
shown in Table 2, the selectivity values are not constant over
where w3II is the mass fraction of [(mebupy)(BF4 )] (3) in the lower
whole two-phase region, and it is decreased as the concentration
layer (IL-rich phase), w1I is the mass fraction of heptadecane (1) in
of propylbenzene increased, it means the higher the concentra-
the upper layer (heptadecane-rich phase), a and b are the fitting
tion of propylbenzene in the feed the lower the selectivity of
parameters of the Othmer–Tobias correlation. The linearity of the
[(mebupy)(BF4 )]. The selectivity values in the systems under study
plot indicates the degree of consistency of the data.
are higher than unity, which ensures the feasibility of separation
The parameters of the Othmer–Tobias correlation as well as the
of propylbenzene and heptadecane from their mixtures. Moreover,
regression coefficients (R2 ) and standard deviation () are given
the use of [(mebupy)(BF4 )], and of ionic liquids in general, in sol-
in Table 3. The regression coefficients are very close to unity and
vent extraction is favorable because it can be easily recovered and
the low values of the standard deviations indicate the degree of
reused.
consistency of the related data at 313 K, 323 K and 333 K.
The operation of liquid–liquid extraction was affected by varia-
tions in the operating temperature and solvent to feed ratio. Thus,
3.2. Distribution coefficient and selectivity the determination of the effects of temperature and solvent to feed
ratio on the values of distribution coefficient and selectivity were
Together with the LLE data, Table 2 includes the corresponding quite crucial for optimum operation of the liquid–liquid extraction
values for the solute distribution ratio (K) and the selectivity (S), process. A final preview of the extraction efficiency parameters (S
which are widely used parameters in assessing the feasibility of and K) against the solvent to feed ratio for the system under study
utilizing the solvent in liquid–liquid extraction.
The distribution coefficient of propylbenzene, which is the mea-
sure of the solvent power or capacity of [(mebupy)(BF4 )], is given
by:

x2II
K= (2)
x2I

Fig. 4 represents the relationship of the solvent to feed ratio ˛stf ,

with the measured distribution coefficients, for the ternary sys-
tem; {heptadecane (1) + propylbenzene (2) + [(mebupy)(BF4 )] (3)}
at temperatures of (313–333) K. The distribution coefficient values
increased as the temperature increased and/or solvent to feed ratio
˛stf decreased.
The effectiveness of a solvent can be expressed by the selectivity
S, of the solvent. The selectivity of [(mebupy)(BF4 )], which is a mea-
sure of the ability of [(mebupy)(BF4 )] to separate propylbenzene
from heptadecane, is given by:

x2II x1I
S= (3) Fig. 6. Distribution coefficient (K) against selectivity (S) at different temperature:
x2I x1II , S; , K.
26 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28

Table 4
UNIQUAC interaction parameters and root mean square deviation (rmsd) at
T = (313–333) K and P = 101.3 kPa.

T/K I j UNIQUAC

aij aji rmsd

313 Heptadecane Propylbenzene −37.72 38.27

Heptadecane (mebupy)(BF4 ) 345.43 256.89
Propylbenzene (mebupy)(BF4 ) 119.58 4.54 0.1977

323 Heptadecane Propylbenzene 12.69 −1.36

Heptadecane (mebupy)(BF4 ) 382.54 270.47
Propylbenzene (mebupy)(BF4 ) 93.53 29.81 0.1881

333 Heptadecane Propylbenzene 173.43 −117.58

Heptadecane (mebupy)(BF4 ) 457.71 333.87
Propylbenzene (mebupy)(BF4 ) 23.71 97.55 0.1839

representation of the tie line data for those systems. However, the
Fig. 7. Conceptual simulation process flowsheet for the separation of propylbenzene UNIQUAC model, fitted to LLE experimental data, is more accurate
and heptadecane from their mixtures using [(mebupy)(BF4 )] as ionic solvent.
than the NRTL model, according to the analysis of the average rmsd
(the average rmsd was 0.1810 for UNIQUAC as compared to 0.2281
is shown in Fig. 6 at temperatures 313 K, 323 K and 333 K. Based for NRTL).
on the values of distribution coefficient (K) and selectivity (S), the
optimum operation for the systems under studied was at 0.4 < ˛stf ,
3.4. Simulation results
<0.6.

