Thermodynamic modelling of liquid-liquid extraction

1 downloads 0 Views 2MB Size Report
The liquid-liquid equilibrium (LLE) data for the ternary mixture of [Cnmim][Phe], dodecane and naphthenic acid were experimentally obtained at a constant ...
Journal of Molecular Liquids 219 (2016) 513–525

Contents lists available at ScienceDirect

Journal of Molecular Liquids journal homepage: www.elsevier.com/locate/molliq

Thermodynamic modelling of liquid-liquid extraction of naphthenic acid from dodecane using imidazolium based phenolate ionic liquids Syed Nasir Shah a,⁎, M.I. Abdul Mutalib d, M. Farid Ismail b, Humbul Suleman d, Kallidanthiyil Chellappan Lethesh a,c, Rashidah Binti Mohd Pilus e a

Centre of Research in Ionic Liquids, Universiti Teknologi Petronas, Bandar Seri Iskandar 32610, Perak, Malaysia Department of Chemistry, Universiti Putra Malaysia, 4330 Serdang, Malaysia Center for Biofuel and Biochemical Research, Universiti Teknologi Petronas, Bandar Seri Iskandar 32610, Perak, Malaysia d Department of Chemical Engineering, Universiti Teknologi Petronas, Bandar Seri Iskandar 32610, Perak, Malaysia e Department of Petroleum Engineering, Universiti Teknologi Petronas, Bandar Seri Iskandar 32610, Perak, Malaysia b c

a r t i c l e

i n f o

Article history: Received 14 September 2015 Accepted 18 March 2016 Available online xxxx Keywords: Ionic liquids (ILs) Naphthenic acid (NA) Phenolate anion (Phe) Alkyl methyl imidazole Liquid-liquid equilibrium (LLE) Non-random two liquid model (NRTL) Universal Quasi Chemical (UNIQUAC)

a b s t r a c t In this work, N-alkyl imidazolium based ionic liquids with phenolate anions have been used for the separation of naphthenic acid from model oil. The liquid-liquid equilibrium (LLE) data for the ternary mixture of [Cnmim][Phe], dodecane and naphthenic acid were experimentally obtained at a constant temperature of 303.15 K and atmospheric pressure. The effect of chain length on the extraction capability was observed by calculating the distribution coefficient. The experimental tie line data was correlated using the non-random two liquid model (NRTL) and Universal Quasi Chemical (UNIQUAC) model and new interaction parameters for the ternary systems are reported. The experimental data provides a good correlation with the modelling data and a very low root mean square deviation (RMSD) value was observed for all the systems. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Ionic liquids (ILs) being designer solvents have found a large number of applications in academia as well as in the industry. The properties like negligible vapour pressure, high thermal and chemical stability, non-flammability; high heat capacity, high ionic conductivity etc. make them a greener choice for a range of industrial applications. In addition, ILs exhibit a very good solubility for organic compounds thus making them an excellent solvent for liquid-liquid extraction process [1,2]. These desirable properties make them suitable in numerous applications in which many conventional solvents are non-applicable and in-effective. Liquid-liquid extraction has found large applications in the separation industry because of the mild process conditions applied during the process i.e. low temperature and pressure. The solvent to be used in liquid-liquid extraction should have a high selectivity towards the solute. Moreover, it should be economical to produce, recyclable, and robust to withstand various processing environments. Most of the organic solvents are carcinogenic, highly volatile and their recyclability is very low. Contrarily, most of the ILs are greener, non-volatile and highly recyclable [3–6]. But the core issue that stands in the way of commercialisation is the high cost ⁎ Corresponding author. E-mail address: [email protected] (S.N. Shah).

http://dx.doi.org/10.1016/j.molliq.2016.03.053 0167-7322/© 2016 Elsevier B.V. All rights reserved.

of ILs. However, in one of the recent modelling and simulation studies on the ILs production process shows that ILs can be produced at lower cost ( $1.24 kg−1), which is in comparison with most of the organic solvents such as acetone or ethyl acetate with a cost of $1.30–$1.40 kg−1 [7].Similarly, in another study, the extraction of aromatic hydrocarbon from aliphatic hydrocarbon with 4-methyl-N-butylpyridinium tetrafluoroborate was modeled using ASPEN resulting in a positive margin of about € 20 million per year [8]. These results indicate that ILs are not necessarily expensive, and therefore large-scale ILs-based processes can become a commercial reality. Naphthenic acid removal from acidic crude is one of the major concerns of refiners all over the world. Naphthenic acid presence in heavy crude can cause corrosion in the refinery equipment and storage tanks. Additionally, it affects the combustion characteristic of the finished products [9,10]. Thus the removal of naphthenic acid from crude oil is highly desirable. On the other hand, naphthenic acids have a very complicated moieties and a wide variety of compounds lie in the definition of naphthenic acid. Naphthenic acids can be defined as a mixture of cyclic, aromatic and linear monocarboxylic acids present in the crude oil with the general formula CnH2n + zO2, where n indicates the number of carbon atoms, z indicates the deficiency of hydrogen because of the presence cyclic or aromatic groups. The value of z can be a negative integer or zero [11–14]. Commercial naphthenic acid is obtained from jet fuel, kerosene and diesel fractions by caustic wash followed by acidification of caustic stream with sulfuric

