Toward Modeling the Aromatic/Aliphatic Separation by ExtractiveDistillation with Tricyanomethanide-Based Ionic Liquids Using CPAEoSMiguel Ayuso,† Pablo Navarro,*,‡,§ Andre M. Palma,‡ Marcos Larriba,†,‡ Noemí Delgado-Mellado,†

Julian García,† Francisco Rodríguez,† Joao A.P. Coutinho,‡ and Pedro J. Carvalho‡

†Department of Chemical Engineering, Complutense University of Madrid, Madrid, Spain‡CICECO − Aveiro Institute of Materials, Department of Chemistry, University of Aveiro, Aveiro, Portugal§Department of Chemical Engineering, Autonoma University of Madrid, Madrid, Spain

*S Supporting Information

ABSTRACT: Extractive distillation using tricyanomethanide-based ionic liquids (ILs) has been shown to be a promising andfeasible process for effectively separate aromatics from pyrolysis gasolines. The high performance of these mass agents has beenreported by evaluating simple synthetic n-heptane/toluene binary mixtures, on a wide temperature and solvent to feed (S/F)ratio ranges. However, industrial streams are much more complex with the presence of other aromatic and aliphatic compounds,like benzene, xylenes, and shorter and longer linear alkanes, creating further difficulties to the separation and thus must bestudied. This work covers the phase equilibrium characterization of {n-hexane + benzene + IL} and {n-octane + p-xylene + IL}ternary systems with two tricyanomethanide-based ILs, namely 1-ethyl-3-methylimidazolium tricyanomethanide ([C2C1im]-[TCM]) and 1-butyl-4-methylpyridinium tricyanomethanide ([4-C4C1py][TCM]), addressing also the phase characterizationof the corresponding {hydrocarbon + IL} binary systems. The phase equilibria were determined by headspace gaschromatography (HS-GC) in a wide range of hydrocarbon compositions, temperatures and S/F ratios and modeled using theCubic Plus Association (CPA) Equation of State (EoS). This is the first work proposing a consistent EoS model for amulticomponent separation case involving ILs, demonstrating its suitability for a wide range of conditions in binary and ternarysystems.

1. INTRODUCTION

In the petrochemical industry, the main sources of benzene,toluene, and xylenes, known as BTX, are pyrolysis andreformer gasolines. In these refinery streams, aromatichydrocarbons are mixed with other compounds that presentclose boiling points and, in some cases, azeotropes.1,2 Hence,the separation of pure compounds from these streams, byconventional distillation, stands technically unfeasible. Currenttechnologies, like extractive distillation and liquid−liquidextraction, employ volatile organic compounds, such assulfolane,3−5 as solvents but tend to present a high energydemand and the requirement of several additional separationsteps/units.6,7 The pursuit of overcoming these limitationsstands as one of the most relevant challenges in this field thus,

aiming at improving the aromatic/aliphatic separation, which isalso transferable for other separations. The design of moreefficient separation processes, with lower environmental impactand energy consumption, is essential to develop bettertechnologies capable to meet, or even enhance, the conven-tional processes standards.Recently, ionic liquids (ILs) have gathered a great deal of

attention as agents in separation processes due to their uniqueproperties, like the possibility of fine-tuning their physical and

Received: August 12, 2019Revised: September 12, 2019Accepted: September 25, 2019Published: September 25, 2019

Article

pubs.acs.org/IECRCite This: Ind. Eng. Chem. Res. 2019, 58, 19681−19692

© 2019 American Chemical Society 19681 DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

Dow

nloa

ded

via

UN

IV O

F A

VE

IRO

003

00 o

n D

ecem

ber

20, 2

019

at 1

0:48

:48

(UT

C).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

extractive properties by a simple combination of their ions,their negligible vapor pressures, low melting points, and highthermal stability, which entail a wide liquid range allowing theILs an easy regeneration, high selectivity, and high extractioncapacity.8−13 After approaching by liquid−liquid extractionand technically solving further purifications and the ILregeneration,14−20 IL-based extractive distillation has beenproposed as an enhancer technology with promising resultsable to overcome the limitations found in the liquid−liquidextraction processes.21−27 In previous works,28,29 the feasibilityof the aromatic/aliphatic separation by extractive distillationusing ILs has been proved in a simplified model composed ofn-heptane and toluene. After screening the most promising ILsin the aromatic/aliphatic separation, tricyanomethanide-basedILs were selected due to their high n-heptane/toluene relativevolatility, low viscosities, and high thermal stability.13

Furthermore, the knowledge on the n-heptane/tolueneseparation was improved by studying the [C2C1im][TCM]and [4-C4C1py][TCM] as mass agents in a wide range oftemperature and S/F ratios for the whole hydrocarboncomposition range. The VLLE/VLE data for {n-heptane/toluene + IL} binary systems and {n-heptane + toluene + IL}ternary systems were properly modeled to CPA EoS, proposinga coarse-grain model with ILs’ molecular parametersindependently regressed from the properties of the solventand with transferable, small, low temperature dependent,binary interaction parameters.29

