Extraction of aromatic hydrocarbons from pyrolysis gasoline using tetrathiocyanatocobaltate-based ionic liquids: Experimental study and simulation

Fuel Processing Technology 159 (2017) 96–110

Contents lists available at ScienceDirect

Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc

Extraction of aromatic hydrocarbons from pyrolysis gasoline using tetrathiocyanatocobaltate-based ionic liquids: Experimental study and simulation Marcos Larriba a, Pablo Navarro a, Noemí Delgado-Mellado a, Victor Stanisci b, Julián García a,⁎, Francisco Rodríguez a a b

Department of Chemical Engineering, Complutense University of Madrid, E–28040 Madrid, Spain Centro de Ciências Naturais e Humanas, Universidade Federal do ABC, Santo André, SP, Brazil

a r t i c l e

i n f o

Article history: Received 7 December 2016 Received in revised form 18 January 2017 Accepted 18 January 2017 Available online xxxx Keywords: Aromatic/aliphatic separation Ionic liquids Liquid–liquid extraction Vapor-liquid separation Thermophysical characterization Process simulation

a b s t r a c t The pyrolysis gasoline is one of the main sources of aromatic hydrocarbons as a result of their high content in these compounds. Organic solvents such as sulfolane are currently employed in the extraction of aromatic but the ionic liquids (ILs) have been recently proposed as potential replacement. In this work, we have studied the use of the bis(1-ethyl-3-methylimidazolium) tetrathiocyanatocobaltate ([emim]2[Co(SCN)4]) and bis(1-butyl3-methylimidazolium) tetrathiocyanatocobaltate ([bmim]2[Co(SCN)4]) ILs in the extraction of aromatic hydrocarbons from pyrolysis gasoline. The extractive properties of both tetrathiocyanatocobaltate-based ILs were compared to those of other promising ILs and sulfolane, showing the highest values. To perform the simulation of the whole process, we have experimentally studied the liquid-liquid extraction of aromatics from pyrolysis gasoline and the recovery of the extracted hydrocarbons from the ILs. In addition, a thermophysical characterization of the ionic solvents was performed measuring their densities, viscosities, thermal stabilities, maximum operation temperatures, and specific heats. Employing the experimental data, the extractor was simulated using the Kremser equation whereas the recovery section formed by flash distillation units was simulated using a new algorithm specifically design to the case of a high concentration of non-volatile compounds. © 2017 Elsevier B.V. All rights reserved.

1. Introduction The ionic liquids (ILs) are solvents entirely formed by ions, being the nonvolatile nature their more remarkable property [1]. In addition, their thermophysical and extractive properties can be modified by changing the structure of the ions forming the ILs. A wide variety of works on the potential use of ILs in the dearomatization, denitrogenation, or desulfurization of fuels have been recently published [2–11]. The extraction of aromatic hydrocarbons is currently performed by liquid-liquid extraction processes, being the Sulfolane process the most employed at industrial scale [12]. This process is formed by a liquid-liquid extraction column, an extractive stripper, and a distillation column to recover the extracted hydrocarbons. As result of the high boiling point of the sulfolane, the operative costs of the process are high and this drawback could be overcome using ILs as solvents due to their non-volatility [13]. The recovery section of the extraction process using ILs could be formed by simple flash distillations obtaining a liquid stream of the regenerated IL and vapor streams composed of ⁎ Corresponding author. E-mail address: [email protected] (J. García).

http://dx.doi.org/10.1016/j.fuproc.2017.01.027 0378-3820/© 2017 Elsevier B.V. All rights reserved.

the selectively recovered hydrocarbons. In the case of sulfolane, flash distillations are not used because of the volatility of the organic solvent. However, the majority of papers on the separation of aromatic hydrocarbons from alkanes using ILs have been only focused on the liquid-liquid extraction step without studying the recovery of the extracted hydrocarbons from the ILs [14]. Another advantage of ILs with respect to sulfolane is the negligible solubility of ILs in the aliphatics [14], whereas the sulfolane is partially solved in the raffinate and a washing column to recover the solvent from this stream is required to ensure the economic viability of the process. Nevertheless, the current price of the ILs and the higher viscosity of the ILs with respect of sulfolane have so far limited their application as replacement of sulfolane. In this work, we have studied the applicability of the [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs in the extraction of benzene, toluene, and xylene (BTX) from pyrolysis gasoline obtaining experimentally the required data to simulate the extraction and recovery sections. To evaluate the performance of the ILs in the extraction of aromatic hydrocarbons, the experimental extractive properties were compared to those of sulfolane and other promising ILs. To select the ILs, we have considered our previous results obtained in the extraction of toluene from n-heptane using the [emim][SCN] and

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

the [bmim][SCN], being the [emim][SCN] the IL with the highest values of toluene/n-heptane selectivity among the ILs studied so far [14,15]. The presence of metal transition salts dissolved in ILs has been revealed as a useful tool to improve the separation of aromatics and olefins from alkanes [16–18]. Because of this, we also previously tested the performance of the [emim][SCN] with dissolved Co(SCN)2, observing an slight increase in the values of toluene distribution ratios [19]. Against this background, we decided to employ the [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4], since these ILs have a cobalt atom incorporated in the structure of the anion and an improvement of the extractive properties with respect to the [emim][SCN] and the [bmim][SCN] was expected. 2. Materials and methods 2.1. Materials [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were purchased from Iolitec GmbH with a purity higher than 99% in mass basis. Water content in ILs was determined by the supplier using Karl-Fischer titration, being 0.04% in the [emim]2[Co(SCN)4] and 0.21% for the [bmim]2[Co(SCN)4]. Hydrocarbons were supplied by Sigma-Aldrich with the specifications listed in Table 1. All chemicals were used as received without further purification. ILs were stored in a desiccator and handled in a glove box under dry nitrogen to avoid water hydration. 2.2. Thermophysical characterization of the ILs Densities and viscosities of pure ILs were measured at temperatures between (293.15 and 353.15) ± 0.01 K. An Anton Paar DMA-5000 oscillating U-tube was used in the determination of the density, whereas an Anton Paar Automated Micro Viscometer (AMVn) was used to measure the viscosity. The reliability of these methods has been checked with literature data. Average deviation between literature and experimental densities was 0.57% for [emim]2[Co(SCN)4] and 0.03% for [bmim]2[Co(SCN)4] [20], and between literature and experimental viscosities was 6.95% for [emim]2[Co(SCN)4] and 9.83% for [bmim]2[Co(SCN)4] [21]. Higher deviations of experimental viscosities are due to the strong influence of measurement techniques and impurities in ILs on the value of this property [22]. A thermogravimetrical analysis was employed to determine the thermal stability of the tetrathiocyanatocobaltate-based ILs, using a Mettler Toledo TGA/DSC 1 at heating rates of 5, 10, and 20 K·min−1. A complete description of this method can be found in our previous publication [23]. Dynamic TGA essays were performed to calculate the maximum operation temperatures (MOTs) for the ILs fixing the exposure time to 8000 h and following the method proposed by Seeberger et al.

Table 1 Chemicals used in this paper: suppliers, purities, and analysis methods. Chemical

Supplier

Mass fraction purity

[emim]2[Co(SCN)4]a

Iolitec GmbH

0.99

[bmim]2[Co(SCN)4]d Iolitec GmbH

0.99

n-Hexane n-Heptane n-Octane Benzene Toluene p-Xylene

0.99 0.997 0.99 0.995 0.995 0.99

a b c d e

Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich

Analysis method NMRb and ICc NMRb and ICc GCe GCe GCe GCe GCe GCe

Abbreviation

ILs hexa hepta octa benz tol p-xyl

[emim]2[Co(SCN)4] = bis(1-ethyl-3-methylimidazolium) tetrathiocyanatocobaltate. Nuclear Magnetic Resonance. Ion Chromatography. [bmim]2[Co(SCN)4] = bis(1-butyl-3-methylimidazolium) tetrathiocyanatocobaltate. Gas Chromatography.