514

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

acid. This process generates significant volume of sodium sulphate brine that is of the order of 10 L per kg of crude naphthenic acid. Moreover, it contains phenolic compounds that should be disposed in an environmental friendly way. In addition there is some carryover of phenolic and sulphur compounds to the naphthenic acid streams affecting the properties of the final product [10]. There had been a lot of other lab scale methods for removal of naphthenic acid. The details of these methods can be found elsewhere [15–21]. Ionic liquids had also been used to remove naphthenic acid form crude oil. A lot of studies had been conducted to remove naphthenic acid from model oil and crude oil in which the ILs had been recycled for many times without compromising on their extraction efficiency and producing negligible waste [22–27]. In addition, negligible amount of ILs had been used to completely de-acidify crude oil having a very high acid number ( 4.74 (± 0.01) mg of KOH/g ) [26]. However, scarce published data is available for liquid-liquid equilibrium data of commercial naphthenic acid using ionic liquids. For instance, cyclohexane carboxylic acids (CCA) has been used as an analogy to naphthenic acid, but fails to address variety of naphthenic acids present in literature. Furthermore, the liquid-liquid equilibrium data have not been modelled previously using any thermodynamic models [28], creating a bottleneck in expansion of experimental data to modelling and simulation studies. Therefore, this work provides ample experimental data for the liquid-liquid equilibrium data for the discussed ternary system and their thermodynamic equilibrium. Subsequently, this extends our previous findings related to the aforementioned case [25]. In the current study four different ternary systems were used with dodecane as model oil, commercial naphthenic acid and [Cnmim][Phe]. The effect of chain length on the ionic liquids on separation efficiency was also studied. The distribution coefficient of the ionic liquids were calculated from the LLE data. The LLE data was correlated using NRTL and UNIQUAC model. New interaction parameters for both thermodynamic approaches are reported. Additionally, the volume parameter, R and the surface area parameter, Q for UNIQUAC model are determined using Density Functional Theory (DFT) and the FIXPVA method [29] for determining the cavity size as typically used in the Polarizable Continuum Model (PCM) method. The correlated results are in excellent agreement with experimental values.

2. Material and methods The chemicals used for the synthesis of ILs were purchased from Acros Organics (Geel Belgium) and Sigma Aldrich (Bornem, Belgium). The synthesis of the ILs were carried out according to the already

reported procedure [30]. The properties and structures of all the synthesized ionic liquids are shown in Table 1. 2.1. UNIQUAC volume and surface area structural parameters To get the surface and the volume parameters, the geometry of the cation and the anion of the ionic liquid is optimized using the density functional theory (DFT) method. The B3LYP functional with RIJCOSX SCF approximation, together with the minimally augmented ma-def2-TZVPP basis set [31,32] was used in the geometry optimization step using the ORCA quantum chemistry package [33,34]. The D3BJ dispersion-correction with Becke-Johnson damping [35, 36] was applied to account for the dispersion error of the DFT method. From the optimized geometry, the surface and volume values for the ILs had been calculated using the Fixed Points with Variable Areas (FIXPVA) method [29]. The FIXPVA method is a method used to calculate the cavity size for Polarizable Continuum Model (PCM) [37] as implemented in the GAMESS software package [38]. Similar method was used by Manohar et al. [39] to obtain the surface and volume parameters. 2.2. Liquid–liquid extraction experiments The liquid-liquid equilibrium study for the four ternary systems was performed in 8 mL vials with airtight caps. The vials were weighted and known amount of dodecane, naphthenic acid and ionic liquids were added to the vials. After this the mixture were placed in a shaking incubator for 5 hours to attain thermodynamic equilibrium. The mixture was shaken at a constant speed of 700 RPM at a fixed temperature of 303.15 K (± 0.1 K). To ensure the complete separation between the ionic liquids and hydrocarbon phase the vials were left overnight at a constant temperature of 303.15 K (±0.1 K). The structure of naphthenic acid is very complicated as already mentioned. This complicated structure of naphthenic acid makes it quite impossible to quantify via Gas chromatography or High Pressure liquid chromatography technique [11–14]. For this reason we had quantified the hydrocarbon layer by measuring refractive index. Refractive index had long been used to quantify the concentration of different solutes in aqueous as well as in organic solutions [40]. An ATAGO programmable digital refractometer (RX-5000α) was used with an uncertainty of ±4 × 10−5 and temperature accuracy of ± 0.05 K. The mole fraction against refractive index graphs were drawn as shown in Figs. 1 and 2. All the concentrations of the hydrocarbon phase and IL phase was determined from these graphs. To ensure no carryover of the ionic liquids to the hydrocarbon layer the 1H NMR of the hydrocarbon layer was performed using Bruker

Table 1 Structures of synthesized ionic liquids plus water content and halide content. ILs

Chemical structure

Water cotent (ppm)

Halide (ppm)

[C4mim][Phe]

170

102

[C6mim][Phe]

153

93

[C8mim][Phe]

149

90

[C10mim][Phe]