This work pursues the challenge of studying and modelingthe aromatic/aliphatic separation by extractive distillation withtricyanomethanide-based ILs using the same experimental-modeling strategy, which combines HS-GC techniques andCPA EoS modeling. The VLLE/VLE for other aliphatic-aromatic hydrocarbons pairs with different molecular weight,that is, n-hexane/benzene and n-octane/p-xylene, wereexperimentally studied and modeled. The selection of thesesystems allows the estimation of all CPA EoS binaryinteractions parameters, dependent on the hydrocarbonmolecular weight and temperature, to properly represent theseparation of BTX from a real multicomponent stream, that is,pyrolysis gasoline by extractive distillation with [C2C1im]-[TCM] and [4-C4C1py][TCM] ILs. VLLE and VLE of binary{aliphatic/aromatic + IL} and ternary {aliphatic + aromatic +IL} systems have been determined in the whole hydrocarboncomposition range, from 323.2 to 423.2 K, and an S/F ratio of10, selected in a previous work29 studying the effect of thesolvent concentration is the {hydrocarbon + IL} binarysystems and shown to be representative for ternary systems,as well.

2. EXPERIMENTAL SECTION2.1. Chemicals. The ILs 1-ethyl-3-methylimidazolium

tricyanomethanide, [C2C1im][TCM], and 1-butyl-4-methyl-pyridinium tricyanomethanide, [4-C4C1py][TCM], wereacquired from Iolitec GmbH with a mass fraction purityhigher than 98%. Before its use, the ILs were dried undervacuum (0.1 Pa), moderate temperature (313 K) and undercontinuous stirring for a period never smaller than 48 h.Benzene, p-xylene, n-hexane, and n-octane, were acquired fromSigma-Aldrich with a mass fraction purity higher than 99% andwere kept into their original vessels, with 3 Å molecular sievesto maintain negligible the compounds’ water content. Thecompounds description, abbreviation, supplier, purity, andwater content are reported in Table 1.

2.2. Determination of VLE/VLLE. The VLE and VLLEdata were determined by HS-GC technique based on anAgilent 7697A headspace injector coupled to an Agilent 7890gas chromatograph, fully detailed in previous publications.15,30

The mixtures were gravimetrically prepared in 20 mLhermetically sealed vials by adding the IL and the hydro-carbons using a Mettler Toledo XS205 mass balance, with aprecision of 10−5 g. Equilibration and phase splitting timeswere 2 and 1 h, respectively. Both equilibration and splittingtimes were selected based on previous publications and foundenough to ensure the phase equilibrium.28,29 Knowing the feed(zi), vapor phase volume and compositions, the overall liquidphase composition (xi) were determined by mass balance:

=· − · ·

∑ · − · ·=

xz F p V R T

z F p V R T

( / )

( ( / ))i

i i

i i i

G

13

G (1)

where F denotes the molar amount of the feed, VG the vialheadspace volume and R the ideal gas law constant.Hydrocarbon partial pressures (pi) were also determinedfrom the vapor phase composition, using the relationshipbetween the peak areas of the hydrocarbons (Ai) and the peakareas of each pure hydrocarbon at the same conditions (Ai

0).

=·

pp A

Ai i

ii

0

0(2)

where pi0 refers to the vapor pressure of each pure

hydrocarbon, taken from the literature.31 The total pressurewas calculated as the sum of the partial pressures. However, forthe VLLE measurements, the hydrocarbon distribution,between the liquid phases cannot be calculated withoutadditional experimental information. In these cases, eachmixture point was prepared in duplicate; one for characterizingthe vapor phase as in the VLE assays, and the other tocharacterize the IL-rich liquid phase. The mixtures were mixedduring 2 h in a Labnet Vortemp 1550 shaking incubator toreach equilibrium at the same temperature used in the HS-GCoven. The gas phase was sampled with the Agilent 7697AHeadspace injector and analyzed by the 7890A GC, asexplained above. On the other hand, the IL-rich liquid phasewas directly sampled with a syringe (BD Micro-Fine, 0.3 mL)and analyzed by multiple headspace extraction (MHE)technique. The details and background of the MHE methodare widely explained in a previous work.32 By knowing the feedcomposition of feed, vapor phase and IL-rich liquid phase, and

Table 1. Compound Description, Supplier, Mass FractionPurity and Water Content of the Studied Compounds

chemical supplierpurityin wt %

watercontent in

ppm

1-ethyl-3-methylimidazoliumtricyanomethanide [C2C1im][TCM]

IolitecGmbH

98 <300

1-butyl-4-methylpyridiniumtricyanomethanide [4-C4C1py][TCM]

IolitecGmbH

98 <300

n-hexane Sigma-Aldrich

≥99.0 anhydrous

n-octane Sigma-Aldrich

≥99.0 anhydrous

benzene Sigma-Aldrich

99.8 anhydrous

p-xylene Sigma-Aldrich

≥99.0 anhydrous

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19682

taking into account ILs’ negligible vapor pressures andpractically null solubility in the hydrocarbon-rich liquidphase, the molar amount and composition of hydrocarbon-rich phase were determined by mass balance.