97

[24]. These values of MOTs are important to guarantee the thermal stability of the ILs in the extraction process. Finally, the specific heats for both tetrathiocyanatocobaltate-based ILs were determined from (283 to 393) K following the sapphire method (ASTM E 1269-01) and using differential scanning calorimetry (DSC) in a Mettler Toledo DSC821e. Prior to the measurements, an in situ drying was included to minimize the water content of the ILs. Further experimental details were included in our previous work [23]. 2.3. Pyrolysis gasoline models According to Franck and Stadelhofer [25] a typical pyrolysis gasoline obtained by severe cracking has the following composition: benzene (33.8 wt.%), toluene (19.4 wt.%), ethylbenzene, xylenes and styrene (13.0 wt.%), and non-aromatics (33.9 wt.%). To simplify the experimental model, ethylbenzene, xylenes and styrene were represented by p-xylene. In addition, non-aromatic content in the model was formed by 11.3 wt.% of n-hexane, 11.3 wt.% of n-heptane, and 11.3 wt.% of n-octane. Table S1 in the Supplementary Data shows the composition in mass basis of the pyrolysis gasoline employed. In the preparation of this gasoline a Mettler Toledo XS 205 balance with a precision of ±1·10−5 g was employed. 2.4. Liquid-liquid extraction. Experimental procedure and analysis First, the performance of the tetrathiocyanatocobaltate-based ILs in the liquid-liquid extraction of aromatics was studied in the separation of toluene from n-heptane, since the wide majority of papers on the extraction of aromatic hydrocarbons using ILs have studied the separation of these two hydrocarbons [26]. From the experimental results, toluene distribution ratios and toluene/n-heptane selectivities using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were calculated and compared with the extractive properties of other promising ILs and those of sulfolane. Subsequently, the extraction of benzene, toluene, and p-xylene from the pyrolysis gasoline using both tetrathiocyanatocobaltate-based ILs was also performed at 313.2 K and a solvent to feed (S/F) ratios in mass of 1.0, 3.0, 5.0, and 7.0. Hydrocarbons and ILs were gravimetrically added to 8 mL vials using a Mettler Toledo XS 205 balance with a precision of ±1·10−5 g. The liquid-liquid equilibrium between both phases were reached inside a Labnet Vortemp 1550 shaking incubator for 5 h at (313.2 ± 0.5) K. To guarantee the correct separation of these phases, vials were transferred to a Labnet Accublock dry bath for 12 h at (313.2 ± 0.3) K. Three raffinate samples were analyzed using an Agilent 7890A gas chromatograph (GC) equipped with a flame ionization detector (FID). A complete description of this analytical method can be found in our previous publication [27]. Same samples from the raffinate phases were also analyzed using a Bruker Avance 500 MHz NMR spectrometer to ensure that ILs were not soluble in these phases and that their presence was undetectable. The analysis of the composition of the extract phases were done with a multiple headspace extraction (MHE) method. Triplicate samples of 100 μL were analyzed employing an Agilent 7890A GC coupled with a Headspace Sampler Agilent 7697A. A more detailed description of this MHE method can be found elsewhere [27]. 2.5. Simulation of the liquid-liquid extraction column by the Kremser method The liquid-liquid extraction column of BTX from the pyrolysis gasoline was simulated using the Kremser method employing the experimental extractive properties of both tetrathiocyanatocobaltate-based ILs. This method was successfully employed in our previous publication to simulate the extraction of BTX from pyrolysis gasoline using a binary IL mixture [11]. Using the distribution ratios (Di) obtained from the experimental results at the optimal value of S/F ratio, the extraction factor

98

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

(Ei) and the reciprocal of E (Ui) of the Kremser method were calculated as follows: Ei ¼ Di Ui ¼

V L

ð1Þ

1 L ¼ Ei Di V

ð2Þ

where L is the mass flow of the raffinate stream and V is the mass flow of the extract. Flows of raffinate and extract streams obtained in the extractor were calculated using an iterative method implemented in Microsoft Excel as a function of the number of equilibrium stages in the extractor. As basis of calculation, a flow of 1000 t/h of the pyrolysis gasoline model was introduced to the extraction column with the composition showed in Table S1 in the Supplementary Data. Considering the optimal value of S/F ratio selected from the experimental results, the flow of the solvent fed to the extractor was calculated. The complete description of the Kremser method used in the simulation of the extractor can be found elsewhere [11,28]. Once the extractor was simulated, the optimal value of equilibrium stages was selected and the flow and composition of the extract stream obtained were used in the subsequent study on the hydrocarbon recovery section.

Fig. 1. Algorithm to simulate adiabatic flash distillation units with a high concentration of non-volatile substances.

2.6. Vapor-liquid separation procedure and analysis To study the vapor-liquid selective recovery of the hydrocarbon extracted from the ILs, an Agilent headspace (HS) 7697A injector coupled to an Agilent gas chromatograph 7890A was used. Mixtures of aromatic and aliphatic hydrocarbons and the ILs were prepared using a Mettler Toledo XS205 with the composition of the extract stream obtained in the simulation of the extractor. The closed vials were introduced into the oven of the HS injector at different temperatures, being analyzed the vapor phase by GC. All temperatures used in the experimental vapor-liquid separation of the hydrocarbons and the ILs were lower than the maximum operation temperature previously estimated for the ILs by TGA. To calculate the composition of the liquid phase the next equation was employed: xi ¼

zi  F−ðP i  V G =R  T Þ 7 ∑i¼1 ðzi  F−ðP i  V G =R 

T ÞÞ

ð3Þ

where xi is the mole fraction of each compound in the liquid phase, zi is the mole fraction in the feed, F indicates the molar amount of the feed, VG is the vapor volume of the vial, and Pi is the partial pressure of each hydrocarbon calculated as: Pi ¼

Ai A0i

P 0i

By using this algorithm, we have calculated different temperature (TV) – pressure (PV) scenarios in the flash to select the optimal operational conditions. As TV are the experimental temperatures, the first selected PV is the experimentally obtained value. Accordingly, the individual molar flows in the vapor phase are known to the first simulation, which is done for checking purpose. Thus, the total vapor flow and the mole fractions in the vapor stream are calculated. Using the experimental K-values, it is possible to calculate the composition of the liquid phase. Therefore, the objective function was defined as the difference between the mole fractions in the liquid phase calculated from the chemical equilibrium (C) and those determined by mass balance (D). The second simulation at the next PV will be done starting the algorithm with the previous individual molar flows, which are closely near to the solution for the new simulation. The same procedure is followed to complete all TV-PV scenarios. Finally, from the results obtained in the simulation, an enthalpy balance was made and the feed temperature was calculated checking that this temperature is lower than the MOT of the ILs. We have selected adiabatic flash distillation units as their costs are lower in comparison with isothermal units [28]. The enthalpy balance was solved using the vaporization heats and specific heats obtained from literature for the hydrocarbons [30] and the specific heats of the ILs reported in this work.

