lable at ScienceDirect

Fluid Phase Equilibria 419 (2016) 57e66

Contents lists avai

Fluid Phase Equilibria

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

Designing high ionicity ionic liquids based on 1-ethyl-3-methylimidazolium ethyl sulphate for effective azeotrope breaking

Filipe S. Oliveira a, Ralf Dohrn b, Ana B. Pereiro a, Jo~ao M.M. Araújo a, Luís P.N. Rebelo a,Isabel M. Marrucho a, *

a Instituto de Tecnologia Química e Biol�ogica Ant�onio Xavier, Universidade Nova de Lisboa, Apartado 127, 2780-157 Oeiras, Portugalb Bayer Technology Services GmbH, 51368 Leverkusen, Germany

a r t i c l e i n f o

Article history:Received 12 October 2015Received in revised form3 March 2016Accepted 3 March 2016Available online 15 March 2016

Keywords:Ionic liquidsInorganic saltsAzeotrope mixturesIonicityLiquideliquid equilibria

* Corresponding author.E-mail address: imarrucho@itqb.unl.pt (I.M. Marru

http://dx.doi.org/10.1016/j.fluid.2016.03.0040378-3812/© 2016 Elsevier B.V. All rights reserved.

a b s t r a c t

Separation of alkanols and alkanes is a complex process in the production of oxygenated additives andfuels due their close boiling points and azeotrope formation. In this work, liquideliquid extraction ofethanol from n-heptane þ ethanol mixtures using high ionicity ionic liquids (HIILs) as extraction solventsis studied. Three different HIILs, with different ionicities were prepared through the addition of differentamounts of ammonium thiocyanate inorganic salt to 1-ethyl-3-methylimidazolium ethyl sulphate ionicliquid, and used. Liquideliquid equilibria of ternary mixtures of n-heptane þ ethanol þ 3 different HIILswere experimentally measured at 298.15 K and 0.1 MPa. Modelling of the experimental tie lines andbinodal curves was performed by using the non-random two-liquid (NRTL) activity coefficient model;each HIIL (IL þ salt) was treated as a pseudo-component. Both the selectivity and the distribution co-efficient were used in the assessment of the extraction solvent feasibility and a correlation of theseparameters with the ionicity of the HIILs was established. A comparison between the different HIILs andthe neat ionic liquid is also made.

© 2016 Elsevier B.V. All rights reserved.

1. Introduction

In many areas of industry, from commodities to fine chemicalprocesses, the accumulation of hazardous solvent mixtures, due torecycling difficulties, is a serious problem. In order to meet sus-tainability criteria, the separation of these mixtures into their purecomponents is mandatory so that they can be reused. For example,numerous problems related to the trade-off between efficiency andsoot gases emissions are posed in the production and commerci-alisation of fuels. In these processes, alkanols and alkanes arebrought together to produce oxygenated additives for gasolines ordiluted hydrocarbon fuels. Several studies on the separation ofazeotropes of either n-hexane or n-heptane with methanol orethanol have already been reported due to difficulties in separatingthese compounds [1e18]. Specifically, the separation of the azeo-trope n-heptane þ ethanol has been intensively studied by manyauthors, using ionic liquids (ILs) and liquideliquid separation pro-cesses [19].

cho).

Seoane et al. [18] studied the effect of increasing the alkyl chainof the 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([C2MIM][NTf2]) on the separation of n-heptane þ ethanolmixtures and found that this increase lead to lower selectivityvalues, while the distribution coefficient values were not greatlyaffected. Gonz�alez et al. [7] tested two pyridinium-based ILs, 1-ethyl-3-methylpyridinium bis(trifluoromethylsulfonyl)imide ([C2-3-C1Py][NTf2]) and 1-propyl-3-methylpyridinium bis(tri-fluoromethylsulfonyl)imide ([C3-3-C1Py][NTf2]) and demonstratedthe lower performance of these ILs when compared to their imi-dazolium counterparts. However, in another work the same au-thors [6] showed that pyrrolidinium-based ILs could provide betterresults than other ILs families. In this work these authors combinedthe 1-butyl-1-methylpyrrolidinium cation ([C4-3-C1pyr]þ) with thedicyanamide ([DCA]�) and the trifluoromethanesulfonate ([OTf]�)anions. The results showed that the [C4-3-C1pyr][DCA] was themore feasible extraction solvent for the separation of n-heptaneþ ethanol mixtures showing distribution coefficient valuessuperior to 8 and selectivity values ranging from 80 to 1500. Arandaet al. [1] used two tetraalkyl ammoniums, butyl-trimethylammonium bis(trifluoromethylsulfonyl)imide ([BTMA][NTf2]) and tributylmethylammonium bis(trifluoromethylsulfonyl)