3. CPA EOS MODELINGA simplified CPA EoS was proposed by Kontegeorgis etal.,33,34 combining the Soave−Redlich−Kwong (SRK) physicalcontribution29,30 together with an association term thatincorporates intermolecular interaction, that is, hydrogenbonding and solvation. The CPA EoS approach is expressedin terms of compressibility factor (Z) as the sum of physicaland association contributions:

= +Z Z Zphys assoc (3)

In this work, the three parameters from the physical term (a, b,and c0) for the hydrocarbons (nonassociative compounds)were taken from literature where they were regressed fromtheir vapor pressures and liquid densities,35 whereas those fortricyanomethanide-based ILs, which were considered asassociative compounds (scheme B), were taken from aprevious work where they were regressed from their densitiesand heat capacities, also controlling their nonvolatilecharacter.29 Aiming at modeling the binary and ternarymixtures determined experimentally, a and b were calculatedfollowing the well-known vdW-1f mixing rules:

∑ ∑= = −a x x a a a a k; ( ) (1 )i j

i j jij ij i1/2

ij(4)

∑=b x bi

i i(5)

where kij are the binary interaction parameters, which are theonly adjustable binary parameters used in this work.

4. RESULTS AND DISCUSSION4.1. VLE and VLLE for Binary Systems Composed of

Aliphatics, Aromatics, and Tricyanomethanide-based

Figure 1. VLE Data for (a) {n-hexane (1) + benzene (2)} and (b) {n-octane (1) + p-xylene (2)} binary systems phase diagram. Symbols denotedata from ref 37 and the solid lines the CPA EoS, with k12, fitting to the data.

Table 2. CPA EoS Binary Interaction Parameters (kij) forthe Aliphatic (1) + Aromatic (2) + IL (3) Systems

k13 k23

T/K k12[C2C1im][TCM]

[4-C4C1py][TCM]

[C2C1im][TCM]

[4-C4C1py][TCM]

n-Hexane + Benzene + IL323.2 0.013 0.010 0.045 −0.060 −0.045363.2 0.010 0.020 0.050 −0.055 −0.040403.2 0.007 0.030 0.055 −0.050 −0.035n-Octane + p-Xylene + IL323.2 0.013 −0.040 0.015 −0.050 −0.025363.2 0.010 −0.030 0.020 −0.040 −0.020403.2 0.007 −0.020 0.025 −0.035 −0.015

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19683

ILs: Experimental vs CPA EoS Description. Aiming atchecking the suitability of CPA EoS to describe the VLEbehavior of the studied hydrocarbon, VLE data for the {n-hexane + benzene} and {n-octane + p-xylene} binary systemswere taken from literature, in the temperature range of(323.2−403.2) K,36 and modeled with the EoS. The results arepresented in Figure 1, showing the ability of CPA EoS tocorrectly describe these binary systems using a temperaturedependent k12 binary interaction parameter, as reported inTable 2. Furthermore, the k12 binary interaction parameterproposed for the binary system {n-heptane + toluene},reported in a previous publication, was shown adequate tofurther describe both the {n-hexane + benzene}, {n-octane + p-

xylene} studied here. Having the same binary interactionparameters denote similar deviations to ideality among thealiphatic and aromatic hydrocarbons controlled by the numberof carbon atoms difference, between the species, rather than bythe molecular weight of the species.The VLE/VLLE of the {hydrocarbon + [C2C1im][TCM]/

[4-C4C1py][TCM]} binary systems were determined at 323.2,363.2, and 403.2 K. The VLE/VLLE data are depicted inFigure 2 and reported in Supporting Information (SI) TablesS1 and S2. As expected, the pTx diagrams showed twodifferent trends. First, a VLE region is observed with increasingpressures, as the hydrocarbon mole fraction increases,extending up to the solubility limit of the hydrocarbon in

Figure 2. VLE/VLLE Data for the {[C2C1im][TCM]/[4-C4C1py][TCM] + n-hexane/benzene/n-octane/p-xylene} binary systems. Symbolsdenote experimental data: △, 323.2 K; ○, 363.2 K; ◊, 403.2 K, and the solid lines the CPA EoS with ki3 listed in Table 2.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19684

Figure 3. CPA EoS binary interaction parameters as a function of temperature for the {aliphatic (1) + aromatic (2) + IL (3)} ternary systems: k13(n-hexane, Ο; n-heptane, ◆ (from ref 29); n-octane, △) and k23 (benzene, □; toluene, × (from ref 29); p-xylene, ).