ð4Þ 3. Results and discussion

where Ai are the peak areas of the hydrocarbons in the presence of the IL obtained by GC, whereas A0i and P0i are the peak areas of the pure hydrocarbons and the vapor pressures of the hydrocarbons under the same conditions, respectively. A complete description of this method can be found in our previous publication [29]. 2.7. Adiabatic flash distillation simulation Once the vapor-liquid separation of the hydrocarbons and the ILs was experimentally studied, the recovery section formed by flash distillation units was simulated and designed. To simulate the flash distillation units a new algorithm was developed for systems with a high concentration of non-volatile compounds. The scheme of the algorithm used in the simulation of the flash distillations is plotted in Fig. 1, whereas Table S2 in the Supplementary Data shows the independent equations used to simulate the flash distillations.

3.1. Thermophysical characterization of the ionic liquids Densities, dynamic viscosities, and thermal stabilities of [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were measured in order to evaluate their potential use in aromatic extraction. Experimental values of densities and viscosities between 293.2 K and 353.2 K are shown in Table S3 in the Supplementary Data. In relation to density, [emim]2[Co(SCN)4] (1.2657 g·cm− 3 at 313.2 K) showed a density slightly higher than that of [bmim]2[Co(SCN)4] (1.2112 g·cm− 3 at 313.2 K). Compared to published density of sulfolane (1.2532 g·cm−3 at 313.2 K) [27], both ILs showed densities similar to sulfolane and considerably higher than those of hydrocarbons forming the pyrolysis gasoline; therefore, a fast separation of raffinate and extract phases could be achieved using the ILs as aromatic extraction solvents.

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

99

Table 2 Activation energies (EA), frequency factors (k0), and predicted maximum operation temperatures (MOT) for tetracyanocobaltate-based ILs. ILs

EA/J·mol−1

k0/s−1

MOT/K

[emim]2[Co(SCN)4] [bmim]2[Co(SCN)4]

1.40·105 1.49·105

1.24·109 8.64·109

393.0 400.7

Regarding viscosities, these tetrathiocyanatocobaltate-based ILs presented high values compared to sulfolane. The high viscosity of the ILs studied in this work could difficult the mass transfer of extracted hydrocarbons and could increase the mixing and pumping costs. Even though the difference in this physical property with sulfolane is noticeable, values of viscosity at 313.2 K of [emim]2[Co(SCN)4] (59.6 mPa·s) and [bmim]2[Co(SCN)4] (136.4 mPa·s) are similar to those of other promising ILs in this area: [bmim][MeSO4] (84.7 mPa·s) [31], and [bmim][PF6] (109 mPa·s) [32]. According to previous work published by Galán Sánchez et al., an increase of the length of the alkyl chain in the imidazolium cation causes an increase in the viscosity [33]. In tetrathiocyanatocobaltate-based ILs this effect was higher as each IL presents two cations. Different parameters were studied to determine the thermal stability of the tetrathiocyanatocobaltate-based ILs: onset temperature (Tonset), temperatures at which 10% (T10%) and 50% (T50%) of IL mass is lost and the percentage of ashes that remains at 1123 K. These values were obtained from the TGA curves at three different heating rates, namely, 5, 10, and 20 K·min−1. The thermal stability parameters regarding these experiments are listed in Table S4 in the Supplementary Data. As can be seen, the thermal stability of both ILs is quite similar as Tonset, the commonly factor to compare thermal stability, has near values at 10 K·min−1, Tonset for [emim]2[Co(SCN)4] was 587.6 K whereas the Tonset for [bmim]2[Co(SCN)4] was 588.8 K. The nature of the cation has a slight influence in Tonset and the little difference in the decomposition temperature is due to the increase in the alkyl chain length of the cation, since a longer alkyl chain in the cation causes higher Tonset, as published Huddleston et al. [34] and Crosthwaite et al. [35] for imidazolium-based ILs. Both ILs showed a Tonset at 10 K·min−1 higher than other cyano-based ILs proved in aromatic extraction such as the [amim][DCA] (549.0 K), [bzmim][DCA] (551.4 K), or [emim][SCN] (538.6 K) [23,36], which confirms that tetrathiocyanatocobaltate anion is adequate to form an IL with high thermal stability. In fact, comparing Tonset for [emim]2[Co(SCN)4] and [emim][SCN], the high effect of the cobalt atom on the thermal stability for thiocyanate anions is clearly seen. As studied before in our previous publications, Tonset overestimates the temperature at which an IL can operate in an industrial process without thermal decomposition [23], therefore, it is necessary to calculate the maximum operation temperature to give a more realistic limit of application in long-term scenarios. Using the dynamic TGA data obtained at the heating rate of 5 K·min− 1, the activation energies and the frequency factors were determined. Table 2 shows these parameters along with MOT values calculated for both ILs as explained in our previous work [23]. By comparing experimental Tonset and predicted MOT, the latter were approximately 180 K lower than the common decomposition parameter calculated directly from dynamic TGA curves. Therefore, the limit of application for [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs are 393 K and 401 K, respectively. Considering these values, the temperatures used in the extraction of BTX from pyrolysis gasoline and in the selective separation of the extracted hydrocarbons from the ILs were always lower than their values of MOTs to guarantee the thermal stability of ILs in the process.

Fig. 2. Liquid-liquid equilibria for the ternary systems {n-heptane (1) + toluene (2) + IL (3)} at T = 313.2 K and P = 0.1 MPa. Solid lines and full points represent experimental tie-line whereas dashed lines and empty squares are the regressed data using the NRTL model.

Finally, the specific heats of both tetrathiocyanatocobaltate-based ILs have been determined from (283.2 to 393.2) K taking into account that at higher temperatures it is not possible to operate with these ILs due to their just commented values of MOT. Experimental results were listed in Table S5 in the Supplementary Data, whereas the parameters of the adjustments of specific heats to a second order polynomial expansion model as a function of temperature were included in Table 3 along with the goodness of these fittings. There is only one data collection available in literature of specific heats for [bmim]2[Co(SCN)4] and to the best of our knowledge there is no data in the literature for [emim]2[Co(SCN)4]. Nóvoa-López et al. (2016) reported specific heats for [bmim]2[Co(SCN)4] from (283.2 to 333.2) K with a good agreement with our values as is shown by the absolute average deviation (AAD) of 2.6% [37]. 3.2. Extractive properties of [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] in the extraction of toluene from n-heptane To make a comparative analysis between the extractive properties of both tetrathiocyanatocobaltate-based ILs in the extraction of aromatic hydrocarbons from alkanes and those of other promising ILs and

Table 3 Adjustments of specific heats for tetrathiocyanatocobaltate-based ILs. IL [emim]2[Co(SCN)4] [bmim]2[Co(SCN)4]

a/J·(g·K)−1 0.110 0.085

b/J·g−1·K−2 −3

6.044·10 6.331·10−3

c/J·g−1·K−3 −6

−2.022·10 −2.456·10−6

R2

Trange/K

0.9998 0.9998

283.2–393.2 283.2–393.2

100

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

sulfolane, the liquid-liquid extraction of toluene from n-heptane was studied. Experimental compositions of raffinate and extract phases obtained in the extraction of toluene from n-heptane over the whole range of compositions at 313.2 K using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs as solvents are listed in Table S6 in the Supplementary Data. The reliability of the obtained data was studied employing the Othmer-Tobias correlation [38]:

ln

1−wIIIL wIIIL

! ¼ a þ b ln

1−wIhepta wIhepta

! ð5Þ

where wIIIL indicates the IL mass fraction in the extract phase, wIhepta denotes the mass fraction of n-heptane in the raffinate phase, whereas a and b are the parameters of the Othmer-Tobias correlation. As can be observed in the Table S7 in the Supplementary Data, regression coefficients (R2) higher than 0.99 were obtained for the fitting of both ternary systems to the Othmer-Tobias correlation; therefore, the reliability of the experimental data was checked.