A conceptual single stage extraction process was simulated

3.3. Correlations
using HYSYS v.3.2 (from Aspen Technology, Cambridge, MA). The
equilibrium data was generated using the UNIQUAC equation fitted
The UNIQUAC model of Abrams and Prausnitz [41] and the NRTL
to the experimental tie-line data. The simulation investigated the
model of Renon and Prausnitz [42] were used to correlate our
economic feasibility of aromatics removal from paraffin versus the
experimental data.
mole fraction of aromatics in the feed stream.
The LLE experimental data were used to determine the opti-
The flow sheet of the process is shown in Fig. 7. Propylbenzene
mum UNIQUAC and NRTL binary interaction parameters between
and heptadecane mixture was fed from the bottom and the raffinate
[(mebupy)(BF4 )], propylbenzene, and heptadecane. The UNIQUAC
stream was collected from the top of the liquid–liquid extractor.
and the NRTL models were fitted to experimental data using an
The pure solvent stream was fed countercurrent to the top of the
iterative computer program, based on flash calculation method,
extractor from the evaporator as regenerated ionic solvent was
developed by S∅rensen [46]. The objective function used in this
used.
case was determined by minimizing the square of the difference
The solvent to feed ratio was kept constant at 0.5 while mole
between the mole fractions predicted by the respective method and
fraction of propylbenzene in the feed stream was varied from 0.15
these experimentally measured over all the tie lines in the ternary
to 0.55. The temperature was set at 313 K for highest selectivity. The
system. For the UNIQUAC correlation the pure component struc-
key cost figure is the heat supplied at the evaporator to regenerate
tural parameters (r and q) listed in Table 1, were calculated from the
the ionic solvent. This is a prerequisite for an economically feasible
group contribution data [47] or taken from literature [48]. The NRTL
process due to high prices of ionic solvents as otherwise, alternative
model was fitted with fixed values of the third non-randomness
processes like azeotropic distillation may be favorable.
parameter, ˛ij , for each pair of components (˛ij = 0.2 or 0.3) during
calculations, and the correlation with ˛ij = 0.2 gave the best results
according to rmsd values.
The objective function OF, used is:
 2
OF = min (xijk,exp − xijk,cal ) (4)
k j i

where x is mole fraction, subscripts exp, cal, i, j, and k are experi-

mental, calculated, components, phases and tie lines respectively.
The optimization results were judged by calculating the corre-
sponding relative mean square deviation (rmsd) values using the
following equation:
   1/2
k j i
(xijk,exp − xijk,cal )2
rmsd = 100 (5)
6n

where n is number of tie lines.

The values of interaction parameters and the relative mean
square deviation for the UNIQUAC and the NRTL models at dif-
ferent temperatures are shown in Tables 4 and 5, respectively.
Fig. 8. Heat duty of solvent regenerator versus percentage removal of propylben-
These parameters are used to calculate LLE tie lines for the zene from feed stream at different aromatic mole fractions in the feed stream; ,
present systems. The calculation based on both models gave good 0.15; , 0.2; , 0.25; , 0.3; , 0.35; , 0.4; ♦, 0.45; , 0.5; , 0.55.
 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28 27

Table 5
NRTL interaction parameters and root mean square deviation (rmsd) at T = (313–333) K and P = 101.3 kPa.

T/K i j NRTL (˛ = 0.2)

aij aji rmsd

313 Heptadecane Propylbenzene −609.74 483.75

Heptadecane (mebupy)(BF4 ) 1553.7 1263.1
Propylbenzene (mebupy)(BF4 ) 1533.1 −277.28 0.2324

323 Heptadecane Propylbenzene −607.83 506.68

Heptadecane (mebupy)(BF4 ) 1610.2 1282.1
Propylbenzene (mebupy)(BF4 ) 1576.6 −302.94 0.2221

333 Heptadecane Propylbenzene −620.24 537.01

Heptadecane (mebupy)(BF4 ) 1682 1301.2
Propylbenzene (mebupy)(BF4 ) 1594.1 −337.19 0.2464

T/K i j NRTL (˛ = 0.3)

aij aji rmsd

313 Heptadecane Propylbenzene −565.33 687.00

Heptadecane (mebupy)(BF4 ) 1533.0 1540.4
Propylbenzene (mebupy)(BF4 ) 1530.1 0.66557 0.2507