161

97

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

515

Table 2 Experimental LLE data for ternary system Dodecane (1) + [C4mim][Phe] (2) + Naphthenic Acid (3) on mole fraction basis. Ionic liquid rich phase

ẋ1

ẋ2

ẋ3

ẍ1

ẍ2

ẍ3

0.9956 0.9470 0.9010 0.8557 0.8103 0.7660 0.7217 0.6788

0 0 0 0 0 0 0 0

0.0044 0.0530 0.0990 0.1443 0.1897 0.2340 0.2783 0.3212

0 0 0 0 0 0 0 0

0.9853 0.9695 0.9494 0.9286 0.9097 0.8901 0.8735 0.8548

0.0147 0.0305 0.0506 0.0714 0.0903 0.1099 0.1265 0.1452

a

Fig. 1. Refractive index vs. concentration curve for naphthenic acid.

Avance 500 MHz NMR spectrometer. The 1H NMR spectra (Figure 1 in supporting information) showed no peaks from the ionic liquid thus ensuring no carry-over of the ionic liquid to hydrocarbon layer. In addition to ensure no carry-over of dodecane in to the IL layer, the IL layer was analysed using GC-MS [41]. A small portion from the IL layer

βa

Hydrocarbon rich phase

3.3173 0.5746 0.5114 0.4944 0.4759 0.4696 0.4544 0.4522

β = distribution coefficient calculated via Eq. (1).

was taken and was analysed for GC-MS using the selected method. The GC-MS spectrum shows no traces of dodecane in the IL layer. The spectrum for pure dodecane is attached at Figure-2 in the supporting information. The spectrum for the IL layer is shown in Figure-3 in supporting information. The GC spectra show no peak for dodecane thus ensuring no carryover of dodecane into the IL layer. 3. Results and discussion Commercial naphthenic acid was used to determine ternary liquid-liquid equilibrium data. Dodecane was used as model oil

Fig. 2. Refractive index vs. concentration curve for ionic liquids layer.

516

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

Fig. 3. Experimental liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C4mim][Phe] at T = 303.23.

because most of the world commercial naphthenic acid is produced from kerosene and dodecane is considered as a perfect analogy to kerosene. The composition of feed for all the four ILs was taken as

approximately constant in an attempt to determine the potential of IL for the extraction of naphthenic acid. The mole fraction of naphthenic acid was varied from 0.007 to 0.33 to replicate the actual processing

Fig. 4. Experimental liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C6mim][Phe] at T = 303.23.

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

517

Table 3 Experimental LLE data for ternary system Dodecane (1) + [C6mim][Phe] (2) +

3.1. LLE data of Dodecane (1) + [Cnmim][Phe] (2) + Naphthenic Acid (3)

Naphthenic Acid (3) on mole fraction basis.

The LLE data for [C4mim][Phe] is given in Table 2 and the ternary plot for [C4mim][Phe] is plotted in Fig. 3. A closer look at the figure reveals that most of the tie lines have a negative slope, which indicates that more IL is required for the better separation of naphthenic acid from dodecane. Furthermore, the slope of the tie line is positive at very low concentration of naphthenic acid. The LLE plot for [C6mim][Phe] is shown in Fig. 4. The figure shows a negative slope for all the tie lines except at very low concentration of naphthenic acid thus indicating that more solvent ( IL) required for the extraction of naphthenic acid from dodecane. Moreover, Table 3 reveals that [C6mim][Phe] is capable of extracting more naphthenic acid as compared to [C4mim][Phe]. The LLE data for [C8mim][Phe] is given in Table 4 indicates that the concentration of dodecane in the extract phase continues to increase and the maximum value of 0.093 was observed in the extract phase. [C8mim][Phe] shows better extraction of naphthenic acid, when compared to [C4mim][Phe] and [C6mim][Phe]. The ternary plot in Fig. 5 shows a negative slope for all the concentrations except at very low concentration of naphthenic acid, where the slope shows a positive trend. The ternary diagram for [C10mim][Phe] shown in Fig. 6 displays the highest ability to extract naphthenic acid from dodecane. Three tie lines have a positive slope when compared to the rest of the three ILs discussed above, which had only one and two positive slope tie lines respectively. Table 5 indicates that the concentration of naphthenic acid is the highest in extract phase as compared to [C 4mim][Phe], [C 6mim][Phe] and [C8mim][Phe] thus showing that [C10mim][Phe] had the highest extraction capability. This is in agreement with previous works [42,43], where increase in chain length results in increase of extraction capacity. The extraction capability in terms of naphthenic acid for all the four ILs is in the following order [C 10mim][Phe] N [C 8mim][Phe] N [C 6 mim][Phe] N [C4mim][Phe].

Hydrocarbon rich phase

βa

Ionic liquid rich phase

ẋ1

ẋ2

ẋ3

ẍ1

ẍ2

ẍ3

0.9967 0.9502 0.9037 0.8585 0.8125 0.7671 0.7232 0.6809

0 0 0 0 0 0 0 0

0.0033 0.0498 0.0963 0.1415 0.1875 0.2329 0.2768 0.3191

0 0 0 0 0 0 0 0

0.9843 0.9603 0.9377 0.9134 0.8939 0.8762 0.8576 0.8369

0.0157 0.0397 0.0623 0.0866 0.1061 0.1238 0.1424 0.1631

a

4.765 0.798 0.647 0.612 0.566 0.532 0.514 0.511

β = distribution coefficient calculated via Eq. (1).