The experimental data obtained in the extraction of toluene from using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were plotted as ternary diagrams in Fig. 2. As observed, the raffinate phase had a negligible IL content, whereas the solubility of n-heptane in the ILs was substantially lower than the solubility of toluene. In the same figure, regressed tie-lines using the Non-Random Two Liquids (NRTL) model are depicted together with the experimental tie-lines. As can be seen, the experimental and regressed tie-lines were almost coincident. Hence, the experimental data obtained in the extraction of toluene from n-heptane using both tetrathiocyanatocobaltate-based ILs were successfully fitted to the NRTL model. Parameters and deviations of the fitting to the NRTL model are listed in the Table S8 in the Supplementary Data. From the experimental compositions of raffinate and extract phases, hydrocarbons distribution ratios (Di) and toluene/n-heptane selectivities (αtol,hepta) using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were calculated using the following expressions:

Di

  xII mol ¼ iI mol xi

ð6Þ

Fig. 3. Mass-based toluene distribution ratios (a) and toluene/n-heptane selectivities (b) for the ternary systems {n-heptane (1) + toluene (2) + solvent (3)} at T = 313.2 K against toluene mole fraction in the raffinate phase: ▲, [emim]2[Co(SCN)4]; ●, [bmim]2[Co(SCN)4]; *, sulfolane [39]. Dashed lines are second order polynomial correlations of experimental properties.

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

Di

  wII g ¼ iI g wi

Stol;hepta ¼

Dtol Dhepta

ð7Þ

ð8Þ

where xi and wi are the mole and mass fraction of each hydrocarbon, superscript I indicates raffinate phase, and superscript II refers to the extract phase. Calculated values for the extractive properties of the ILs can be found in Table S6 in the Supplementary Data along with the experimental compositions. To evaluate the performance of the tetrathiocyanatocobaltate-based ILs in the extraction of toluene from n-heptane, mass-based toluene distribution ratios and toluene/n-heptane selectivities of the ILs were plotted in Fig. 3 together with the literature extractive properties of sulfolane under the same conditions [39]. We have decided to compare the toluene distribution ratios of the solvent in mass basis, since Canales and Brennecke recommended to use the mass fraction basis to compare adequately the extractive properties of ILs in the extraction of aromatics [14]. As observed, the toluene distribution ratios of the ILs were very similar to the sulfolane values and they were even slightly higher than the extractive property of the conventional solvent at low aromatic contents. This fact suggests that the replacement of sulfolane by the ILs studied in this work could be performed using similar values of S/F ratio in mass in the extractor achieving comparable values of extraction yield of aromatics. Regarding to the toluene/n-heptane selectivity, the ILs have exhibited values of Stol,hepta approximately double than the sulfolane values. Therefore, the extracted toluene by the ILs had a considerably higher purity than that extracted by the sulfolane. Although a wide number of ILs has been proved in the extraction of toluene from n-heptane, the ILs tested until the date have exhibited mass-based toluene distribution ratios lower than the sulfolane [14]. In Fig. 4, the extractive properties of the [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs determined in this work have been plotted together with literature values of other promising ILs and sulfolane [15, 27,39–42]. We have represented the extractive properties of the solvents in the extraction of a 10 wt.% of toluene from its mixture with nheptane, since this composition has been the most employed to compare the extractive properties of ILs in the extraction of aromatics [26]. As can be observed, the [bmim]2[Co(SCN)4] is the IL that presented simultaneously mass-based toluene distribution ratios and toluene/n-

101

heptane selectivities considerably higher than the sulfolane. In addition, the [emim]2[Co(SCN)4] also exhibited extractive properties considerably better than those of the others ILs. It is important to highlight the significant improvement observed in Fig. 4 comparing the extractive properties of the [emim][SCN] and [bmim][SCN] with respect to those of [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4]. Therefore, the introduction of the atom of cobalt in the anion has increased the solubility of aromatics in the ILs, increasing considerably the value of mass-based toluene distribution ratios. 3.3. Liquid-liquid extraction of aromatics from aliphatics in pyrolysis gasoline Once the extractive properties of the tetrathiocyanatocobaltatebased ILs were determined in the extraction of toluene from n-heptane and the potential of these ILs was confirmed by comparing their performance with other promising ILs, the extraction of BTX from the pyrolysis gasoline model using both ILs was studied. The liquid-liquid extraction experiments were performed at 313.2 K using solvent to feed ratios in mass of 1.0, 3.0, 5.0, and 7.0 to optimize this variable. The experimental compositions in the extraction of BTX from the pyrolysis gasoline as a function of the S/F ratio can be found in Table S9 in the Supplementary Data. From the experimental compositions and performing mass balances in the vials, values of yields of extraction of aromatics (Yldarom) and relative purities of extracted aromatic hydrocarbons in the extract phase (Puarom) were calculated as follows: Yldarom ð%Þ ¼ 100

Puarom ð%Þ ¼ 100 

extract mextract þ mextract benz þ mtol p−xyl

mfeed þ mfeed þ mfeed benz tol p−xyl

ð9Þ

extract wextract þ wextract benz þ wtol p−xyl    extract þ wextract þ wextract þ wextract þ wextract wextract þ w octa hexa hepta benz tol p−xyl

ð10Þ where mi is the mass of the aromatic hydrocarbon in the extract phase or in the feed whereas wi is the hydrocarbon mass fraction in the extract phase. In Fig. 5, values of Yldarom and Puarom as a function of the S/F ratio are plotted, whereas the values of these extractive properties are also listed in Table S10 in the Supplementary Data. As seen in Fig. 5, the [emim]2[Co(SCN)4] showed slightly higher values of relative purity of

Fig. 4. Mass-based toluene distribution ratios and toluene/n-heptane selectivities in the extraction of toluene from a mixture {n-heptane (90% wt.) + toluene (10% wt.)} at T = 313.2 K: [emim]2[Co(SCN)4] (this work); [bmim]2[Co(SCN)4] (this work); [emim][SCN] [15]; [bmim][SCN] [15]; [emim][DCA] [27]; [emim][TCM] [27]; [emim][Tf2N] [40]; [emim]][CH3SO3] [26]; [4bmpy][BF4] [26]; [4empy][Tf2N] [41]; {[4empy][Tf2N] (0.3) + [emim][DCA] (0.7)} [41]; [bmim][TCM] [42]; [4bmpy][TCM] [42]. White point is the toluene distribution ratio and toluene/ n-heptane selectivity for the sulfolane under the same conditions to be employed as benchmark [39].