323 Heptadecane Propylbenzene −558.75 699.14

Heptadecane (mebupy)(BF4 ) 1578.3 1575.0
Propylbenzene (mebupy)(BF4 ) 1571.9 −16.348 0.2474

333 Heptadecane Propylbenzene −570.68 −16.34

Heptadecane (mebupy)(BF4 ) 1625.5 1612.8
Propylbenzene (mebupy)(BF4 ) 1607.3 −55.183 0.2792

Table 6
Simulation data for the extraction of propylbenzene from heptadecane using ionic solvent at 313 K and 101.3 kPa.

Stream Stream

Feed Solvent Raffinate Out Extract

Flow rate (mol/h) 50.00 25.00 49.58 0.42 25.42

×Propylbenzene 0.1500 0.0000 0.1419 0.9072 0.0151
×Heptadecane 0.8500 0.0000 0.8581 0.0928 0.0015
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.9833

Flow rate (mol/h) 50.00 25.00 49.14 0.86 25.86

×Propylbenzene 0.2000 0.0000 0.1852 0.9516 0.0316
×Heptadecane 0.8000 0.0000 0.8148 0.0484 0.0016
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.9668

Flow rate (mol/h) 50.00 25.00 48.68 1.32 26.32

×Propylbenzene 0.2500 0.0000 0.2288 0.9665 0.0483
×Heptadecane 0.7500 0.0000 0.7712 0.0335 0.0017
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.9500

Flow rate (mol/h) 50.00 25.00 48.21 1.79 26.79

×Propylbenzene 0.3000 0.0000 0.2730 0.9740 0.0652
×Heptadecane 0.7000 0.0000 0.7270 0.0260 0.0017
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.9331

Flow rate (mol/h) 50.00 25.00 47.70 2.30 27.30

×Propylbenzene 0.3500 0.0000 0.3177 0.9785 0.0823
×Heptadecane 0.6500 0.0000 0.6823 0.0215 0.0018
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.9159

Flow rate (mol/h) 50.00 25.00 47.18 2.82 27.82

×Propylbenzene 0.4000 0.0000 0.3630 0.9816 0.0996
×Heptadecane 0.6000 0.0000 0.6370 0.0184 0.0019
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.8985

Flow rate (mol/h) 50.00 25.00 46.62 3.38 28.38

×Propylbenzene 0.4500 0.0000 0.4090 0.9837 0.1171
×Heptadecane 0.5500 0.0000 0.5910 0.0163 0.0019
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.8809

Flow rate (mol/h) 50.00 25.00 46.03 3.97 28.97

×Propylbenzene 0.5000 0.0000 0.4556 0.9853 0.1350
×Heptadecane 0.5000 0.0000 0.5444 0.0147 0.0020
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.8630

Flow rate (mol/h) 50.00 25.00 45.41 4.59 29.59

×Propylbenzene 0.5500 0.0000 0.5031 0.9866 0.1531
×Heptadecane 0.4500 0.0000 0.4969 0.0134 0.0021
×(mebupy)(BF4 ) 0.0000 1.0000 0.0000 0.0000 0.8448
28 M.S. Al-Tuwaim et al. / Fluid Phase Equilibria 315 (2012) 21–28