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e6658

imide ([TBMA][NTf2]), for the separation of n-heptane or n-hexane þ ethanol or methanol mixtures, where the best resultswere obtained for [BTMA][NTf2] in the n-hexane þ methanol sys-tem, whereas for the n-heptane þ ethanol system the studied ILstested performed more poorly than the pyridinium-based ILsstudied by Gonz�alez et al. [7]. Pereiro et al. studied the separation ofn-heptane þ ethanol, using the ILs 1-butyl-3-methylimidazoliummethyl sulphate [C4MIM][C1SO4] [10], 1,3-dimethylimidazoliummethyl sulphate ([C1MIM][C1SO4]) [13] and also 1-ethyl-3-methylimidazolium ethyl sulphate ([C2MIM][C2SO4]) [9]. The re-sults obtained showed that all of the ILs used were suitableextraction solvents for the extraction of ethanol from its azeotropicmixture, especially in mixtures of n-heptane þ ethanol, whereparticularly high efficiencies were obtained. In addition, it wasshown that a shorter alkyl chain on the imidazolium ion led tobetter n-heptane/ethanol efficiencies. Moreover, Pereiro et al.[11,15] also tested hexafluorophosphate-based ILs, 1-hexyl-3-methylimidazolium ([C6MIM][PF6]) and 1-octyl-3-methylimidazolium hexafluorophosphate ([C8MIM][PF6]), for theseparation of n-heptane þ ethanol mixtures. Both the ILs proved tobreak the azeotropic point of the testedmixtures, with the [C6MIM][PF6] IL showing very high selectivity values. However, the lowdistribution coefficients and its solutropic behaviour makes the useof this IL inadvisable. In addition, Cai et al. [2] tested ILs based on di-alkyl phosphate anions for the extraction of ethanol from n-heptane þ ethanol azeotropic mixtures. The extraction capabilitiesof the ILs tested showed to decrease with the increase in the alkylchain of both cation and anion. The IL 1,3-dimethylimidazoliumdimethylphosphate ([C1MIM][DMP]) displayed both higher distri-bution coefficient and selectivity values than the other two ILstested, 1-ethyl-3-methylimidazolium diethylphosphate ([C2MIM][DEP]) and 1-butyl-3-methylimidazolium dibutylphosphate([C4MIM][DBP]). However, in comparison with other neat ILs,[C1MIM][DMP] still presents lower selectivity and distribution co-efficient values than the alkyl sulphate-based ILs.

In an effort to boost the ILs power as separation agents forazeotropic mixtures, Lei et al. [20] recently combined ILs withinorganic salts (ISs) and used them in the separation of ethanol andwater azeotropic mixtures. These authors tested several differentISs based on potassium and sodium cations combined with acetate,chloride and thiocyanate anions, among others. The experimentalresults showed that in all cases the combination of IL with IS yiel-ded better results than the neat IL. Up to date, no other studiesregarding the use of IL þ IS mixtures for the separation of azeo-tropicmixture have been found. Nevertheless, in order to use ILþ ISmixtures as separation agents is mandatory to previously study thesolubility of the IS in the IL, as well as the impact of the former inthe physical and chemical properties of the IL.

Lately, our group studied the solubility of common ISs in a widerange of different ILs [21] and showed that their solubilisation inthe IL media can increase the Coulombic character of the latter, thusincreasing the ionicity of ILs at very low cost, while the liquid statestatus is still preserved [22e24]. The resulting purely ionic liquidmedia produced was named high ionicity ionic liquids (HIILs). Theionicity of an IL is related to its ionic nature, which can be controlledby the magnitude and balance of the interactive forces of the IL. ILspresent a complex nature where several interactions, such asCoulombic (the predominant), van der Waals, hydrogen-bondingand pep interactions, are present. In addition, the formation ofaggregates or clusters in ILs may also occur to some extent,affecting their structure and obviously their physicochemicalproperties such as viscosity, conductivity, and diffusion coefficients.Other IL's properties, namely vapour pressure and hydrogenacceptor or donor character, have also been linked to their ionicity[25]. Thus, the evaluation of the ionicity or the degree of

dissociation/association of ILs and its correlation to their macro-scopic properties has become an interesting and importantparameter for the characterisation of ILs [26,27].

In previous works [22,24], we studied the effects of the additionof ammonium thiocyanate on the thermophysical properties ofthree ILs based on the 1-ethyl-3-methylimidazolium cation com-bined with ethyl sulphate, ethyl sulfonate and acetate anions. NMR,Raman and MD calculations showed that by solubilising this saltinto the IL media, modifications on the IL's initial structure werepromoted and the Coulombic character of the IL altered. Althoughan increase in the ionicity was observed in all the studied systems,this property did not change in a linear manner with compositionindicating the formation of different complexes/aggregatesdepending on the IS concentration, in other words, different ion's“availability” to participate in solvation schemes.

In the present work, our aim is to evaluate the performance ofstudied HIILs as extraction solvents for n-heptane þ ethanolazeotropic mixtures in liquideliquid extraction and, if possible, tolink their ionicity to their extraction efficiency and capacity. For thatpurpose, liquideliquid equilibria data for ternary systems of n-heptane þ ethanol þ HIIL were measured at 298.15 K and 0.1 MPa.Literature shows that alkyl sulphate-based ILs are some of the mostpromising ILs, presenting high efficiency in the separation ofazeotropic mixtures, namely in the separation of mixtures ofbenzene þ C6-C9 aliphatic compounds [28,29], ethylacetate þ ethanol [30] or 2-propanol [31], ethyl tert-butylether þ ethanol [32] and n-hexane or n-heptane þ ethanol[9,10,12e14]. This family of ILs, specifically those with a cationderived from imidazolium, exhibit good chemical and thermalstability, low melting points and relatively low viscosities [33]. Inparticular, the 1-ethyl-3-methylimidazolium ethyl sulphate IL ishighly efficient in the separation of the n-heptane þ ethanolazeotropic mixture [9]. Consequently, in this work, three differentHIILs, based on 1-ethyl-3-methylimidazolium ethyl sulphate ILwith different amounts of ammonium thiocyanate, were used.Their separation capacity was evaluated through the calculation ofthe decisive parameters: distribution coefficient and selectivity.These HIILs were selected for several reasons: i) the IL selectedshows good efficiency in the separation of the targetedmixture [9];ii) the IS chosen displays high solubility in the IL [21], allowing thestudy of different concentrations of IS; and iii) the commercialavailability and cheap price of the IL when comparing to other shortchain alkyl sulphate-based imidazolium ILs.