Table 4 Experimental LLE data for ternary system Dodecane (1) + [C8mim][Phe] (2) + Naphthenic Acid (3) on mole fraction basis. Hydrocarbon rich phase

βa

Ionic liquid rich phase

ẋ1

ẋ2

ẋ3

ẍ1

ẍ2

ẍ3

0.9967 0.9521 0.9043 0.8587 0.8134 0.7684 0.7255 0.6841

0 0 0 0 0 0 0 0

0.0033 0.0479 0.0957 0.1413 0.1866 0.2316 0.2745 0.3159

0 0 0 0 0 0 0 0

0.9822 0.9519 0.9305 0.9065 0.8851 0.8669 0.8465 0.8251

0.0178 0.0481 0.0695 0.0935 0.1149 0.1331 0.1535 0.1749

a

5.476 1.003 0.726 0.662 0.615 0.575 0.559 0.554

β = distribution coefficient calculated via Eq. (1).

conditions [10]. The bulk of raffinate phase consisted of dodecane, while the extract phase was rich in ionic liquids. The composition of dodecane in the feed was varied from 0.67 to 0.53. Similarly, the mole fraction of IL in the feed was decreased from 0.32 to 0.13.

Fig. 5. Experimental liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C8mim][Phe] at T = 303.23.

518

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

Fig. 6. Experimental liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C10mim][Phe] at T = 303.23.

3.2. Distribution coefficient Distribution coefficient quantitatively indicates the distribution of solute between the extract and raffinate phase. It is the ratio of solute in the extract phase to the raffinate phase. β¼

xENA : xRNA

ð1Þ

The values of distribution coefficient for all the four ILs are shown in Tables 2, 3, 4 and 5 correspondingly. The plot of distribution coefficient as a function of naphthenic acid mole fraction in dodecane phase is shown in Fig. 7. The values of the distribution coefficient decrease for all the four ILs with the increase in the naphthenic acid concentration in the raffinate phase. The value of distribution coefficient greater than unity is obtained at very low concentration of naphthenic acid only. Thus, more IL would be required to extract the naphthenic acid from dodecane at higher concentrations. The highest

value of the distribution coefficient was observed for [C10mim][Phe]. This conforms to the published data where the distribution coefficient increases with the increase in chain length. For the extraction of toluene using [Cnmim][NTf2] the distribution coefficient increases with the increase in the chain length [44]. Similarly, an increase in chain length results in an increase of distribution coefficient for the extraction of cyclohexane carboxylic acid from dodecane [28].

3.3. Thermodynamic framework The thermodynamic framework for the LLE ternary data has been developed by using Non Random Two Liquid (NRTL) equation developed

Table 5 Experimental LLE data for ternary system Dodecane (1) + [C10mim][Phe] (2) + Naphthenic Acid (3) on mole fraction basis. Hydrocarbon rich phase

βa

Ionic liquid rich phase

ẋ1

ẋ2

ẋ3

ẍ1

ẍ2

ẍ3

0.9973 0.9606 0.9109 0.8640 0.8195 0.7751 0.7322 0.6908

0 0 0 0 0 0 0 0

0.0027 0.0394 0.0891 0.1360 0.1805 0.2249 0.2678 0.3092

0 0 0 0 0 0 0 0

0.9813 0.9304 0.9079 0.8825 0.8538 0.8283 0.8022 0.7750

0.0187 0.0696 0.0921 0.1175 0.1462 0.1717 0.1978 0.2250

a

β = distribution coefficient calculated via Eq. (1).

6.994 1.768 1.034 0.863 0.810 0.764 0.739 0.728

Fig. 7. Distribution coefficient of naphthenic acid as a function of mole fraction of naphthenic acid in dodecane phase.

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525 Table 6 Values of the NRTL parameters obtained from LLE data by regression at T = 303.2 K. Components (i-j)

Aij (K)

Aij (K)

Dodecane + Naphthenic Acid + [C4mim][Phe] 1–2 -4108.02 -6875.84 1–3 -2.333 2.335 2–3 3.276 -3.273 Dodecane + Naphthenic Acid + [C6mim][Phe] 1–2 -3277.403 -4493.76 1–3 -5.6135 5.624 2–3 7.976 -7.955 Dodecane + Naphthenic Acid + [C8mim][Phe] 1–2 2201.26 1416.14 1–3 -7.247 7.2646 2–3 10.134 -10.10 Dodecane + Naphthenic Acid + [C10mim][Phe] 1–2 -884.46 -1086.78 1–3 -7.4594 7.4779 2–3 8.4879 -8.4641 a b

Fa

RMSDb

3.744 × 10-01

2.744 × 10-05

519

species in the solution. The reference states are taken as pure liquids of all species at system temperature and pressure. Thus, the total Gibbs energy (per mole of mixture) is given by the molar Gibbs energy of mixing gM, which is g M Xn gE ¼ x ln xi þ i¼1 i RT RT

3.716 × 10-01

1.974 × 10-05

4.009 × 10-01

2.365 × 10-05

4.519 × 10-01

2.292 × 10-05

where, xi represents the mole fraction of component i, gE is the molar excess Gibbs energy, R is the universal gas constant and n is the number of species. This model predicts large heat of mixing, which is the characteristics of electrolyte solution [45]. The non-ideal liquid phase activity coefficient γi of component i is given as follows; Xm ln γi ¼

F = calculated via Eq. (6). RMSD = calculated via Eq. (7).