102

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

Fig. 5. Extraction yield of aromatic hydrocarbons from the pyrolysis gasoline (■) and relative purity of the extracted aromatics (○) as a function of the S/F ratio in mass using a) [emim]2[Co(SCN)4] and b) [bmim]2[Co(SCN)4] at 313.2 K. Solid lines are to guide the eye.

the extracted hydrocarbons and somewhat lower yield of extraction of aromatics than those of [bmim]2[Co(SCN)4] at a constant value of S/F. Both results are in agreement with the extractive properties of the ILs determined in the extraction of toluene from n-heptane, since the values of toluene/n-heptane selectivities for the [emim]2[Co(SCN)4] were higher whereas the toluene distribution ratios were lower than those of the [bmim]2[Co(SCN)4]. As can be also observed in Fig. 5, an increase in S/F ratio caused an increase in Yldarom and a decrease in the values of Puarom using both ILs. However, the values of Yldarom at S/F ratios of 5.0 and 7.0 were similar, especially using the [emim]2[Co(SCN)4]. Therefore, to maximize the values of Yldarom obtaining high relative purities of the extracted hydrocarbons and to moderate the flow needed of IL, an S/F ratio of 5.0 was selected as the optimal value. 3.4. Simulation of the countercurrent extraction column by the Kremser equation As explained in the experimental section, the simulation of the extractors was performed using the Kremser equation and the experimental data obtained at the optimal value of S/F ratio. The aim of the

simulation was to determine the number of equilibrium stages (Ns) that provides the same or higher value of yields of extraction of BTX than those using the Sulfolane process: extraction yield of benzene of 99.9%, 99.0% for toluene, and 97.0% for xylenes [12]. In Fig. 6, extraction yields of BTX using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] as solvents as a function of the number of equilibrium stages in the countercurrent extraction column are depicted. As seen, the extraction yields for the [emim]2[Co(SCN)4] were higher than the sulfolane values with N 8 equilibrium stages, whereas for the [bmim]2[Co(SCN)4] 6 equilibrium stages were needed. However, we decided to select 14 equilibrium stages to obtain extraction yields of benzene and toluene from the pyrolysis gasoline model using both ILs higher than 99.9% and an extraction yield of p-xylene higher than 99.8%. Flows and compositions of the streams obtained in the simulation of the extraction column with a S/F ratio of 5.0 and a Ns of 14 are listed in Table 4. Therefore, according to the results obtained in the simulation, the hypothetical process using the tetrathiocyanatocobaltate-based ILs could achieve extraction yields of BTX higher than those of the Sulfolane process with 14 equilibrium stages in the extractor. In addition, the ILs studied in this paper have exhibited extraction yields of aromatics substantially higher than the {[4empy][Tf2N] + [emim][DCA]} mixture

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

103

Fig. 6. Extraction yield of aromatic hydrocarbons from the pyrolysis gasoline: ■, Benzene; ○, Toluene; ▲, p-Xylene as a function of the number of equilibrium stages in the countercurrent extraction column using a) [emim]2[Co(SCN)4] and b) [bmim]2[Co(SCN)4] at 313.2 K and a solvent to feed ratio of 5.0. Dashed lines are to guide the eye.

tested in our previous publication, since the ILs proved in this work showed toluene distribution ratios in mass considerably higher than those of the binary IL mixture [11]. In addition to the extraction yields of aromatic hydrocarbons, the purity of the extracted BTX using [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs were also studied as a function of Ns, as shown in Fig. 7. In the same figure, the results of purity of the BTX extracted from the same pyrolysis gasoline model using sulfolane published in our previous work were also represented [11]. The purity of the BTX in the extract stream by the [emim]2[Co(SCN)4] with 14 equilibrium stages was 98.3%, using the [bmim]2[Co(SCN)4] was 97.4%, whereas the purity of the BTX extracted by the sulfolane was 94.2% [11]. This order of purity is the same that was observed in the values of toluene/n-heptane selectivity previously. Hence, the extraction of BTX from the pyrolysis gasoline was more selective using the tetrathiocyanatocobaltate-based ILs than the extraction using sulfolane according to the values of purity. 3.5. Simulation of the recovery section based on adiabatic flash distillation units After studying the aromatic extraction, the extract stream requires further separation in order to selectively separate the extracted

aromatics and the IL-based solvent. As was aforementioned, for that purposed vapor-liquid equilibria data for the mixture of extracted hydrocarbons and the ILs is necessary to plan the recovery section. As a consequence, the vapor-liquid equilibria for {n-hexane + nheptane + n-octane + benzene + toluene + p-xylene + tetrathiocyanatocobaltate-based ILs} systems were measured between 323.2 K and 363.2 K with the extract stream compositions obtained in Section 3.4. The equilibrium data are listed in Tables S11 and S12 in the Supplementary Data as function of temperature for both ILs. From the experimental data the aliphatic/aromatic relative volatilities were calculated as follows: K aliph K arom 32 3−1 2 yhexa þ yhepta þ yocta ybenz þ ytol þ yp‐xyl  54  5 ¼ 4 xhexa þ xhepta þ xocta xbenz þ xtol þ xp‐xyl

α aliph;arom ¼

ð11Þ

where K are the K-values for the aliphatics or aromatics, y denote the molar vapor compositions, and x refer to the molar liquid compositions. The aliphatic/aromatic relative volatilities are listed along with the experimental VLE data to quantitatively show the selectivity in the

104

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

Table 4 Results of the simulation of a countercurrent extraction with 14 equilibrium stages using the Kremser equation in the liquid-liquid extraction of aromatics from a model of pyrolysis gasoline employing the [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] at S/F = 5.0 and T = 313.2 K.

[emim]2[Co(SCN)4] Flow (t/h) Benzene (wt.%) Toluene (wt.%) p-Xylene (wt.%) Aromatics (wt.%) n-Hexane (wt.%) n-Heptane (wt.%) n-Octane (wt.%) n-Alkanes (wt.%) IL (wt.%) [bmim]2[Co(SCN)4] Flow (t/h) Benzene (wt.%) Toluene (wt.%) p-Xylene (wt.%) Aromatics (wt.%) n-Hexane (wt.%) n-Heptane (wt.%) n-Octane (wt.%) n-Alkanes (wt.%) IL (wt.%)

Feed

Solvent

Raffinate stream

Extract stream

1000.0 33.80 19.30 13.00 66.10 11.30 11.30 11.30 33.90 0.00

5000.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 100.00

328.0 0.00 0.00 0.09 0.09 33.16 33.23 33.52 99.91 0.00

5672.0 5.96 3.40 2.29 11.65 0.07 0.07 0.05 0.20 88.15

was the use of two flash distillation units in Configuration A (Fig. 8). Flash 1 is aimed in removing the aliphatics from the extract stream and Flash 2 to perform the aromatics/IL separation. The aforementioned algorithm developed to simulate flash distillation units for systems with a high content of non-volatile liquids (Fig. 1), as is the present case, was used to study several pressure-temperature scenarios for Flash 1 based exclusively on the experimental vapor-liquid equilibrium data with the extract stream compositions obtained in Section 3.4. In order to improve the understanding of the effect of the temperature or pressure on the aliphatic/aromatic separations, the recovery (Rarom) and purity (Pu) of the aromatics in the liquid stream are graphically represented in Figs. 9 and 10 for [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4], respectively, calculated as follows:  Puarom ð%Þ ¼ 100 