2. Materials, methods and modelling

2.1. Chemicals

N-heptane and ethanol were supplied by Riedel-de-Ha€en with99 wt% purity and by Scharlau with 99.9 wt% purity, respectively.Ammonium thiocyanate ([NH4][SCN]) was provided by Sigma-eAldrich with a purity content superior to 99.0 wt%, while the IL 1-ethyl-3-methylimidazolium ethyl sulphate ([C2MIM][C2SO4]) waspurchased from Merck with a purity of �99 wt%. To reduce thewater and other volatile substances contents, vacuum (0.1 Pa) andmoderate temperature (no more than 323 K) were always appliedto the ionic liquid and inorganic salt for at least 3 days prior to theiruse. After drying, the ionic liquid purity was checked by 1H NMR. Asummary of the sample provenance and purity is presented inTable 1.

2.2. HIILs

Three different concentrations of binary mixtures of [C2MIM][C2SO4] þ [NH4][SCN] were prepared with the following molar

Table 1Provenance and purity of the used chemicals.

Chemicals Provenance Final mass fraction Purification method

N-heptane Riedel-de-Ha€en 99 wt% NoneEthanol Scharlau 99 wt% NoneAmmonium thiocyanate SigmaeAldrich �99.0 wt%, None1-ethyl-3-methylimidazolium ethyl sulphate Merck �99 wt% Vacuum and moderate temperature

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e66 59

fractions: xIL ¼ 0.8296 þ xIS ¼ 0.1704 (HIIL17),xIL ¼ 0.6698 þ xIS ¼ 0.3302 (HIIL33) and xIL ¼ 0.5492 þ xIS ¼ 0.4508(HIIL45). These concentrations were chosen taking into account thesolubility limits [21] and the ionicity of the system [24]. Table 2presents the ionicity of the used HIILs, calculated by the WaldenPlot method.

In this method, the ionicity is determined by the deviation of thecorresponding HIIL to the ideal Walden line (behaviour of idealelectrolyte). A higher deviation to the ideal line corresponds to a“less ionic” ionic liquid. The HIIL33 was chosen due to its highestionicity (the lowest deviation to the ideal Walden line), while theother two (HIIL17 and HIIL45) present similar ionicity values,despite the fact that HIIL45 has a much higher viscosity and lowerconductivity than HIIL17 [24]. The samples were prepared byweighing known masses of the each component into stopperedflasks using an analytical high-precision balance with an uncer-tainty of ±10�5 g, under inert atmosphere. Afterwards, they weremixed with a magnetic stirring until a clear solution was obtained.In Fig. 1, the chemical structures of [C2MIM][C2SO4] and [NH4][SCN]are presented.

2.3. Liquideliquid equilibria measurements

The ternary LLE experiments were performed at 298.15 K and0.1 MPa in a glass cell thermostatically regulated by a water jacketconnected to an external water bath controlled to ±0.01 K. Thetemperature in the cell was measured with a platinum resistancethermometer coupled to a Keithley 199 System DMM/Scanner thatwas calibrated with higheaccuracy mercury thermometers(0.01 K). The mixing was assured by a magnetic stirrer.

The solubility curve of the ternary system was determined bypreparing several binary mixtures of HIIL þ n-heptane in theimmiscible region and then ethanol was added until the ternarymixture became miscible. Afterwards, the refractive index of thoseternary mixtures was determined in triplicate. The compositions ofthemiscible ternarymixtures used in the solubility curve, as well astheir corresponding refractive index, are depicted in Tables S1eS3in the ESI for each of the HIILs studied in this work. The measure-ments of refractive index were performed at 303.15 K and 0.1 MPa,to ensure no phase splitting occurred, using an automated AntonPaar Refractometer Abbemat 500 with an absolute uncertainty inthe measurement of ±0.00005.

The determination of the tie-lines, was carried out by preparingternary mixtures of known composition that were vigorously

Table 2Ionicity (DW) calculated from theWalden Plot deviations at 298.15 K and 0.1 MPa forthe neat IL and the HIILs tested in this work.

Solvent Mole fraction of IS DWa

[C2MIM][C2SO4] 0 0.164HIIL17 0.1704 0.088HIIL33 0.3302 0.061HIIL45 0.4508 0.084

a The ionicity data was taken from Ref. [24]. Standard uncertainties u areu(x) ¼ 0.0001.

stirred for at least 1 h and left to settle for at least 12 h at 298.15 K.Then, samples from both phases were taken with a syringe andtheir refractive indexes measured in triplicate. The composition ofboth phases in equilibrium was determined by MATLAB, using thefitting of the refractive indexes at 303.15 K with the composition,along the solubility curve, using the following equations:

nD ¼ A,w1 þ B,w21 þ C,w3

1 þ D,w41 þ E,w2 þ F,w2

2 þ G,w32

þ H,w42 þ I,w3 þ J,w2

3 þ K,w33 þ L,w4

3

(1)

w3 ¼ M,exph�

N,w0:51

���O,w3

1

�i(2)

w3 ¼ 1� ðw1 þw2Þ (3)

where,w1,w2 andw3 correspond to themass fraction compositionsof n-heptane, ethanol and HIIL, respectively, and the parameters Ato O are adjustable parameters, which are given in Table S4 in theESI for the studied systems. The use of only one physical property,refractive index, can be accepted as correct due to the proximity ofthe studied compositions to the binary system.