Figure 4. VLE p-x,y and y-x diagrams for the {n-hexane (1) + benzene (2) + [C2C1im][TCM] (3)} ternary system with S/F = 10. Symbols: blackdenotes IL-rich liquid phase, gray refers to hydrocarbon-rich liquid phase, white represents vapor phase, and striped denotes global liquid phasecomposition. Solid lines represent the CPA EoS wth ki3 from Table 2, dotted lines represent the CPA EoS with k13 and ki3 modified at 403.2 K to−0.050 and −0.090, respectively, and dashed refers to benchmark hydrocarbon binary system.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19685

the IL. Then, a VLLE region, in which both hydrocarbon-richand an IL-rich liquid phases coexist, can be identified atconstant pressure as the hydrocarbon mole fraction increases.As depicted in Figure 2, the hydrocarbon solubility in the IL ismostly temperature independent, as previously reported for{hydrocarbon + IL} systems.30,37

The experimental VLE/VLLE data was modeled with CPAEoS with high accuracy. The corresponding binary interactionparameters between all hydrocarbons and the two ILs (ki3) arereported in Table 2 and depicted in Figure 3, as a function oftemperature and number of carbon atoms. As can be seen, theki3 presents a small dependency with temperature, increasingwith increasing temperature. Furthermore, for the same ionicliquid the binary interaction parameters present similarbehavior (similar slope) independently of the studied hydro-carbon. The only exception is observed for the toluene forwhich the toluene-IL binary interaction parameter shows a lesspronounced increase between 323.2 and 363.2 K but increasesabruptly for temperatures above 363.2 K. Moreover, thealiphatics present higher binary interaction parameters thanthose reported for the aromatics, for which a decrease in thebinary interaction parameter is observed with the increase ofthe aliphatic molecular weight; an opposite behavior to thatobserved for the aromatics.

4.2. VLE/VLLE for Aliphatic + Aromatic + IL TernarySystems: Experimental and CPA EoS Description. TheVLE/VLLE for the {n-hexane + benzene + IL} and {n-octane+ p-xylene + IL} were determined at 323.2, 363.2, and 403.2 Kalong the whole hydrocarbon composition range at an S/Fratio of 10, as reported in SI Tables S3−S6.The y-x and p-x,y diagrams, shown in Figures 4, 5, 6−7,

allow to infer that in the case of {n-octane + p-xylene + IL}ternary systems, a homogeneous extractive distillation is notfeasible along the whole hydrocarbon composition range, evenat the highest temperature evaluated (403.2 K). As previouslyreported,28 the region with higher aliphatic content requireshigher temperature to promote a homogeneous vapor−liquidequilibrium. Thus, the lower vapor pressure of n-octane, incomparison with shorter linear alkanes, implies that 403.2 K isnot enough to implement a homogeneous extractivedistillation. Nonetheless, the VLE region, in the whole rangeof compositions, would be found above 403.2 K. In addition,the liquid−liquid equilibria (LLE) of these mixtures could notbe determined at 403.2 K due to technical limitations onsampling reliably of the liquid phase. For this reason in Figures6 and 7 the striped symbols refer to the overall liquid phase,not to the IL-rich and hydrocarbon-rich liquid phases. Bycontrast, in the case of {n-hexane + benzene + IL} ternarysystems, two regions as a function of composition, VLLE and

Figure 5. VLE p-x,y and y-x diagrams for the {n-hexane (1) + benzene (2) + [4-C4C1py][TCM] (3)} ternary system with S/F = 10. Symbols: blackdenotes IL-rich liquid phase, gray refers to hydrocarbon-rich liquid phase, white represents vapor phase, and striped denotes global liquid phasecomposition. Solid lines represent the CPA EoS wth ki3 from Table 2, dotted lines represents the CPA EoS with k13 and ki3 modified at 403.2 K to0.000 and −0.060 respectively, and dashed refers to benchmark hydrocarbon binary system.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19686

VLE, are observed. The former only appears at high n-hexaneconcentrations at the temperatures of 323.2 and 363.2 K.Aiming at clarify the present phases, a scheme is shown in insetof the p-x,y diagrams where black and gray stand for the liquidphases and white for the gas phase. Similarly, the VLE regionincreases as temperature rises, for n-hexane, below its solubilitylimit. Nonetheless, the slope of the p-x,y diagram decreaseswith this increase, meaning a moderate negative impact in theseparation.On the other hand, ternary phase diagrams are depicted in

Figure 8 to allow a better analysis of the LLE region. A widezone of immiscibility between IL and hydrocarbon-rich phaseis observed. In fact, the amount of the IL in the hydrocarbon-rich liquid phase is negligible, whereas the values of thealiphatic and aromatic molar fraction in the IL-rich liquidphase are low and high, respectively, confirming the highselectivity and capacity of the ILs. As expected, an increase intemperature leads to a decrease in LLE region due to thehigher amount of vapor phase and the higher amount ofhydrocarbon in IL-rich liquid phase. Furthermore, ILs tend toremain in the IL-rich liquid phase, and their extractiveproperties remaining almost constant in the 323.2 and 363.2K temperature range.In addition, in Figures 4−8, the excellent description of the