1000.0 33.80 19.30 13.00 66.10 11.30 11.30 11.30 33.90 0.00

5000.0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 100.00

321.4 0.00 0.00 0.03 0.03 33.21 33.15 33.60 99.97 0.00

5678.6 5.95 3.40 2.29 11.64 0.11 0.11 0.09 0.31 88.05

aliphatic separation. To evaluate the effect of the ILs on the relative volatilities, the vapor-liquid separation of a mixture of hydrocarbons with the same composition of the extract stream but without IL was also determined. The values of aliphatic/aromatic relative volatility in the presence of ILs were 15 to 40, substantially higher in comparison with the aliphatic/aromatic relative volatilities without ILs, which were between 1.3 and 1.8 as can be observed in Tables S11 and S12 in the Supplementary Data. Therefore, the removing of the aliphatics from the extract stream seems to be selective enough to assay its feasibility by a reduce number of consecutive flash distillations. 3.5.1. Equipment and conditions to selectively remove the aliphatics from the extract stream Considering the values of the aliphatics/aromatics relative volatilities, the first approach to conceptually design the recovery section

mbenz þ mtol þ mp‐xyl

 stream

mhexa þ mhepta þ mocta þ mbenz þ mtol þ mp‐xyl

 stream

ð12Þ   mbenz þ mtol þ mp‐xyl stream  Rarom ð%Þ ¼ 100  mbenz þ mtol þ mp‐xyl feed

ð13Þ

where m refers to the mass of each hydrocarbon in each stream that was calculated in the simulations. As can be observed from these figures, the aromatic purity in the bottom liquid stream increases as the value of temperature does or the pressure decreases; however, the recovery of aromatics in the liquid stream drastically decreases at high temperatures or low pressures. Thus, taking into account that only one flash distillation unit is destined to remove the aliphatics from the extract stream, it is not possible to simultaneously achieve the total removing of aliphatics with high values of recovery of aromatics in the liquid stream. At this time, in order to reduce the flow of the recycling streams, the recovery section was planned in three flash distillation units, as shown in Fig. 11, the first two (Flash 1 and Flash 2) are destined to selectively remove the aliphatics and the latter (Flash 3) to perform the aromatic/solvent separation just commented before. The removing of aliphatics is planned in two flash distillation units to work at moderate temperatures and pressure and, therefore, to increase the aliphatic/aromatic relative volatilities in these flash distillators and reduce the flow for the recycling streams.

Fig. 7. Relative purities of the aromatics in the extract phase from the pyrolysis gasoline as a function of the number of equilibrium stages in the countercurrent extraction column at 313.2 K and a solvent to feed ratio of 5.0. ●, using [emim]2[Co(SCN)4]; ▲, using [bmim]2[Co(SCN)4]; ◊, employing sulfolane as solvent at 303.2 K and a solvent to feed ratio of 5.0 [11]. Dashed lines are to guide the eye.

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

105

Fig. 8. Process conceptual design related to Configuration A: liquid-liquid extraction section and separation and recovery section with two flash distillation units.

Fig. 9. Aromatic purity (Puarom) and aromatic recovery (Rarom) in the liquid stream as a function of temperature and pressure for the first flash with [emim]2[Co(SCN)4] ionic liquid. *, T = 323.2 K; ×, T = 333.2 K; △, T = 343.2 K; □, T = 353.2 K; ○, T = 363.2 K. Solid lines are used to guide the eyes.

106

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

Fig. 10. Aromatic purity (Puarom) and aromatic recovery (Rarom) in the liquid stream as a function of temperature and pressure for the first flash with [bmim]2[Co(SCN)4] ionic liquid. *, T = 323.2 K; ×, T = 333.2 K; △, T = 343.2 K; □, T = 353.2 K; ○, T = 363.2 K. Solid lines are used to guide the eyes.

Fig. 11. Process conceptual design related to Configuration B: liquid-liquid extraction section and separation and recovery section with three flash distillation units.

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

The PV-TV selection for Flash 1 in the cases of both tetrathiocyanatocobaltate-based ILs was planned in order to maximize the recovery of aromatics in the liquid stream increasing the maximum their purity in this stream. Accordingly, Flash 1 would work at 343.2 K and 19 kPa for [emim]2[Co(SCN)4] and 343.2 K and 20 kPa for the [bmim]2[Co(SCN)4]. To simulate Flash 2 in Configuration B, it was necessary to perform additional experiments to reproduce the vapor-liquid equilibrium data of the hydrocarbons and ILs forming the liquid stream obtained in the first flash distillation unit. These new VLE data are listed in Table S11 and S12 in the Supplementary Data. Following the same procedure previously described, the recovery of the aromatics and their purity in the liquid stream were calculated at 353.2 K and at several values of PV between the corresponding operation pressure for Flash 1 and 10 kPa and plotted in Fig. 12 for both ILs used in this work. As can be noticed, the design value for the purity of the aromatics at the bottom of the extractive stripping in the Sulfolane process is also represented to be used as benchmark. We have selected the values of pressure and temperature that permit to recover the maximum flow of aromatics in the liquid

107

phase with a better or comparable value of purity of the aromatics in comparison with that obtained in the Sulfolane process (99.9 wt.%) [12]. To achieve this purpose, the temperature was fixed to 353.2 K to assay the total removing of the aliphatics and because higher values of temperature would imply lower recoveries of aromatics in the liquid stream. In the case of [emim]2[Co(SCN)4], the pressure was set again in 19 kPa as this value lets to achieve a higher value of purity for the aromatics (N 99.9 wt.%) with a recovery of these compounds near to 80%. In the case of [bmim]2[Co(SCN)4], the working pressure would be 20 kPa, value that permits to achieve an aromatic purity in the liquid stream higher than 99.8 wt.% and a recovery of aromatics near to 70%. Comparing both tetrathiocyanatocobaltate-based ILs, it is possible to establish a relationship between the best performance of the [emim]2[Co(SCN)4] IL and its extractive properties. The aromatic/aliphatic selectivity is quite higher for that IL than for the other. This extractive property, in addition to reduce the aliphatic content in the extract stream, is related to the aliphatic/aromatic relative volatility as was just studied in our previous work [43]. Therefore, the low aliphatic content in the extract stream and the higher aliphatic/aromatic relative

Fig. 12. Aromatic purity (Puarom, ○) and aromatics recovery in the liquid stream (Rarom, ■) for the second flash at 353.2 K as a function of pressure with a) [emim]2[Co(SCN)4] and b) [bmim]2[Co(SCN)4]. Dashed line denotes the aromatic purity an aromatic purity of 99.9 wt.% and solid lines are used to guide the eyes.

108

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

volatility values achieved using [emim]2[Co(SCN)4] has led to notably improve both the aromatic purity and recovery. 3.5.2. Equipment and conditions to separate the product stream of aromatics and the solvent Following the same procedure previously described, the third flash distillation unit to perform the aromatic/IL separation was studied by determining the vapor-liquid equilibrium for the liquid stream obtained from Flash 2 listed in Tables S11 and S12 in the Supplementary Data. The purity and recovery of aromatics are graphically shown in Figs. 13 and 14, but in this particular case referring to the vapor stream since this flash is destined to recover the aromatics in this stream. As can be seen in both cases, the unique combination that assays a good recovery of aromatics is the highest temperature checked (383.2 K) and the lowest pressure proved (3 kPa). In addition to this, the aromatic purity in the vapor stream and the IL purity in the liquid stream are incremented as the aromatic recovery does as a result of the almost full recovery of the aliphatic hydrocarbons at all scenarios simulated. Thus, for both cases, the third flash would work at 383.2 K and 3 kPa to improve the three parameters: recovery of aromatics, purity of aromatics, and purity of the regenerated solvent. The purity the

aromatics obtained in the vapor stream of the third flash was 99.9 wt.% for the [emim]2[Co(SCN)4] and 99.7 wt.% for the [bmim]2[Co(SCN)4], whereas the ILs obtained in the liquid stream of Flash 3 had a purity higher than 99.3 wt.%. Once all streams are characterized in both flow and compositions, these values are listed together with their conditions in Tables S13 and S14 showing the flows and compositions of numbered streams in Fig. S1 also in the Supplementary Data for [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4], respectively. It is important to mention that the temperature of the feed stream for the three flash distillation units have been calculated by enthalpy balance as indicated in Fig. 1. As can be noticed, the calculated temperatures are below the estimated MOT for both tetrathiocyanatocobaltate-based ILs, fact that ensures no thermal decomposition of the solvent. 4. Conclusions In this paper, we have tested the applicability of the [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] ILs in the separation of aromatic hydrocarbons from pyrolysis gasoline. We have experimentally determined all the required data to simulate and design the whole

Fig. 13. Aromatic purity (Puarom) and aromatics recovery (Rarom) in the vapor stream as a function of temperature and pressure for the third flash with [emim]2[Co(SCN)4] IL. *, T = 363.2 K; ×, T = 373.2 K; △, T = 383.2 K. Solid lines are used to guide the eyes.