Table 6 lists streams flow rates and compositions. Fig. 8 shows [9] H.G. James, E.H. Glenn, Petroleum Refining Technology and Economics, Marcel
that at low aromatic mole fraction, percentage removal of aro- Dekker, Inc., New York, Basel, 2001.
[10] R. Krishna, H. Singh, K. Kishore, G.S. Choudhary, R.S. Kaushik, Revue-de-L’ Insti-
matic can be enhanced without much increase in the operating tut Français du Petrole 42 (6) (1987) 827–836.
cost (energy), while at higher aromatic mole fractions an increase [11] V.R. Nigmatullin, R.R. Mukhametova, I.R. Nigmatullin, Chem. Technol. Fuels Oils
of operating cost is noticeable and requires careful assessment of 44 (2008) 10–12.
[12] A.A. Gaile, G.D. Zalishchevskii, L.V. Varshavskii, L.V. Semenov, Russ. J. App.
economic feasibility. Chem. 74 (5) (2001) 864–869.
[13] J. Mahmoudi, M.N. Lotfollahi, J. Chem. Thermodyn. 42 (2010) 466–471.
4. Conclusions [14] A. Marciniak, Fluid Phase Equilib. 294 (2010) 213–233.
[15] G.W. Meindersma, A.J.G. Podt, A.B. de Haan, Fuel Process. Technol. 87 (2005)
59–70.
An experimental investigation of equilibrium behavior [16] G.W. Meindersma, A.J.G. Podt, A.B. de Haan, Fluid Phase Equilib. 247 (2006)
of liquid–liquid equilibria of the ternary system heptade- 158–168.
cane + propylbenzene + [(mebupy)(BF4 )] was carried out at [17] U. Domańska, A. Pobudkowska, M. Królikowski, Fluid Phase Equilib. 259 (2007)
173–179.
temperatures of 313–333 K and at atmospheric pressure. While the [18] J. García, S. García, J.S. Torrecilla, M. Oliet, F. Rodríguez, J. Chem. Thermodyn. 42
temperature has no effect on the solubility of [(mebupy)(BF4 )] in (2010) 1004–1008.
heptadecane rich phase, it has a little effect upon the solubility of [19] A. Arce, M.J. Early, H. Rodríguez, K.R. Seddon, Green Chem. 9 (2007)
70–74.
heptadecane in ionic solvent rich phase. The solubility of heptade- [20] A. Arce, M.J. Early, H. Rodríguez, K.R. Seddon, J. Phys. Chem. B 111 (2007)
cane in ionic solvent rich phase increased as the concentration of 4732–4736.
propylbenzene increases, but it has no effect upon the solubility of [21] A. Arce, M.J. Early, H. Rodríguez, K.R. Seddon, A. Soto, Green Chem. 10 (2008)
1294–1300.
[(mebupy)(BF4 )] in heptadecane rich phase. A good degree of con-
[22] R. Wang, J. Wang, H. Meng, C. Li, Z. Wang, J. Chem. Eng. Data 53 (2008)
sistency of the experimental LLE data was ascertained by applying 1159–1162.
the Othmer–Tobias correlation with an average regression coeffi- [23] R. Wang, J. Wang, H. Meng, C. Li, Z. Wang, J. Chem. Eng. Data 53 (2008)
cient (R2 ) equal to 0.9994. In addition, the UNIQUAC and the NRTL 2170–2174.
[24] R.M. Maduro, M. Aznar, Fluid Phase Equilib. 265 (2008) 129–138.
models satisfactorily correlate the LLE experimental data, however [25] E.J. González, N. Calvar, B. González, A. Domínguez, J. Chem. Thermodyn. 41
the former model was more suitable for the studied systems (the (2009) 1215–1221.
average rmsd was 0.1810 for UNIQUAC as compared to 0.2281 [26] A.B. Pereiro, A. Rodriguez, J. Chem. Thermodyn. 41 (2009) 951–956.
[27] A. Arce, M.J. Early, H. Rodríguez, K.R. Seddon, A. Soto, Green Chem. 11 (2009)
for NRTL). The effects of temperature and the solvent to feed 365–372.
ratio on distribution coefficient and selectivity were evaluated. As [28] J. García, A. Fernández, J.S. Torrecilla, M. Oliet, F. Rodríguez, Fluid Phase Equilib.
the operating temperature increased the distribution coefficient 282 (2009) 117–120.
[29] J. García, A. Fernandez, J.S. Torrecilla, M. Oliet, F. Rodríguez, J. Chem. Eng. Data