The method was validated using experimental data for theternary system of n-heptane þ ethanol þ [C2MIM][C2SO4] from theliterature. The measurements obtained were estimated to be ac-curate to ±0.009 in mass fraction in comparison with the methodused in literature [9] and the uncertainty in the composition isestimated to be ±0.006 in mass fraction. In addition, to assure thatthe HIIL did not decompose, by the preferential dissolution of oneof the compounds (IL or IS) in one of the phases, a ternary mixtureof known composition (23.5 wt% HIIL þ 34 wt% ethanol þ 42.5 wt%n-heptane) was prepared in the immiscible region and vigorouslystirred for at least 1 h and left to settle for at least 12 h at 298.15 K.Afterwards, the two phases were analysed by 1H NMRwhere it waspossible to confirm, by the peak integration, that the IL and the ISdo not separate from each other. This procedure was performed foreach of the HIILs. Figs. 2e4 depict the 1H spectrum of the pureHIIL45, the HIIL-rich phase and the n-heptane-rich phase, respec-tively. Figs. S1e6 in the ESI represent the same for the other twoHIILs.

2.4. Modelling

Excess Gibbs free energy models, also called activity coefficientmodels, are widely used for the calculation of phase equilibriaincluding liquideliquid equilibria. During the last decade they havebeen increasingly applied to the modelling of phase equilibria insystems containing ILs, e.g. using the non-random two-liquid(NRTL) model [34e37] or the universal quasi-chemical (UNIQUAC)model [35,37e40]. For reasons of flexibility and simplicity con-cerning IL þ IS systems, the NRTL model was used in this work.Using the NRTL model, the activity coefficient of component i iscalculated as follows:

Fig. 1. Chemical structures of both ionic liquid and inorganic salt used in this work, a) [C2MIM][C2SO4] and b) [NH4][SCN].

Fig. 2. 1H spectrum of the pure HIIL45.

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e6660

ln gi ¼

PjtjiGjixj

PkGkixk

þXj

GijxjPkGkjxk

0B@tij �

PntnjGnjxn

PkGkjxk

1CA (4)

with

Gij ¼ exp�� aij,tij

�(5)

Usually, the temperature dependence of interaction parametertij is described by an approach with two adjustable parameters aijand bij

tij ¼ aij þbijT

(6)

In this study, no temperature dependence was used and all bijwere set to zero since the experimental datawere taken at the sametemperature, 298.15 K. The three investigated HIILs (IL þ certainamount of salt) were treated as a pseudo-component, each, ofwhich the molar mass was calculated from the molar mass of the ILand of the salt according to the total composition of the HIIL.

The non-randomness factors aij for the interactions between n-heptane (1) and ethanol (2) as well as n-heptane (1) and HIIL (3)were set to the standard value of Aspen Plus, which is 0.3. Adifferent value was used for the interaction between ethanol (2)and HIIL (3), a23 ¼ �0.1, which leads in all three ternary systems

Fig. 3. 1H spectrum of the HIIL-rich phase for the ternary mixture of n-heptane þ ethanol þ HIIL45.

Fig. 4. 1H spectrum of the n-heptane-rich phase for the ternary mixture of n-heptane þ ethanol þ HIIL45.

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e66 61

investigated to a significantly better description of the binodalcurve and of the tie lines as compared to the standard value

a23 ¼ 0.3. In Table 3 all binary interaction parameters used for themodelling are given. The calculation of the phase equilibria was

Table 3Binary interaction parameters (NRTL model) used for the calculations.

Component i Component j aij aji aij

n-Heptane Ethanol 1,5 0,9 0.3HIIL17 4.0 4.1 0.3HIIL33 3.6 4.6 0.3HIIL45 3.2 5.1 0.3

Ethanol HIIL17 �0.4 �10.1 �0.1HIIL33 �0,7 �12 �0.1HIIL45 �1.6 �13 �0.1

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e6662

performed with the help of the process modelling system AspenPlus 8.4.

Fig. 6. Ternary diagram for the system n-heptane (1) þ ethanol (2) þ HIIL33 (3) at298.15 K and 0.1 MPa. The red dots represent the solubility curve and the red squaresthe composition of the initial mixture; the black dots and full lines represent thecomposition of the co-exiting phases and the experimental tie-lines, respectively; theblack points and dotted lines represent the calculated tie-lines.

3. Results and discussion

The ternary diagrams for the system n-heptaneþ ethanolþHIILat 298.15 K and 0.1 MPa are presented in Figs. 5e7. It can beobserved that an increase in the molar ratio of IS in the HIIL leads toan increase in the immiscibility region of the ternary diagram. Inaddition, the tie-lines of the three studied systems present positiveslopes meaning that the extraction of ethanol from n-heptane isalways favourable. The composition of the phases in equilibrium,along with the distribution coefficient and the selectivity values arepresented in Table 4.

In Figs. 5e7 the results of the calculated phase equilibria arecompared with the experimental data. The slopes of the tie linesagree well. Also, the size of the liquideliquid region is well repre-sented by the NRTLmodel. Close to the critical point the parametersused here lead to too wide calculated liquideliquid region. The setof binary interactions parameters used for the modelling (Table 3)describes the increasing size of the liquideliquid region fromHIIL17 to HIIL33 and to HIIL45 very well. There is almost a lineardependence between the size of the parameters and the salt con-tent of the mixture. Therefore, it would be possible to calculate thephase behaviour depending on the salt content.

The distribution coefficient and the selectivity are two requiredparameters to assess the performance of the extraction solvent in

Fig. 5. Ternary diagram for the system n-heptane (1) þ ethanol (2) þ HIIL17 (3) at298.15 K and 0.1 MPa. The green dots represent the solubility curve and the greensquares the composition of the initial mixture; the black dots and full lines representthe composition of the co-existing phases and the experimental tie-lines, respectively;the black points and dotted lines represent the calculated tie-lines. (For interpretationof the references to colour in this figure legend, the reader is referred to the webversion of this article.)