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

109

Fig. 14. Aromatic purity (Puarom) and aromatics recovery (Rarom) in the vapor stream as a function of temperature and pressure for the third flash with [bmim]2[Co(SCN)4] IL. *, T = 363.2 K; ×, T = 373.2 K; △, T = 383.2 K. Solid lines are used to guide the eyes.

process of extraction of BTX using ILs. First, a thermophysical characterization of the ILs was performed. The tetrathiocyanatocobaltate-based ILs showed densities comparable to the sulfolane but their viscosities were higher than those of the organic solvent. The higher viscosity values of tetrathiocyanatocobaltate-based ILs, compared to sulfolane viscosity, would limit the mass transfer in the separations. In this respect, additional kinetic studies, besides the thermodynamic experimentation, should be done to complete the characterization of the process. The thermal stability of the ILs was also studied, being the maximum operation temperatures of the ILs higher than 393 K to be employed for 8000 h without thermal decomposition. In the separation of toluene from n-heptane, [emim]2[Co(SCN)4] and [bmim]2[Co(SCN)4] exhibited extractive properties substantially higher than those of sulfolane and other promising ILs. A solvent to feed ratio in mass of 5.0 was selected as the optimal in the extraction of BTX from the pyrolysis gasoline using both ILs. The liquid-liquid extraction column was simulated using the Kremser method, selecting 14 equilibrium stages in the extractor to ensure yields of extraction of aromatics and purities of the extracted BTX higher using the ILs than those employing sulfolane. The vapor-liquid equilibrium between the extracted aromatics and the ILs was also studied to simulate and design the recovery section. This section was formed by three flash distillations and works at

moderate temperatures and requiring also vacuum to achieve competitive recoveries and purities. At these conditions the best overall results were obtained using [emim]2[Co(SCN)4]; the stream of recovered aromatics had a 99.9 wt.% of purity, the aromatic content in the pyrolysis gasoline was reduced from 66.1 wt.% to 0.09 wt.%, using a low recycling flow of aromatics in the process. All temperatures in the process were selected under the MOT of the ILs to ensure their thermal stability and the ILs were regenerated in the recovery section obtaining the IL with a purity higher than 99.3 wt.%. To completely evaluate the potential use of tetrathiocyanatocobaltate-based ILs at industrial scale, future studies should be focused on the study of mass transfer of hydrocarbons in the solvent and on the recycle and reuse of the ILs. In addition, an economical evaluation of the proposed process should be made considering energy consumptions and solvent cost. Acknowledgments The authors are grateful to Ministerio de Economía y Competitividad (MINECO) of Spain and Comunidad Autónoma de Madrid for financial support of Projects CTQ2014–53655-R and S2013/MAE-2800, respectively. Pablo Navarro and Noemí Delgado-Mellado also thank MINECO for awarding them an FPI grant (Reference BES–2012–052312 and

110

M. Larriba et al. / Fuel Processing Technology 159 (2017) 96–110

Reference BES–2015–072855, respectively). Victor Stanisci thanks the financial support of Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) of Brasil. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.fuproc.2017.01.027. References [1] N.V. Plechkova, K.R. Seddon, Applications of ionic liquids in the chemical industry, Chem. Soc. Rev. 37 (2008) 123–150. [2] Y. Nie, C. Li, H. Meng, Z. Wang, N,N-dialkylimidazolium dialkylphosphate ionic liquids: their extractive performance for thiophene series compounds from fuel oils versus the length of alkyl group, Fuel Process. Technol. 89 (2008) 978–983. [3] R. Anantharaj, T. Banerjee, COSMO-RS based predictions for the desulphurization of diesel oil using ionic liquids: effect of cation and anion combination, Fuel Process. Technol. 92 (2011) 39–52. [4] C.D. Wilfred, C.F. Kiat, Z. Man, M.A. Bustam, M.I.M. Mutalib, C.Z. Phak, Extraction of dibenzothiophene from dodecane using ionic liquids, Fuel Process. Technol. 93 (2012) 85–89. [5] G.W. Meindersma, A.J.G. Podt, A.B. de Haan, Selection of ionic liquids for the extraction of aromatic hydrocarbons from aromatic/aliphatic mixtures, Fuel Process. Technol. 87 (2005) 59–70. [6] G.O. Yahaya, F. Hamad, A. Bahamdan, V.V.R. Tammana, E.Z. Hamad, Supported ionic liquid membrane and liquid–liquid extraction using membrane for removal of sulfur compounds from diesel/crude oil, Fuel Process. Technol. 113 (2013) 123–129. [7] U. Maity, J.K. Basu, S. Sengupta, Performance study of extraction and oxidation–extraction coupling processes in the removal of thiophenic compounds, Fuel Process. Technol. 121 (2014) 119–124. [8] S.A. Dharaskar, K.L. Wasewar, M.N. Varma, D.Z. Shende, K.K. Tadi, C.K. Yoo, Synthesis, characterization, and application of novel trihexyl tetradecyl phosphonium bis (2,4,4-trimethylpentyl) phosphinate for extractive desulfurization of liquid fuel, Fuel Process. Technol. 123 (2014) 1–10. [9] I. Dominguez, E.J. González, A. Dominguez, Liquid extraction of aromatic/cyclic aliphatic hydrocarbon mixtures using ionic liquids as solvent: literature review and new experimental LLE data, Fuel Process. Technol. 125 (2014) 207–216. [10] G.C. Laredo, N.V. Likhanova, I.V. Lijanova, B. Rodriguez-Heredia, J.J. Castillo, P. PerezRomo, Synthesis of ionic liquids and their use for extracting nitrogen compounds from gas oil feeds towards diesel fuel production, Fuel Process. Technol. 130 (2015) 38–45. [11] M. Larriba, P. Navarro, E.J. González, J. García, F. Rodríguez, Dearomatization of pyrolysis gasolines from mild and severe cracking by liquid–liquid extraction using a binary mixture of [4empy][Tf2N] and [emim][DCA] ionic liquids, Fuel Process. Technol. 137 (2015) 269–282. [12] J. Gary, G. Handwerk, M. Kaiser, Petroleum Refining Technology and Economics, CRC Press, Boca Raton, FL, 2007. [13] G.W. Meindersma, A.B. de Haan, Conceptual process design for aromatic/aliphatic separation with ionic liquids, Chem. Eng. Res. Des. 86 (2008) 745–752. [14] R.I. Canales, J.F. Brennecke, Comparison of ionic liquids to conventional organic solvents for extraction of aromatics from aliphatics, J. Chem. Eng. Data 61 (2016) 1685–1699. [15] M. Larriba, P. Navarro, J. García, F. Rodríguez, Selective extraction of toluene from nheptane using [emim][SCN] and [bmim][SCN] ionic liquids as solvents, J. Chem. Thermodyn. 79 (2014) 266–271. [16] M. Charkaborty, H.J. Bart, Highly selective and efficient transport of toluene in bulk ionic liquid membranes containing Ag+ as carrier, Fuel Process. Technol. 88 (2007) 43–49. [17] D. Gorri, A. Ruiz, A. Ortiz, I. Ortiz, The use of ionic liquids as efficient extraction medium in the reactive separation of cycloolefins from cyclohexane, Chem. Eng. J. 154 (2009) 241–245. [18] R. Li, H. Xing, Q. Yang, X. Zhao, B. Su, Z. Bao, Y. Yang, Q. Ren, Selective extraction of 1hexene against n-hexane in ionic liquids with or without silver salt, Ind. Eng. Chem. Res. 51 (2012) 8588–8597. [19] M. Larriba, P. Navarro, J. García, F. Rodríguez, Highly selective extraction of toluene from n–heptane using [emim][SCN] pure ionic liquid and its mixtures with several transition metal salts, Chem. Eng. Trans. 32 (2013) 1957–1962. [20] T. Peppel, M. Köckerling, M. Geppert-Rybczyńska, R.V. Ralys, J.K. Lehmann, S.P. Verevkin, A. Heintz, Low viscosity paramagnetic ionic liquids with doubly charged [Co(NCS)4]2− ions, Angew. Chem. 49 (2010) 7116–7119.