τij ¼ by Renon and Prausnitz [45] and Universal Quasi Chemical (UNIQUAC) model developed by Abrams and Prausnitz [46].

ð2Þ

xτ G j¼1 j ji ji Xm x G k¼1 k ki

  g ij −g jj RT

  Gij ¼ exp −α ij τ ij :

þ

Xm j¼1

0 x j Gij

Xm

x G k¼1 k kj

@τij −

Xm

1

xr τrj Grj A Xr¼1 m x G k¼1 k kj

ð3Þ

ð4Þ

ð5Þ

3.4. NRTL model The NRTL model had been successfully used to correlate the ternary LLE systems having ILs. This model presumes the ILs as completely associated compounds [47]. This assumption helps to simplify the model by presuming that each cation is fully paired with an anion and hence considering ILs to be a single molecular

In the above equation, g represents an energy parameters that represents the interaction between species, γ represents the activity coefficient, αij = αji represents the non-randomness in the mixture (αij = 0 represents complete randomness, or an ideal solution). Although αij can be considered as an adjustable parameter, it is deemed as a constant

Fig. 8. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C4mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by NRTL model.

520

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

Fig. 9. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C6mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by NRTL model.

in order to limit number of binary parameters. In case of IL/solvent and IL/co-solvent binary LLE data, mutual solubilities are used for the parameter estimation, which can be determined by using the equal

activity equations [48]. For a model having no liquid phase splitting, there will be no parameter solution to this equation. In the current system, αij = αji = 0.2 is used for immiscible binaries following Sørenson

Fig. 10. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C8mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by NRTL model.

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

521

Fig. 11. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C10mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by NRTL model.

and Arlt [48]. The interaction parameters Δgij = Tij are predicted from experimental data by minimizing the objective function, OF given below in Eq. (6). 2 2 3 exp calc x −x X X X ijk n 3 3 6 ijk 7 F a ¼ arg min k¼1 4 5: j¼1 i¼1 6n

ð6Þ

The interaction parameters calculated using NRTL correlation and the values of root mean square deviation (RMSD) is given in Table 6. RMSD was used to calculate the goodness of the fit and is given in Eq. (7).

RMSD ¼

ð7Þ

> > ;

6n

Table 8 Values of the UNIQUAC parameters obtained from LLE data by regression at T = 303.2 K. Components (i-j)

8X X X  2 91 =2 exp calc > > > > x −x < = i i k j i > > :

where, x is the experimental or calculated mole fraction. Similarly subscripts i, j, and k represent the component, phase, and tie line, respectively. The value of n designates the number of components. The fitting of liquid-liquid equilibrium data via NRTL model for [C4mim][Phe], [C6mim][Phe], [C6mim][Phe] and [C10mim][Phe] is shown in Figs. 8, 9, 10 and 11. From all these figures and the negligibly small values of root mean square deviation given in Table 6 we can conclude that NRTL model gives an extremely good fit for all these ionic liquids.

Aij (K)

Aij (K)

Dodecane + Naphthenic Acid + [C4mim][Phe] 1–2 -1016.66 -257.52 1–3 -2254.06 1991.75 2–3 -1007.87 1445.84 Dodecane + Naphthenic Acid + [C6mim][Phe] 1–2 1757.10 -2226.11 1–3 -2225.14 6538.58 2–3 -4615.47 3194.36

Table 7 UNIQUAC volume and surface area parameters for all the components. Component

V (Å3)

A (Å2)

r

q

Dodecane Naphthenic Acid [C4mim][Phe] [C6mim][Phe] [C8mim][Phe] [C10mim][Phe]

110.894 216.103 246.171 273.041 300.793

138.648 188.202 202.593 222.588 238.983

8.5462 [52] 4.323 [53] 8.5800 9.7738 10.8406 11.9425

7.096 [52] 3.344 [53] 4.5342 4.8809 5.3626 5.7576

Dodecane + Naphthenic Acid + [C8mim][Phe] 1–2 -650.62 -888.37 1–3 -2245.39 2256.47 2–3 1406.05 -935.46 Dodecane + Naphthenic Acid + [C10mim][Phe] 1–2 -1021.1481 241.2558 1–3 -2155.958 1329.9191 2–3 1372.7850 -874.0472 a b

F = calculated via Eq. (6). RMSD = calculated via Eq. (7).

Fa

RMSDb

3.744 × 10-01

2.891 × 10-02

3.716 × 10-01

2.835 × 10-02

4.009 × 10-01

2.901 × 10-02

4.519 × 10-01

2.843 × 10-02

522

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

Fig. 12. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C4mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by UNIQUAC model.