Fig. 7. Ternary diagram for the system n-heptane (1) þ ethanol (2) þ HIIL45 (3) at298.15 K and 0.1 MPa. The orange dots represent the solubility curve and the orangesquares the composition of the initial mixture; the black dots and full lines representthe composition of the co-exiting phases and the experimental tie-lines, respectively;the black points and dotted lines represent the calculated tie-lines. (For interpretationof the references to colour in this figure legend, the reader is referred to the webversion of this article.)

liquideliquid extraction systems. The first accounts for the solute-carrying capacity as well as the amount of extraction solvent (HIIL,in this case) required for the extraction process, while the latterevaluates the efficiency of the extraction solvent, indicating theease of extraction of a solute (ethanol) from a diluent or carrier(heptane). These two parameters can be determined by thefollowing equations:

b2 ¼ wII2

wI2

(7)

Table 4Composition of the experimental tie-lines, ethanol distribution coefficient (b2) andselectivity (S) for the ternary system n-Heptane þ Ethanol þ HIIL at 298.15 K and0.1 MPa.

Heptane-rich phase HIIL-rich phase b2 S

w1I w2

I w1II w2

II

N-heptane (1) þ Ethanol (2) þ HIIL17 (3)0.994 0.004 0.002 0.1130.989 0.009 0.004 0.1750.984 0.014 0.009 0.242 17.3 18900.974 0.024 0.014 0.282 11.8 8170.959 0.039 0.024 0.347 8.9 3560.944 0.053 0.035 0.393 7.4 2000.913 0.083 0.057 0.452 5.4 87.20.898 0.097 0.095 0.507 5.2 49.40.849 0.143 0.176 0.543 3.8 18.3N-heptane (1) þ Ethanol (2) þ HIIL33 (3)0.995 0.004 0.000 0.0400.989 0.010 0.001 0.0890.984 0.014 0.002 0.138 9.9 48500.969 0.029 0.003 0.176 6.1 19600.959 0.039 0.008 0.251 6.4 7720.949 0.049 0.021 0.349 7.1 3220.929 0.068 0.032 0.404 5.9 1720.908 0.088 0.053 0.466 5.3 90.70.868 0.127 0.081 0.513 4.0 43.30.847 0.147 0.117 0.545 3.7 26.80.827 0.166 0.199 0.562 3.4 14.1N-heptane (1) þ Ethanol (2) þ HIIL45 (3)0.994 0.004 0.001 0.1100.989 0.009 0.003 0.1740.979 0.018 0.005 0.228 12.7 24800.969 0.028 0.008 0.278 9.9 12030.958 0.038 0.013 0.325 8.6 6300.943 0.053 0.021 0.386 7.3 3270.928 0.068 0.029 0.422 6.2 1990.913 0.082 0.043 0.475 5.8 1230.903 0.092 0.063 0.523 5.7 81.50.856 0.136 0.104 0.572 4.2 34.6

Standard uncertainties u are u(T) ¼ 0.01 K, u(x) ¼ 0.006.

Fig. 8. Distribution coefficient values, b2, for the ternary systems n-heptane(1) þ ethanol (2) þ HIIL (3), as a function of ethanol mass fraction in n-heptane-richphase at 298.15 K and 0.1 MPa. The data for the neat IL is taken from literature [9].

Fig. 9. Selectivity, S, for the ternary systems n-heptane (1) þ ethanol (2) þ HIIL (3), as afunction of the ethanol mass fraction in the n-heptane-rich phase at 298.15 K and0.1 MPa. The lines are just for eyes' guidance and the data for the neat IL is taken fromliterature [9].

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e66 63

S ¼ wI1

wII1

� b2 (8)

where, b2 is the distribution coefficient of ethanol, S is the selec-tivity, w1 and w2 the mass fractions of n-heptane and ethanol,respectively and superscripts I and II indicate the n-heptane-richphase (upper phase) and HIIL-rich phase (lower phase), respec-tively. A high selectivity allows fewer stages in the process ofextraction and smaller amount of residual diluent or carrier in theextract, while a high distribution coefficient value usually leads to alower solvent flow rate, a smaller-diameter column and loweroperating costs [19].

Fig. 8 depict the distribution coefficient values obtained for theHIILs tested, while the data is listed in Table 4. In addition, thedistribution coefficient values obtained for the neat IL, [C2MIM][C2SO4], were also plotted for comparison. It can be observed thataddition of IS did not promote any major effect on the distributioncoefficient in comparison with the neat IL. In all systems, HIILs andIL, the b2 values increase with the decrease of the ethanol massfraction in the n-heptane-rich phase, showing similar values in thewhole range with the exception of HIIL33. For mass fraction valuesbelow 0.05, the b2 values of HIIL33 drop. This was not expectedsince HIIL33 presents the highest ionicity. This fact indicates thatthe distribution coefficient is not related with the extraction sol-vent ionicity. On the other hand, MD studies [24] showed that, atthis concentration (xIS ¼ 0.33), the original IL structure starts tobreak apart, and the ions are less ordered which might explain the

less solute-carrying capacity of HIIL33.The selectivity values obtained for the ternary systems con-

taining HIILs studied in this work are plotted in Fig. 9 and the datapresented in Table 4. The selectivity values obtained for the neat[C2MIM][C2SO4] were also plotted for comparison. It can beobserved that the addition of IS to the IL increased the selectivityvalues in comparison to the pure IL. The results obtained also showthat a correlation between the ionicity of the extraction solventand its selectivity can be established. HIIL33 is the extraction sol-vent that presented the highest selectivity values in accordancewith its highest ionicity. When comparing HIIL33 with the neat IL,the HIIL always presents higher selectivity values than the neat IL.Even at low ethanol concentrations, where the neat IL displays itshighest selectivity, HIIL33 has a selectivity of more than 3 timeshigher than that of the neat IL. Regarding the other two HIILs, theselectivity values obtained are also higher than those of the neat IL,with the HIIL45 presenting higher selectivity values than HIIL17.For mass fractions of ethanol lower than 0.05 it is possible toestablish a trend, where the selectivity increases with the increase

Fig. 10. Effect of the ionicity in the selectivity of the HIIL for the system n-heptane(1) þ ethanol (2) þ HIIL (3), at a three fixed ethanol mass fractions in n-heptane-richphase, at 298.15 K and 0.1 MPa. The data for the neat IL is taken from literature [9].