CPA EoS, with an specially accurate description of theequilibrium at lower temperatures is shown, that is, 323.2 and

363.2 K. For the {n-octane + p-xylene + IL} systems, theresults are more reliable due to the lower total pressuresreported. Knowing that the combined uncertainty in pressurewas found to be ur(p) = 0.02, the absolute dispersion in morevolatile systems is higher as pressures are also greater.However, deviations between CPA EoS and experimentalresults, at the highest temperature, imply even qualitativedeviations related to limitations known to the EoS.29 Asreflected in a previous work,29 small changes on the k13 and k23for high temperatures allows reaching an adequate descriptionat these conditions at the low aliphatic concentration region.Since high temperatures are related to high aromatic content,in the extractive distillation column, this approach can beconsidered as suitable from an industrial point of view.For the {n-octane + p-xylene + IL} ternary systems no

additional corrections were required, since the CPA EoS wasable, with the proposed binary interaction parameters, toachieve a good description of the lower volatility systems, in awider range of temperatures. Finally, comparing the y-x and p-x,y diagrams, the CPA EoS provides a better description of theformer, meaning that CPA EoS properly represents theseparation in the whole range of conditions.

4.3. Analysis of the Aliphatic/Aromatic RelativeVolatility. Hereunder, experimental aliphatic/aromatic rela-tive volatility are also evaluated and compared to thoseestimated by the CPA EoS, allowing to confirm the EoS

Figure 6. VLE p-x,y and y-x diagrams for the {n-octane (1) + p-xylene (2) + [C2C1im][TCM] (3)} ternary system with S/F = 10. Symbols: blackdenotes IL-rich liquid phase, gray refers to hydrocarbon-rich liquid phase, white represents vapor pase, and striped denotes global liquid phasecomposition. Solid lines represent the CPA EoS wth ki3 from Table 2 and dashed refers to benchmark hydrocarbon binary system.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19687

robustness for simulation purposes. From the experimentalVLLE/VLE data, the aliphatic/aromatic relative volatilitieshave been calculated as follows:

α = =KK

y x

y x

/

/121

2

1 1

2 2 (6)

where K is the K-value, yi and xi denote the vapor and liquidphase mole fractions of component i, respectively. Thealiphatic/aromatic relative volatility values are depicted inFigure 9, as a function of free-IL hydrocarbon molar fraction,and reported in SI Tables S.3−S.6. In Figure 9, it is possible toidentify similar behaviors among the different systems denotingsimilar separations. The aliphatic/aromatic relative volatilitydecreases with the increase in the aliphatic molar fraction. Inthese cases, a less significant impact of the VLE or VLLEregions, in comparison with the trends observed for n-heptane/toluene is observed.29 This is explained by the higher recoveryof n-hexane, even for VLE, and the lower recovery of n-octane,even for VLLE, in comparison with n-heptane. In this way, alower impact of the type of phase equilibrium, for lower orhigher volatile compounds than n-heptane, is observed.Comparing the IL-based mass agents, [C2C1im][TCM] standsas the most effective and selective IL, on the VLE region, forthe n-hexane/benzene separation, while for the VLLE regionthe [4-C4C1py][TCM] offers higher relative volatility due to

its higher capacity. Conversely, as the n-octane solubility in theIL is much lower, showing almost only VLLE, [4-C4C1py]-[TCM] presents the best separation. As can be seen, overall,CPA EoS model describes these behaviors adequatelyconfirming its suitability for describing the studied systems,even allowing an adequate representation of the derivedseparation properties calculated from the phase equilibria.Higher deviations were found for the n-hexane/benzenesystem, at 363.2 and 403.2 K, explained by the aforementionedlimitations at high temperatures.Overall, the experimental and computational results

evidence that the extractive distillation using ILs stands avery efficient separation of BTX from linear alkanes. Aliphatic/aromatic relative volatilities are highly increased in thepresence of both ILs at any temperature, as shown in Figures4−7. Considering two hydrocarbons with the same number ofcarbon atoms (one aliphatic and one aromatic), the separationis enhanced by reducing the number of the hydrocarboncarbon atoms, that is, the separation of n-hexane from benzeneis easier than n-heptane from toluene, because the aliphatic ismore volatile and the aromatic is more soluble in the IL with awider VLE region.