It is found that low values for root mean square deviation and the virtually identical experimental and calculated data observed in the triangular diagrams give an idea of the goodness of the NRTL model. The

interaction parameters calculated using NRTL can be used to calculate the enthalpy of the mixture as well as to calculate the number of stages for any given concentration of the feed in ASPEN-HYSYS.

Fig. 13. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C6mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by UNIQUAC model.

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

3.5. UNIQUAC volume and surface area structural parameters For the calculation of r and q, we have used the simple relationships:





 3 V 1  108 NA V vw

 2 A 1  108 NA Avw

ð8Þ

523

rx ∅i ¼ Xn i i rx j¼1 j j

ð13Þ

qx θi ¼ Xn i i qx j¼1 j j

ð14Þ

  −Δuij : τij ¼ exp RT

ð15Þ

ð9Þ

where, NA is the Avogadro's number (6.023 × 1023 mol−1). For the standard segment area AVW (2.5 × 109 cm2/mol) and volume VVW (15.17 cm3/mol), values were taken from Bondi [49]. The value of r and q is shown in Table 7. 3.6. UNIQUAC model The UNIQUAC model [46] gives the following equation for the nonideal activity coefficient for component i g E g comb g res ¼ þ RT RT RT

ð10Þ

Xn g comb Xn ∅ θ ¼ x ln i þ 5 i¼1 qi xi ln i i¼1 i RT xi ∅i

ð11Þ

Xn  Xn g res ¼ − i¼1 qi xi ln θ j τ ji j¼1 RT

ð12Þ

In the above equation qi and ri represents the surface area and relative volume fractions for component i. Also θi, Φi are surface area fractions and volume fractions for component i. The binary interaction parameters (Δuij = τij) are calculated from the experimental data by minimizing the objective function given in Eq. (6). Similarly the root mean square deviation was calculated using Eq. (7). The binary interaction for quaternary phases as well as the objective function and root mean square for all these four ILs can be found in Table 8. The fitting of LLE data via UNIQUAC model for [C4mim][Phe], [C6mim][Phe], [C8mim][Phe] and [C10mim][Phe] is shown in Figs. 12, 13, 14, and 15. From all these figures it is obvious that the fitting of LLE data via UNIQUAC model seems to be overestimated at higher concentrations of naphthenic acid as compared to lower concentration of naphthenic acid. Similarly the deviation between experimental and modelling values is more obvious in the hydrocarbon phase as compared to the ionic liquid phase. These results are in agreement with the previous reported literature studies [39]. These interaction parameters can be used in ASPEN-HYSYS simulators for designing of extraction column. The relatively higher values for square deviation and the difference in experimental and calculated data observed in the triangular

Fig. 14. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C8mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by UNIQUAC model.

524

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525

Fig. 15. Comparison of liquid-liquid equilibrium plot for the ternary system Dodecane (1) + Naphthenic Acid + [C10mim][Phe]. Solid lines and square indicates experimental tie lines and dashed lines and upper triangle indicate calculated data by UNIQUAC model.

diagrams indicates that NRTL model gives a better fit as compared to the UNIQUAC model. This is congruent with review of these thermodynamic techniques [51]. Likewise it is also obvious from previous literature reports that for LLE modelling having ionic liquids NRTL model gives a better fit as compared to UNIQUAC model [50]. 4. Conclusions LLE data for the system of Dodecane, NA and [Cnmim][Phe] (n = 4, 6, 8, 10) at different concentrations of naphthenic acid in the feed were determined experimentally at 303.2 K and atmospheric pressure. The solubility of naphthenic acid and dodecane are higher when the ILs contains longer alkyl chain in the imidazolium ring, which increases the distribution ratio of naphthenic acid. The experiments with [C10mim][Phe] IL have shown good results in the separation of naphthenic acid from dodecane comparing with the distribution ratio in the whole range of composition. Hence, they could be considered as suitable solvent for liquid-liquid extraction of naphthenic acid from dodecane. Moreover, the absence of IL in the raffinate phase may improve the liquid-liquid extraction process by avoiding the purification step for recovering the solvent from the raffinate phase. The NRTL model was found to be more effective as compared to the UNIQUAC model in correlating the experimental LLE data for the studied ternary systems. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.molliq.2016.03.053. References [1] N.V. Plechkova, K.R. Seddon, Applications of ionic liquids in the chemical industry, Chem. Soc. Rev. 37 (2008) 123–150.