Fig. 12. Comparison of the selectivity, S, for the systems studied in this work and thosefound in literature [1,2,6,7,9e11,13,18] at an ethanol mass fraction between 0.02 and0.03 wt% in n-heptane-rich phase, at 298.15 K and 0.1 MPa.

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e6664

in the ionicity, following the order: HIIL33 > HIIL45 > HIIL17 > neatIL.

In order to better evaluate the effect of the addition of IS to the ILin the efficiency of the HIILs as extraction solvents, the ethanolmass fractionwas fixed and the selectivity values were compared inFig. 10. Three different mass fractions of ethanol, 0.014, 0.030 and0.090, were chosen to draw a direct comparison between the HIILcontaining systems and the system with the neat IL. As the massfraction of ethanol increases in the n-heptane-rich phase, the effectof the ionicity becomes less pronounced, as shown in Figs.10 and S7in the ESI. Nonetheless, the addition of [NH4][SCN] to the [C2MIM][C2SO4] always enhances the selectivity of the neat IL and HIIL33always displays the highest selectivity values at the 3 studiedethanol compositions, in accordance to its highest ionicity.

In Figs. 11 and 12, a comparison between the distribution coef-ficient and selectivity values of the HIILs tested in this work and theILs found in literature, for the n-heptane þ ethanol azeotrope, at afixed ethanol mass fraction in n-heptane-rich phase, is presented.

Fig. 11. Comparison of the distribution coefficient values, b2, for the systems studied inthis work and those from literature [1,2,6,7,9e11,13,18] at an ethanol mass fractionbetween 0.02 and 0.03 wt% in n-heptane-rich phase at 298.15 K and 0.1 MPa.

The comparison in the full range of ethanol mass fraction is shownin Figs. S7eS8 in the ESI. Most of the literature data on the sepa-ration of n-heptane þ ethanol azeotropic mixture using ILs asextraction solvents is based on ILs containing the bis(tri-fluoromethylsulfonyl)imide anion.

When comparing the performance of the HIILs used in this workwith the [NTf2]-based ILs found in literature, it can be seen thatHIILs greatly outperform ILs in terms of both the distribution co-efficient and the selectivity values, in the hole range of ethanolcomposition. In the case of the di-alkyl phosphate-based ILs, theHIILs tested presented slightly higher distribution coefficient valuesthan the [C1MIM][DMP] and higher than the other two ILs. How-ever, HIILs have much higher selectivity values than [C1MIM][DMP]. The IL [C4-3-C1pyr][DCA] was, so far, the neat IL that yieldedthe highest distribution coefficient values. However, in terms ofselectivity it still falls short in comparison with the alkyl sulphate-based ILs and HIILs tested in this work.

Regarding the alkyl sulphate-based ILs, in terms of the distri-bution coefficient values, HIIL17 and 45 displayed values in thesame order of magnitude as the other alkyl sulphate-based ILsfound in literature [9,10,13], whereas for the selectivity, it can beseen that the addition of inorganic salt greatly enhances it incomparison with the neat IL, and in the case of HIIL33 higherselectivity values than those of [C1MIM][C1SO4] IL can beobtained.

4. Conclusions

In this work, the effect of the ionicity of the ionic liquid wasassessed for the separation of n-heptane þ ethanol azeotropicmixtures. To increase the ionicity of the IL 1-ethyl-3-methylimidazolium ethyl sulphate, the IS ammonium thiocyanatewas added in three different concentrations, xIS ¼ 0.17, 0.33 and0.45, resulting in the formation of HIILs.

The LLE for the ternary systems n-heptane þ ethanol þ HIIL wasmeasured at 298.15 K and 0.1 MPa, and in order to analyse the HIILs'extraction capacity, the distribution coefficient and selectivityvalues were determined. The results obtained show that the HIILstested are good candidates as extraction solvents for the n-heptaneþ ethanol azeotropic mixture breaking using liquideliquid

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e66 65

extraction process. The HIILs tested in this work, proved to havehigher selectivity values than the majority of the neat ILs found inliterature, as well as distribution coefficient values higher or in thesame order of magnitude. Furthermore, by increasing the ionicity ofa given IL, through the addition of an IS, the efficiency of the IL canbe increased and a better separation for azeotropic mixtures can beobtained.

The experimental phase equilibria can be well represented byusing the NRTL model and by treating each HIIL (IL þ IS) as apseudo-component. The slopes of the tie lines agree well and thesize of the liquideliquid region is well described by the model,except for the region close to the critical point. The set of binaryinteractions parameters found in this studywould allow calculatingthe phase behaviour depending on the salt content of the system.

The knowledge gained in this study may open the door for theproduction of more promising extraction solvents for azeotropicmixtures, due to the wide range of combinations that can beestablished between ILs and ISs. Nevertheless, the addition of ISshould always prove to make a large enough boost to the ILextraction capabilities, in order to overcome the additional costsimposed by its presence.