■ CONCLUSIONSIn this work, the VLE/VLLE for {n-hexane + benzene +[C2C1im][TCM] or [4-C4C1py][TCM]} and {n-octane + p-

Figure 7. VLE p-x,y and y-x diagrams for the {n-octane (1) + p-xylene (2) + [4-C4C1py][TCM] (3)} ternary System with S/F = 10. Symbols: blackdenotes IL-rich liquid phase, gray refers to hydrocarbon-rich liquid phase, white represents vapor phase, and striped denotes global liquid phasecomposition. Solid lines represent the CPA EoS wth ki3 from Table 2 and dashed refers to benchmark hydrocarbon binary system.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19688

xylene + [C2C1im][TCM] or [4-C4C1py][TCM]} systemswere determined by HS-GC and modeled by CPA EoS in a

wide range of temperatures and compositions. Both binary andternary systems have been experimentally determined, the

Figure 8. Ternary LLE diagrams from the VLLE data of {n-heptane + toluene + IL} ternary system at an S/F = 10. Full symbols and solid linesdenote experimental data and empty squares and dashed lines the CPA EoS.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19689

latter at S/F ratio of 10, to evaluate the two separations andgive insights regarding how the number of carbon atoms ofaromatics and aliphatics impacts on the aliphatic/aromaticrelative volatility and the type of extractive distillation(homogeneous or heterogeneous). The CPA EoS has beenused to model all phase equilibria, aiming at providing a wholemodel that represents the BTX separation from linear alkanes.The results have shown that the effectiveness of the

aliphatic/aromatic separation, as well as the homogeneousregion, increases as the number of carbon atoms decreases.Since [C2C1im][TCM] presents the highest selectivity, ahigher relative volatility at VLE region is obtained for n-hexane/benzene separation, but for n-octane/p-xylene the [4-C4C1py][TCM] is the best option along the wholecomposition range due to its higher capacity. However, thealiphatic/aromatic relative volatilities are high enough forseparation purposes in all cases. The proposed model properlydescribed the phase equilibria and the analysis of the binaryinteraction parameters has shown its robustness, evendescribing the nonvolatile character of the solvent and itsdensity and heat capacity.Overall, the results show the extractive distillation to be an

efficient technology to separate BTX from linear alkanes andCPA EoS to be a robust model that properly represents themain features required to further explore the BTX removalfrom industrial streams by extractive distillation withtricyanomethanide-based ILs.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.iecr.9b04440.

Vapor−liquid and vapor−liquid−liquid equilibria forhydrocarbon (1) + [C2C1im][TCM]/[4-C4C1py]-[TCM] (2) binary systems. Vapor−liquid and vapor−liquid−liquid equilibria for n-hexane (1) + benzene (2)+ [C2C1im][TCM]/[4-C4C1py][TCM] (3) ternary

systems with S/F = 10. Vapor−liquid and vapor−liquid−liquid equilibriaa for n-octane (1) + p-xylene (2)+ [C2C1im][TCM]/[4-C4C1py][TCM] (3) ternarysystems with S/F = 10 (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: pablo.navarro@uam.es.

ORCIDPablo Navarro: 0000-0002-0017-3898Andre M. Palma: 0000-0002-5580-6883Marcos Larriba: 0000-0003-0058-6841Julian García: 0000-0003-1386-4003Joao A.P. Coutinho: 0000-0002-3841-743XPedro J. Carvalho: 0000-0002-1943-0006NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was developed in the scope of the project CICECO-Aveiro Institute of Materials, POCI-01-0145-FEDER-007679(ref. FCT UID/CTM/50011/2019) funded by FEDERthrough COMPETE2020 − Programa Operacional Compet-itividade e Internacionalizacao (POCI) and by national fundsthrough FCT − Fundacao para a Ciencia e a Tecnologia. Weare also grateful to Ministerio de Economia y Competitividad(MINECO) of Spain and Comunidad Autonoma de Madridfor financial support of Projects CTQ2017-85340-R andS2013/MAE-2800, respectively. P.N. and P.J.C. also thankFCT for awarding their postdoctoral grant (SFRH/BPD/117084/2016) and contract under the Investigator FCT 2015(IF/00758/2015), respectively. N.D.-M. also thanks MINECOfor awarding her an FPI grant (BES-2015-072855). M.L thanksMinisterio de Eduacion, Cultura y Deporte of Spain forawarding him a Jose Castillejo postdoctoral mobility grant

Figure 9. Aliphatic/aromatic Relative volatility as function of the free-IL aliphatic mole fraction for the {aliphatic + aromatic + IL}. Symbolsrepresent the experimental data and the lines the CPA EoS.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19690

(CAS17/00018). We thank Infochem-KBC for the Multiflashprogram used for the CPA EoS calculations.