[21] M. Geppert-Rybczyńska, J.K. Lehmann, T. Peppel, M. Köckerling, A. Heintz, Studies of physicochemical and thermodynamic properties of the paramagnetic1-alkyl-3methylimidazolium ionic liquids (EMIm)2[Co(NCS)4] and (BMIm)2[Co(NCS)4], J. Chem. Eng. Data 55 (2010) 5534–5538. [22] N. Zhao, J. Jacquemin, R. Oozeerally, V. Degirmenci, New method for the estimation of viscosity of pure and mixtures of ionic liquids based on the UNIFAC–VISCO model, J. Chem. Eng. Data 61 (2016) 2160. [23] P. Navarro, M. Larriba, E. Rojo, J. García, F. Rodríguez, Thermal properties of cyanobased ionic liquids, J. Chem. Eng. Data 58 (2013) 2187–2193. [24] A. Seeberger, A.K. Andresen, A. Jess, Prediction of long-term stability of ionic liquids at elevated temperatures by means of non-isothermal thermogravimetrical analysis, Phys. Chem. Chem. Phys. 11 (2009) 9375–9381. [25] H.G. Franck, J.W. Staldelhofer, Industrial Aromatic Chemistry, Springer-Verlag, New York, 1988. [26] G.W. Meindersma, A.R. Hansmeier, A.B. de Haan, Ionic liquids for aromatics extraction. Present status and future outlook, Ind. Eng. Chem. Res. 49 (2010) 7530–7540. [27] M. Larriba, P. Navarro, J. García, F. Rodríguez, Liquid-liquid extraction of toluene from heptane using [emim][DCA], [bmim][DCA], and [emim][TCM] ionic liquids, Ind. Eng. Chem. Res. 52 (2013) 2714–2720. [28] E.J. Henley, J.D. Seader, Equilibrium-Stage Separation Operations in Chemical Engineering, John Wiley and Sons, New York, 1981. [29] P. Navarro, M. Larriba, J. García, E.J. González, F. Rodríguez, Vapor–liquid equilibria of {n-heptane + toluene + [emim][DCA]} system by headspace gas chromatography, Fluid Phase Equilib. 387 (2015) 209–216. [30] R.H. Perry, D.W. Green, J.O. Maloney, Perry's Chemical Engineers' Handbook, McGraw-Hill, New York, 1999. [31] B. González, N. Calvar, E.J. González, A. Dominguez, Density and Viscosity experimental data of the ternary mixtures 1-propanol or 2-propanol + water + 1ethyl-3-methylimidazolium ethylsulfate. Correlation and prediction of physical properties of the ternary systems, J. Chem. Eng. Data 53 (2008) 881. [32] Y. Qiao, F. Yan, S. Xia, S. Yin, P. Ma, Densities and viscosities of [Bmim][PF6] and binary systems [Bmim][PF6] + Ethanol, [Bmim][PF6] + Benzene at several temperatures and pressures: determined by the falling-ball method, J. Chem. Eng. Data 56 (2011) 2379. [33] L. Galán Sánchez, J. Ribé Espel, F. Onink, G.W. Meindersma, A.B. de Haan, Density, viscosity, and surface tension of synthesis grade imidazolium, pyridinium, and pyrrolidinium based room temperature ionic liquids, J. Chem. Eng. Data 54 (2009) 2803. [34] J.G. Huddleston, A.E. Visser, W.R. Reichert, H.D. Willauer, G.A. Broker, R.D. Rogers, Characterization and comparison of hydrophilic and hydrophobic room temperature ionic liquids incorporating the imidazolium cation, Green Chem. 3 (2001) 156–164. [35] J.M. Crosthwaite, M.J. Muldoon, J.K. Dixon, J.L. Anderson, J.F. Brennecke, Phase transition and decomposition temperatures, heat capacities and viscosities of pyridinium ionic liquids, J. Chem. Thermodyn. 37 (2005) 559–568. [36] M. Larriba, P. Navarro, M. Gonzalez-Miquel, S. Omar, J. Palomar, J. García, F. Rodríguez, Dicyanamide-based ionic liquids in the liquid-liquid extraction of aromatics from alkanes: experimental evaluation and computational predictions, Chem. Eng. Res. Des. 109 (2016) 561–572. [37] J.A. Nóvoa-López, E. López Lago, J.A. Seijas, M.P. Vázquez-Tato, J. Troncoso, R. de la Fuente, J.R. Salgueiro, H. Michinel, Nonlinear absorption in ionic liquids with transition metallic atoms in the anion, Opt. Mater. 52 (2016) 144–149. [38] D.F. Othmer, P.E. Tobias, Toluene and acetaldehyde systems, tie line correlation, partial pressures of ternary liquid systems and the prediction of tie lines, Ind. Eng. Chem. 34 (1942) 693–696. [39] G.W. Meindersma, A.J.G. Podt, A.B. de Haan, Ternary liquid-liquid equilibria for mixtures of toluene + n-heptane+ an ionic liquid, Fluid Phase Equilib. 247 (2006) 158–168. [40] S. García, M. Larriba, J.S. Torrecilla, J. García, F. Rodríguez, Liquid-liquid extraction of toluene from heptane using 1-alkyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ionic liquids, J. Chem. Eng. Data 56 (2011) 113–118. [41] M. Larriba, P. Navarro, J. García, F. Rodríguez, Separation of toluene from n-heptane,2,3-dimethylpentane, and cyclohexane using binary mixtures of [4empy][Tf2N] and [emim][DCA] ionic liquids as extraction solvents, Sep. Purif. Technol. 120 (2013) 392–401. [42] M. Larriba, P. Navarro, N. Delgado-Mellado, V. Stanisci, J. García, F. Rodríguez, Separation of aromatics from n-alkanes using tricyanomethanide-based ionic liquids: liquid-liquid extraction, vapor-liquid separation, and thermophysical characterization, J. Mol. Liq. 223 (2016) 880–889. [43] P. Navarro, M. Larriba, J. García, E.J. González, F. Rodríguez, Selective recovery of aliphatics from aromatics in the presence of the {[4empy][Tf2N] + [emim][DCA]} ionic liquid mixture, J. Chem. Thermodyn. 96 (2016) 134–142.