Acknowledgements

Filipe S. Oliveira, Ana B. Pereiro and Jo~ao M. M. Araújo gratefullyacknowledge the financial support of FCT/MCTES (Portugal)through the PhD fellowship SFRH/BD/73761/2010, and the Post-Docfellowships SFRH/BPD/84433/2012 and SFRH/BPD/65981/2009,respectively. Isabel M. Marrucho gratefully acknowledges theFundaç~ao para a Ciencia e Tecnologia for a contract under FCTInvestigator 2012 Program. The authors also acknowledge theFundaç~ao para a Ciencia e a Tecnologia for the financial supportthrough the projects PTDC/EQU-FTT/116015/2009, PTDC/EQU-FTT/1686/2012 and PEST-OE/EQB/LA0004/2013. Eurico J. Cabrita for theassistance in the NMR experiments. The NMR spectrometers arepart of The National NMR Facility, supported by FCT (RECI/BBB-BQB/0230/2012).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.fluid.2016.03.004.

References

[1] N.M. Aranda, B. Gonz�alez, Cation effect of ammonium imide based ionic liq-uids in alcohols extraction from alcohol-alkane azeotropic mixtures, J. Chem.Thermodyn. 68 (2014) 32e39.

[2] F. Cai, G. Xiao, Liquid-liquid equilibria for ternary systemsethanol þ heptane þ phosphoric-based ionic liquids, Fluid Phase Equilib. 386(2015) 155e161.

[3] F. Cai, M. Zhao, Y. Wang, F. Wang, G. Xiao, Phosphoric-based ionic liquids assolvents to separate the azeotropic mixture of ethanol and hexane, J. Chem.Thermodyn. 81 (2015) 177e183.

[4] S. Corderí, B. Gonz�alez, Ethanol extraction from its azeotropic mixture withhexane employing different ionic liquids as solvents, J. Chem. Thermodyn. 55(2012) 138e143.

[5] S. Corderí, B. Gonz�alez, N. Calvar, E. G�omez, Ionic liquids as solvents toseparate the azeotropic mixture hexane/ethanol, Fluid Phase Equilib. 337(2013) 11e17.

[6] B. Gonz�alez, S. Corderí, Capacity of two 1-butyl-1-methylpyrrolidinium-basedionic liquids for the extraction of ethanol from its mixtures with heptane andhexane, Fluid Phase Equilib. 354 (2013) 89e94.

[7] B. Gonz�alez, S. Corderí, A.G. Santamaría, Application of 1-alkyl-3-methylpyridinium bis(trifluoromethylsulfonyl)imide ionic liquids for the ethanolremoval from its mixtures with alkanes, J. Chem. Thermodyn. 60 (2013) 9e14.

[8] F.S. Oliveira, A.B. Pereiro, L.P.N. Rebelo, I.M. Marrucho, Deep eutectic solventsas extraction media for azeotropic mixtures, Green Chem. 15 (2013)1326e1330.

[9] A.B. Pereiro, F.J. Deive, J.M.S.S. Esperança, A. Rodriguez, Alkylsulfate-basedionic liquids to separate azeotropic mixtures, Fluid Phase Equilib. 294 (2010)

49e53.[10] A.B. Pereiro, A. Rodriguez, Azeotrope-breaking using [bmim][meSO4] ionic

liquid in an extraction column, Sep. Purif. Technol. 62 (2008) 733e738.[11] A.B. Pereiro, A. Rodriguez, A study on the liquid-liquid equilibria of 1-alkyl-3-

methylimidazolium hexafluorophosphate with ethanol and alkanes, FluidPhase Equilib. 270 (2008) 23e29.

[12] A.B. Pereiro, A. Rodriguez, Purification of hexane with effective extractionusing ionic liquid as solvent, Green Chem. 11 (2009) 346e350.

[13] A.B. Pereiro, A. Rodriguez, Separation of ethanol-heptane azeotropic mixturesby solvent extraction with an ionic liquid, Ind. Eng. Chem. Res. 48 (2009)1579e1585.

[14] A.B. Pereiro, A. Rodríguez, Effective extraction in packed column of ethanolfrom the azeotropic mixture ethanol þ hexane with an ionic liquid as solvent,Chem. Eng. J. 153 (2009) 80e85.

[15] A.B. Pereiro, E. Tojo, A. Rodriguez, J. Canosa, J. Tojo, HmimPF6 ionic liquid thatseparates the azeotropic mixture ethanol þ heptane, Green Chem. 8 (2006)307e310.

[16] A.-L. Revelli, F. Mutelet, J.-N. Jaubert, Extraction of n-alcohols from n-heptaneusing ionic liquids, J. Chem. Eng. Data 56 (2011) 3873e3880.

[17] N.R. Rodriguez, B.S. Molina, M.C. Kroon, Aliphatic þ ethanol separation vialiquideliquid extraction using low transition temperature mixtures asextracting agents, Fluid Phase Equilib. 394 (2015) 71e82.

[18] R.G. Seoane, E.J. Gonz�alez, B. Gonz�alez, 1-Alkyl-3-methylimidazolium bis(tri-fluoromethylsulfonyl)imide ionic liquids as solvents in the separation ofazeotropic mixtures, J. Chem. Thermodyn. 53 (2012) 152e157.

[19] A.B. Pereiro, J.M.M. Araújo, J.M.S.S. Esperança, I.M. Marrucho, L.P.N. Rebelo,Ionic liquids in separations of azeotropic systems - a review, J. Chem. Ther-modyn. 46 (2012) 2e28.

[20] Z. Lei, X. Xi, C. Dai, J. Zhu, B. Chen, Extractive distillation with the mixture ofionic liquid and solid inorganic salt as entrainers, AlChE J. 60 (2014)2994e3004.