[2] K.C. Lethesh, S.N. Shah, M.A. Mutalib, Synthesis, characterization, physical and thermodynamic properties of diazobicyclo undecene based dicyanamide ionic liquids, J. Mol. Liq. 208 (2015) 253–258. [3] C.M. Gordon, New developments in catalysis using ionic liquids, Appl. Catal. A Gen. 222 (2001) 101–117. [4] J. Huddleston, R. Rogers, Room temperature ionic liquids as novel media for ‘clean’ liquid–liquid extraction, Chem. Commun. (1998) 1765–1766 (DOI). [5] T. Welton, Ionic liquids in catalysis, Coord. Chem. Rev. 248 (2004) 2459–2477. [6] B. Wu, R. Reddy, R. Rogers, Novel ionic liquid thermal storage for solar thermal electric power systems, Sol. Eng. (2001) 445–452 (DOI). [7] L. Chen, M. Sharifzadeh, N. Mac Dowell, T. Welton, N. Shah, J.P. Hallett, Inexpensive ionic liquids:[HSO4]−-based solvent production at bulk scale, Green Chem. 16 (2014) 3098–3106. [8] 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. [9] P.P. Alvisi, V.F. Lins, An overview of naphthenic acid corrosion in a vacuum distillation plant, Eng. Fail. Anal. 18 (2011) 1403–1406. [10] J.A. Brient, P.J. Wessner, M.N. Doyle, Naphthenic acids, Kirk-Othmer Encycl. Chem. Technol. (1995) (DOI). [11] T.P. Fan, Characterization of naphthenic acids in petroleum by fast atom bombardment mass spectrometry, Energy Fuel 5 (1991) 371–375. [12] J. Schmitter, P. Arpino, G. Guiochon, Investigation of high-molecular-weight carboxylic acids in petroleum by different combinations of chromatography (gas and liquid) and mass spectrometry (electron impact and chemical ionization), J. Chromatogr. A 167 (1978) 149–158. [13] C.S. Hsu, G. Dechert, W. Robbins, E. Fukuda, Naphthenic acids in crude oils characterized by mass spectrometry, Energy Fuel 14 (2000) 217–223. [14] W.E. Rudzinski, L. Oehlers, Y. Zhang, B. Najera, Tandem mass spectrometric characterization of commercial naphthenic acids and a Maya crude oil, Energy Fuel 16 (2002) 1178–1185. [15] B.H. Ballinger, G. Sartori, D.W. Savage, Process for Neutralization of Petroleum Acids, Google Patents, 1997. [16] L. Ding, P. Rahimi, R. Hawkins, S. Bhatt, Y. Shi, Naphthenic acid removal from heavy oils on alkaline earth-metal oxides and ZnO catalysts, Appl. Catal. A Gen. 371 (2009) 121–130. [17] V. Gaikar, D. Maiti, Adsorptive recovery of naphthenic acids using ion-exchange resins, React. Funct. Polym. 31 (1996) 155–164. [18] Y. Wang, Z. Chu, B. Qiu, C. Liu, Y. Zhang, Removal of naphthenic acids from a vacuum fraction oil with an ammonia solution of ethylene glycol, Fuel 85 (2006) 2489–2493. [19] Y.Z. Wang, X.Y. Sun, Y.P. Liu, C.G. Liu, Removal of naphthenic acids from a diesel fuel by esterification, Energy Fuel 21 (2007) 941–943. [20] Y.-z. Wang, J.-y. Li, X.-y. Sun, H.-l. Duan, C.-m. Song, M.-m. Zhang, Y.-p. Liu, Removal of naphthenic acids from crude oils by fixed-bed catalytic esterification, Fuel 116 (2014) 723–728.

S.N. Shah et al. / Journal of Molecular Liquids 219 (2016) 513–525 [21] A. Zhang, Q. Ma, K. Wang, X. Liu, P. Shuler, Y. Tang, Naphthenic acid removal from crude oil through catalytic decarboxylation on magnesium oxide, Appl. Catal. A Gen. 303 (2006) 103–109. [22] K. Anderson, M. Atkins, P. Goodrich, C. Hardacre, A. Hussain, R. Pilus, D. Rooney, Naphthenic acid extraction and speciation from Doba crude oil using carbonatebased ionic liquids, Fuel (2015) (DOI). [23] K. Anderson, P. Goodrich, C. Hardacre, A. Hussain, D. Rooney, D. Wassell, Removal of naphthenic acids from crude oil using amino acid ionic liquids, Fuel 108 (2013) 715–722. [24] L.J. Shi, B.X. Shen, G.Q. Wang, Removal of naphthenic acids from Beijiang crude oil by forming ionic liquids, Energy Fuel 22 (2008) 4177–4181. [25] S. Nasir Shah, L. Kallidanthiyil Chellappan, G. Gonfa, M.I.A. Mutalib, R.B.M. Pilus, M.A. Bustam, Extraction of naphthenic acid from highly acidic oil using phenolate based ionic liquids, Chem. Eng. J. 284 (2016) 487–493. [26] S. Nasir Shah, M.I.A. Mutalib, R.B.M. Pilus, K.C. Lethesh, Extraction of naphthenic acid from highly acidic oil using hydroxide-based ionic liquids, Energy Fuel 29 (2014) 106–111. [27] S.N. Shah, K.C. Lethesh, M. Abdul Mutalib, M. Pilus, R. Binti, Extraction and recovery of naphthenic acid from acidic oil using Supported Ionic Liquid Phases (SILPs), Chem. Prod. Process. Model. 10 (2015) 221–228. [28] N.A. Manan, M.P. Atkins, J. Jacquemin, C. Hardacre, D.W. Rooney, Phase equilibria of binary and ternary systems containing ILs, dodecane, and cyclohexanecarboxylic acid, Sep. Sci. Technol. 47 (2012) 312–324. [29] P. Su, H. Li, Continuous and smooth potential energy surface for conductorlike screening solvation model using fixed points with variable areas, J. Chem. Phys. 130 (2009) 074109. [30] S. Nasir Shah, L. Kallidanthiyil Chellappan, M.I.A. Mutalib, R.B.M. Pilus, Evaluation of thermophysical properties of imidazolium based phenolate ionic liquids, Ind. Eng. Chem. Res. (2015) (DOI). [31] F. Weigend, R. Ahlrichs, Balanced basis sets of split valence, triple zeta valence and quadruple zeta valence quality for H to Rn: design and assessment of accuracy, Phys. Chem. Chem. Phys. 7 (2005) 3297–3305. [32] J. Zheng, X. Xu, D.G. Truhlar, Minimally augmented Karlsruhe basis sets, Theor. Chem. Accounts 128 (2011) 295–305. [33] F. Neese, The ORCA program system, Wiley Interdisciplinary Reviews: Computational Molecular Science 2 (2012) 73–78. [34] E.F. Valeev, The LIBINT Programmer's Manual, 2013, http://download2.nust.na/ pub7/sourceforge/l/project/li/libint/libint-for-beginners/progman-2.0.3-stable.pdf. [35] S. Grimme, S. Ehrlich, L. Goerigk, Effect of the damping function in dispersion corrected density functional theory, J. Comput. Chem. 32 (2011) 1456–1465. [36] S. Grimme, J. Antony, S. Ehrlich, H. Krieg, A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu, J. Chem. Phys. 132 (2010) 154104. [37] S. Miertuš, E. Scrocco, J. Tomasi, Electrostatic interaction of a solute with a continuum. A direct utilizaion of AB initio molecular potentials for the prevision of solvent effects, Chem. Phys. 55 (1981) 117–129.