increased and the selectivity decreased. Since the selectivity in all 55 (2010) 258–261.
cases is greater than unity, [(mebupy)(BF4 )] can be used to extract [30] J. Han, C. Yu, Y. Wang, X. Xie, Y. Yan, G. Yin, W. Guan, Fluid Phase Equilib. 295
propylbenzene from its mixtures with heptadecane. Optimized (2010) 98–103.
[31] E.J. Gonzále, N. Calvar, E. Gómez, Á. Domínguez, J. Chem. Thermodyn. 42 (2010)
extraction process for the systems under study at temperatures of 104–109.
(313–333) K was attained at ˛stf , ratio of (0.4–0.6). The proposed [32] A. Heintz, D.V. Kulikov, S.P. Verevkin, J. Chem. Eng. Data 46 (2001)
liquid–liquid extraction process at 313 K is economically favorable 1526–1529.
[33] J. García, S. García, J.S. Torrecilla, M. Oliet, F. Rodríguez, Fluid Phase Equilib. 301
to conventional extraction process due to the simpler solvent
(2011) 62–66.
regeneration. The fact that paraffin mole fraction in the extract [34] S.I. Abu-Eishah, A.M. Dowaidar, J. Chem. Eng. Data 53 (2008) 1708–1712.
phase is very low reduces the energy requirement for solvent [35] G.W. Meindersma, A.J.G. Podt, A.B. de Haan, J. Chem. Eng. Data 47 (2006)
regeneration of the ionic solvent. 1814–1819.
[36] K.H.A.E. Alkhaldi, M.S. Al-Tuwaim, M.S. Fandary, A.S. Al-Jimaz, Fluid Phase Equi-
lib. 309 (2011) 102–107.
Acknowledgements [37] M.S. Al-Tuwaim, K.H.A.E. Alkhaldi, M.S. Fandary, A.S. Al-Jimaz, J. Chem. Ther-
modyn. 43 (2011) 1804–1809.
[38] A.S. Al-Jimaz, M.S. Fandary, J.A. Al-Kandary, M.A. Fahim, Fluid Phase Equilib.
The authors thank the Public Authority for Applied Education 231 (2) (2005) 163–170.
and Training (PAAET-TS-08-012) for the financial support of this [39] A.S. Al-Jimaz, M.S. Fandary, K.H.A.E. Alkhaldi, J.A. Al-Kandary, M.A. Fahim, Ind.
work. Eng. Chem. Res. 46 (2007) 5686–5696.
[40] D.F. Othmer, P.E. Tobias, Liquid–liquid extraction data-the line correlation, Ind.
Eng. Chem. 34 (1942) 693–696.
References [41] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116–128.
[42] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135–144.
[1] H. Singh, K. Kishore, J. Appl. Chem. Biotechnol. 28 (1978) 617–625. [43] A.B. Pereiro, F.J. Deive, J.M.S.S. Esperanca, A. Rodríguez, Fluid Phase Equilib. 291
[2] A. Sequeira, Lubricant Base Oil and Wax Processing, Marcel Dekker, New York, (2010) 13–17.
1994, pp. 81–82. [44] A.B. Pereiro, A. Rodríguez, AIChE J. 56 (2) (2010) 381–386.
[3] V. Toteva, L. Topalova, P. Manolova, J. Univ. Chem. Technol. Metal. 42 (2007) [45] A.B. Pereiro, A. Rodríguez, Chem. Eng. J. 153 (2009) 80–85.
17–20. [46] J.M. S∅rensen, W. Arlt, Vol. V, Part 2, Dechema Chemistry Data Series, 625
[4] B.P. Baruah, K. Puja, Energy Fuels 21 (2007) 2156–2164. Seiten, Preis DM 176, DECHEMA, Frankfort/Main, Germany, 1980.
[5] J.M. Bolard, in: A.G. Lucas (Ed.), Modern Petroleum Technology, 6th edition, [47] T. Magnussen, P. Rasmussen, A. Fredenslund, Ind. Eng. Chem. Process. Des. Dev.
John W. & Sons L, Chichester, UK, 2000, pp. 367–387. 20 (1981) 331–339.
[6] Http://en.wikipedia.org/wiki/Dimethyl sulfoxide. [48] R.S. Santiago, G.R. Santos, M. Aznar, Fluid Phase Equilib. 278 (2009) 54–61.
[7] A.K. Sharipov, Chem. Technol. Fuels Oils 37 (2001) 62–72. [49] J. Ortega, R. Vreekamp, E. Penco, E. Marrero, J. Chem. Thermodyn. 40 (2008)
[8] R.R. Savory, in: A.G. Lucas (Ed.), Modern Petroleum Technology, 6th edition, 1087–1094.
John W. & Sons L, Chichester, UK, 2000, pp. 127–138. [50] CRC Handbook of Chemistry and Physics, 86th edition, 2005–2006.