[21] A.B. Pereiro, J.M.M. Araújo, F.S. Oliveira, J.M.S.S. Esperança, J.N.C. Lopes,I.M. Marrucho, L.P.N. Rebelo, Solubility of inorganic salts in pure ionic liquids,J. Chem. Thermodyn. 55 (2012) 29e36.

[22] F.S. Oliveira, A.B. Pereiro, J.M.M. Araújo, C.E.S. Bernardes, J.N. Canongia Lopes,S. Todorovic, G. Feio, P.L. Almeida, L.P.N. Rebelo, I.M. Marrucho, High ionicityionic liquids (HIILs): comparing the effect of -ethylsulfonate and -ethylsulfateanions, Phys. Chem. Chem. Phys. 15 (2013) 18138e18147.

[23] F.S. Oliveira, L.P.N. Rebelo, I.M. Marrucho, Influence of different inorganic saltson the ionicity and thermophysical properties of 1-ethyl-3-methylimidazolium acetate ionic liquid, J. Chem. Eng. Data 60 (2015)781e789.

[24] A.B. Pereiro, J.M.M. Araújo, F.S. Oliveira, C.E.S. Bernardes, J.M.S.S. Esperança,J.N.C. Lopes, I.M. Marrucho, L.P.N. Rebelo, Inorganic salts in purely ionic liquidmedia: the development of high ionicity ionic liquids (HIILs), Chem. Commun.48 (2012) 3656e3658.

[25] W. Xu, E.I. Cooper, C.A. Angell, Ionic liquids: Ion mobilities, glass tempera-tures, and fragilities, J. Phys. Chem. B 107 (2003) 6170e6178.

[26] H. Tokuda, S. Tsuzuki, M. Susan, K. Hayamizu, M. Watanabe, How ionic areroom-temperature ionic liquids? An indicator of the physicochemical prop-erties, J. Phys. Chem. B 110 (2006) 19593e19600.

[27] K. Ueno, H. Tokuda, M. Watanabe, Ionicity in ionic liquids: correlation withionic structure and physicochemical properties, Phys. Chem. Chem. Phys. 12(2010) 1649e1658.

[28] J. García, A. Fern�andez, J.S. Torrecilla, M. Oliet, F. Rodríguez, Liquideliquidequilibria for {hexane þ benzene þ 1-ethyl-3-methylimidazolium ethyl-sulfate} at (298.2, 313.2 and 328.2) K, Fluid Phase Equilib. 282 (2009)117e120.

[29] E.J. Gonza

lez, N. Calvar, E. Go

mez, A.N. Domi

nguez, Separation of benzenefrom linear alkanes (C6�C9) using 1-ethyl-3-methylimidazolium ethyl sulfateat T ¼ 298.15 K, J. Chem. Eng. Data 55 (2010) 3422e3427.

[30] D. Naydenov, H.-J. Bart, Ternary liquid�liquid equilibria for six systems con-taining ethyl acetate þ ethanol or acetic acid þ an imidazolium-based ionicliquid with a hydrogen sulfate anion at 313.2 K, J. Chem. Eng. Data 52 (2007)2375e2381.

[31] A.B. Pereiro, A. Rodriguez, Ternary (liquid þ liquid) equilibria of the azeotrope(ethyl acetate þ 2-propanol) with different ionic liquids at T¼298.15 K,J. Chem. Thermodyn. 39 (2007) 1608e1613.

[32] A. Arce, H. Rodriguez, A. Soto, Use of a green and cheap ionic liquid to purifygasoline octane boosters, Green Chem. 9 (2007) 247e253.

[33] J.D. Holbrey, W.M. Reichert, R.P. Swatloski, G.A. Broker, W.R. Pitner,K.R. Seddon, R.D. Rogers, Efficient, halide free synthesis of new, low cost ionicliquids: 1,3-dialkylimidazolium salts containing methyl- and ethyl-sulfateanions, Green Chem. 4 (2002) 407e413.

[34] M. Larriba, P. Navarro, J. Garcia, F. Rodriguez, Liquid-liquid extraction oftoluene from n-alkanes using {[4empy][Tf2N] þ [emim][DCA]} ionic liquidmixtures, J. Chem. Eng. Data 59 (2014) 1692e1699.

[35] U.K. Ravilla, T. Banerjee, Liquid liquid equilibria of imidazolium based ionicliquid plus pyridine plus hydrocarbon at 298.15K: experiments and correla-tions, Fluid Phase Equilib. 324 (2012) 17e27.

[36] H. Renon, J.M. Prausnitz, Local compositions in thermodynamic excess func-tions for liquid mixtures, AlChE J. 14 (1968) 135e144.

[37] L.D. Simoni, Y. Lin, J.F. Brennecke, M.A. Stadtherr, Modeling liquid-liquidequilibrium of ionic liquid systems with NRTL, electrolyte-NRTL, and UNI-QUAC, Ind. Eng. Chem. Res. 47 (2008) 256e272.

F.S. Oliveira et al. / Fluid Phase Equilibria 419 (2016) 57e6666

[38] D.S. Abrams, J.M. Prausnitz, Statistical thermodynamics of liquid mixtures: anew expression for the excess Gibbs energy of partly or completely misciblesystems, AlChE J. 21 (1975) 116e128.

[39] T. Banerjee, M.K. Singh, R.K. Sahoo, A. Khanna, Volume, surface and UNIQUACinteraction parameters for imidazolium based ionic liquids via polarizable

continuum model, Fluid Phase Equilib. 234 (2005) 64e76.[40] R.S. Santiago, G.R. Santos, M. Aznar, UNIQUAC correlation of liquid-liquid

equilibrium in systems involving ionic liquids: the DFT-PCM approach. PartII, Fluid Phase Equilib. 293 (2010) 66e72.