525

[38] A.A. Granovsky, Firefly Version 8, www http://classic.chem.msu.su/gran/firefly/ index.html, (DOI) [39] C. Manohar, D. Rabari, A.A.P. Kumar, T. Banerjee, K. Mohanty, Liquid–liquid equilibria studies on ammonium and phosphonium based ionic liquid–aromatic–aliphatic component at T = 298.15 K and p = 1 bar: correlations and a-priori predictions, Fluid Phase Equilib. 360 (2013) 392–400. [40] D.R. Lide, CRC Handbook of Chemistry and Physics, CRC Press, 2004. [41] F.M. Holowenko, M.D. MacKinnon, P.M. Fedorak, Characterization of naphthenic acids in oil sands wastewaters by gas chromatography-mass spectrometry, Water Res. 36 (2002) 2843–2855. [42] Y. Sun, L. Shi, Basic ionic liquids with imidazole anion: new reagents to remove naphthenic acids from crude oil with high total acid number, Fuel 99 (2012) 83–87. [43] J. Duan, Y. Sun, L. Shi, Three different types of heterocycle of nitrogen-containing alkaline ionic liquids treatment of acid oil to remove naphthenic acids, Catal. Today 212 (2013) 180–185. [44] S. García, M. Larriba, J.N. García, J.S. Torrecilla, F. Rodríguez, Liquid − liquid extraction of toluene from heptane using 1-alkyl-3-methylimidazolium bis (trifluoromethylsulfonyl) imide ionic liquids, J. Chem. Eng. Data 56 (2010) 113–118. [45] H. Renon, J.M. Prausnitz, Local compositions in thermodynamic excess functions for liquid mixtures, AICHE J. 14 (1968) 135–144. [46] D.S. Abrams, J.M. Prausnitz, Statistical thermodynamics of liquid mixtures: a new expression for the excess Gibbs energy of partly or completely miscible systems, AICHE J. 21 (1975) 116–128. [47] L.D. Simoni, Y. Lin, J.F. Brennecke, M.A. Stadtherr, Modeling liquid-liquid equilibrium of ionic liquid systems with NRTL, electrolyte-NRTL, and UNIQUAC, Ind. Eng. Chem. Res. 47 (2008) 256–272. [48] J. Sørensen, W. Arlt, Liquid− Liquid Equilibrium Data Collection, DECHEMA, Frankfurt/Main, Germany, 1979− 1980 (There is no corresponding record for this reference, DOI.). [49] A. Bondi, van der Waals volumes and radii, J. Phys. Chem. 68 (1964) 441–451. [50] U.K. Ravilla, T. Banerjee, Liquid liquid equilibria of imidazolium based ionic liquid + pyridine + hydrocarbon at 298.15 K: experiments and correlations, Fluid Phase Equilib. 324 (2012) 17–27. [51] G.M. Kontogeorgis, P. Coutsikos, Thirty years with EoS/GE models - what have we learned? Ind. Eng. Chem. Res. 51 (2012) 4119–4142. [52] R.S. Santiago, G.R. Santos, M. Aznar, UNIQUAC correlation of liquid–liquid equilibrium in systems involving ionic liquids: the DFT–PCM approach, Fluid Phase Equilib. 278 (2009) 54–61. [53] T. Banerjee, M.K. Singh, R.K. Sahoo, A. Khanna, Volume, surface and UNIQUAC interaction parameters for imidazolium based ionic liquids via polarizable continuum model, Fluid Phase Equilib. 234 (2005) 64–76.