■ REFERENCES(1) Franck, H. G.; Stadelhofer, J. W. Industrial Aromatic Chemistry;Springer-Verlag: Berlin, 1988.(2) Gary, J. H.; Handwerk, G. E. Petroleum Refining. Technology andEconomics, 4th ed.; Marcel Dekker: New York, 2001.(3) Chen, J.; Duan, L. P.; Mi, J. G.; Fei, W. Y.; Li, Z. C. Liquid-Liquid Equilibria of Multi-Component Systems Including n-Hexane,n-Octane, Benzene, Toluene, Xylene and Sulfolane at 298.15 K andAtmospheric Pressure. Fluid Phase Equilib. 2000, 173 (1), 109−119.(4) Choi, Y. J.; Cho, K. W.; Cho, B. W.; Yeo, Y. K. Optimization ofthe Sulfolane Extraction Plant Based on Modeling and Simulation.Ind. Eng. Chem. Res. 2002, 41, 5504−5509.(5) Lei, Z.; Li, C.; Chen, B. Extractive Distillation: A Review. Sep.Purif. Rev. 2003, 32 (2), 121−213.(6) Schneider, D. F. Avoid Sulfolane Regeneration Problems. Chem.Eng. Prog. 2004, 100 (7), 34−39.(7) De Graff, R. R. Aromatic Hydrocarbon Recovery Process. U.S.Patent 4048062 1976.(8) Plechkova, N. V.; Seddon, K. R. Applications of Ionic Liquids inthe Chemical Industry. Chem. Soc. Rev. 2008, 37 (1), 123−150.(9) Canales, R. I.; Brennecke, J. F. Comparison of Ionic Liquids toConventional Organic Solvents for Extraction of Aromatics fromAliphatics. J. Chem. Eng. Data 2016, 61 (5), 1685−1699.(10) Meindersma, G. W.; Hansmeier, A. R.; De Haan, A. B. IonicLiquids for Aromatics Extraction. Present Status and Future Outlook.Ind. Eng. Chem. Res. 2010, 49 (16), 7530−7540.(11) Rogers, R. D.; Seddon, K. R. Perspective Article: Ionic LiquidsSolvents of the Future? Science (Washington, DC, U. S.) 2003, 302,792−793.(12) Ferreira, A. R.; Freire, M. G.; Ribeiro, J. C.; Lopes, F. M.;Crespo, J. G.; Coutinho, J. A. P. Overview of the Liquid-LiquidEquilibria of Ternary Systems Composed of Ionic Liquid andAromatic and Aliphatic Hydrocarbons, and Their Modeling byCOSMO-RS. Ind. Eng. Chem. Res. 2012, 51 (8), 3483−3507.(13) Larriba, M.; Navarro, P.; Delgado-Mellado, N.; Stanisci, V.;García, J.; Rodríguez, F. Separation of Aromatics from N-AlkanesUsing Tricyanomethanide-Based Ionic Liquids: Liquid-Liquid Ex-traction, Vapor-Liquid Separation, and Thermophysical Character-ization. J. Mol. Liq. 2016, 223, 880−889.(14) Navarro, P.; Larriba, M.; García, J.; Rodríguez, F. Design of theRecovery Section of the Extracted Aromatics in the Separation ofBTEX from Naphtha Feed to Ethylene Crackers Using [4empy]-[Tf2N] and [Emim][DCA] Mixed Ionic Liquids as Solvent. Sep.Purif. Technol. 2017, 180, 149−156.(15) Navarro, P.; Larriba, M.; García, J.; Rodríguez, F. Design of theHydrocarbon Recovery Section from the Extract Stream of theAromatic Separation from Reformer and Pyrolysis Gasolines Using aBinary Mixture of [4empy][Tf2N] + [Emim][DCA] Ionic Liquids.Energy Fuels 2017, 31 (1), 1035−1043.(16) Larriba, M.; Navarro, P.; García, J.; Rodríguez, F. Separation ofToluene from N-Heptane, 2,3-Dimethylpentane, and CyclohexaneUsing Binary Mixtures of [4empy][Tf2N] and [Emim][DCA] IonicLiquids as Extraction Solvents. Sep. Purif. Technol. 2013, 120, 392−401.(17) Larriba, M.; Navarro, P.; Gonzalez, E. J.; García, J.; Rodríguez,F. Separation of BTEX from a Naphtha Feed to Ethylene CrackersUsing a Binary Mixture of [4empy][Tf2N] and [Emim][DCA] IonicLiquids. Sep. Purif. Technol. 2015, 144, 54−62.(18) Larriba, M.; Navarro, P.; Gonzalez, E. J.; García, J.; Rodríguez,F. Dearomatization of Pyrolysis Gasolines from Mild and SevereCracking by Liquid-Liquid Extraction Using a Binary Mixture of[4empy][Tf2N] and [Emim][DCA] Ionic Liquids. Fuel Process.Technol. 2015, 137, 269−282.(19) Larriba, M.; Navarro, P.; García, J.; Rodríguez, F. Liquid-LiquidExtraction of BTEX from Reformer Gasoline Using Binary Mixtures

of [4empy][Tf2N] and [Emim][DCA] Ionic Liquids. Energy Fuels2014, 28 (10), 6666−6676.(20) Larriba, M.; de Riva, J.; Navarro, P.; Moreno, D.; Delgado-Mellado, N.; García, J.; Ferro, V. R.; Rodríguez, F.; Palomar, J.COSMO-Based/Aspen Plus Process Simulation of the AromaticExtraction from Pyrolysis Gasoline Using the {[4empy][NTf 2 ] +[Emim][DCA]} Ionic Liquid Mixture. Sep. Purif. Technol. 2018, 190(September 2017), 211−227.(21) Lei, Z.; Dai, C.; Zhu, J.; Chen, B. Extractive Distillation withIonic Liquids: A Review. AIChE J. 2014, 60, 3312−3329.(22) Paduszynski, K.; Krolikowski, M.; Zawadzki, M.; Orzeł, P.Computer-Aided Molecular Design of New Task-Specific IonicLiquids for Extractive Desulfurization of Gasoline. ACS SustainableChem. Eng. 2017, 5 (10), 9032−9042.(23) Jongmans, M. T. G.; Schuur, B.; De Haan, A. B. Ionic LiquidScreening for Ethylbenzene/Styrene Separation by ExtractiveDistillation. Ind. Eng. Chem. Res. 2011, 50 (18), 10800−10810.(24) Brondani, L. B.; Flores, G. B.; Soares, R. P. Modeling andSimulation of a Benzene Recovery Process by Extractive Distillation.Braz. J. Chem. Eng. 2015, 32 (1), 283−291.(25) Song, Z.; Li, X.; Chao, H.; Mo, F.; Zhou, T.; Cheng, H.; Chen,L.; Qi, Z. Computer-aided ionic liquid design for alkane/cycloalkaneextractive distillation process. Green Energy Environ. 2019;DOI: 10.1016/j.gee.2018.12.001.(26) Graczova, E.; Sulgan, B.; Barabas, S.; Steltenpohl, P. MethylAcetate−Methanol Mixture Separation by Extractive Distillation:Economic Aspects. Front. Chem. Sci. Eng. 2018, 12 (4), 670−682.(27) Su, Y.; Hu, Y.; Shen, W.; Chien, I.-L.; Jin, S. SystematicApproach for Screening Organic and Ionic Liquid Solvents inHomogeneous Extractive Distillation Exemplified by the Tert-ButanolDehydration. Sep. Purif. Technol. 2019, 211 (September 2018), 723−737.(28) Navarro, P.; Larriba, M.; Delgado-Mellado, N.; Ayuso, M.;Romero, M.; García, J.; Rodríguez, F. Experimental Screening towardsDeveloping Ionic Liquid-Based Extractive Distillation in theDearomatization of Refinery Streams. Sep. Purif. Technol. 2018, 201,268−275.(29) Navarro, P.; Ayuso, M.; Palma, A. M.; Larriba, M.; Delgado-Mellado, N.; García, J.; Rodríguez, F.; Coutinho, J. A. P.; Carvalho, P.J. Toluene/ n -Heptane Separation by Extractive Distillation withTricyanomethanide-Based Ionic Liquids: Experimental and CPA EoSModeling. Ind. Eng. Chem. Res. 2018, 57 (42), 14242−14253.(30) Navarro, P.; Larriba, M.; García, J.; Gonzalez, E. J.; Rodríguez,F. Vapor-Liquid Equilibria of {n-Heptane+toluene+[Emim][DCA]}System by Headspace Gas Chromatography. Fluid Phase Equilib.2015, 387, 209−216.(31) Perry, R. H.; Green, D. W.; Maloney, J. O. Perry’s ChemicalEngineers’ Handbook, 7th ed.; New York, 1997; DOI: 10.1021/ed027p533.1.(32) Larriba, M.; Navarro, P.; García, J.; Rodríguez, F. Liquid-LiquidExtraction of Toluene from Heptane Using [Emim][DCA], [Bmim]-[DCA], and [Emim][TCM] Ionic Liquids. Ind. Eng. Chem. Res. 2013,52 (7), 2714−2720.(33) Kontogeorgis, G. M.; Michelsen, M. L.; Folas, G. K.; Derawi, S.;Von Solms, N.; Stenby, E. H. Ten Years with the CPA (Cubic-Plus-Association) Equation of State. Part 1. Pure Compounds and Self-Associating Systems. Ind. Eng. Chem. Res. 2006, 45 (14), 4855−4868.(34) Kontogeorgis, G. M.; Michelsen, M. L.; Folas, G. K.; Derawi, S.;Von Solms, N.; Stenby, E. H. Ten Years with the CPA (Cubic-Plus-Association) Equation of State. Part 2. Cross-Associating andMulticomponent Systems. Ind. Eng. Chem. Res. 2006, 45 (14),4869−4878.(35) Kontogeorgis, G. M.; Folas, G. K. Thermodynamic Models forIndustrial Applications-From Classical and Advanced Mixing Rules toAssociation Theories; John Wiley & Sons: Hoboken, NJ, 2010.(36) Technology, A. Aspen Plus Version 9, Database. 2016.(37) Navarro, P.; Larriba, M.; García, J.; Gonzalez, E. J.; Rodríguez,F. Selective Recovery of Aliphatics from Aromatics in the Presence of

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19691

the {[4empy][Tf2N] + [Emim][DCA]} Ionic Liquid Mixture. J.Chem. Thermodyn. 2016, 96, 134−142.

Industrial & Engineering Chemistry Research Article

DOI: 10.1021/acs.iecr.9b04440Ind. Eng. Chem. Res. 2019, 58, 19681−19692

19692