Measurement, correlation, and prediction of vapor pressure for binary and ternary systems containing an alkylsulfate-based ionic liquid

Fluid Phase Equilibria 397 (2015) 58–67

Contents lists available at ScienceDirect

Fluid Phase Equilibria j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / fl u i d

Measurement, correlation, and prediction of vapor pressure for binary and ternary systems containing an alkylsulfate-based ionic liquid Yafen Dai, Yixin Qu, Shui Wang, Jidong Wang * Beijing Key Laboratory of Membrane Science and Technology and College of Chemical Engineering, Beijing University of Chemical Technology, Beijing 100029, PR China

A R T I C L E I N F O

A B S T R A C T

Article history: Received 20 December 2014 Received in revised form 16 March 2015 Accepted 30 March 2015 Available online 31 March 2015

Vapor pressure data for water, 1-propanol, 2-propanol, as well as the mixtures of {water + 1-propanol} and {water + 2-propanol}, were measured by a quasi-static ebulliometric method, in the presence of an alkylsulfate-based IL, namely, 1-ethyl-3-methylimidazolium methylsulfate ([EMIM][MS]) or 1-ethyl-3methylimidazolium ethylsulfate ([EMIM][ES]). The experimental vapor pressure data for binary systems containing IL were correlated by NRTL model with an overall relative root mean square deviation (rRMSD) of 0.0053, and the obtained binary NRTL parameters were employed to predict the vapor pressure for two ternary systems with an overall rRMSD of 0.0196. In addition, isobaric vapor–liquid equilibria were predicted for the ternary systems containing [EMIM][MS], [EMIM][ES], 1,3-dimethylimidazolium methylsulfate ([MMIM][MS]), 1-ethyl-3-methylimidazolium tetrafluoroborate ([EMIM] [BF4]), and 1-ethyl-3-methylimidazolium trifluoromethanesulfonate [EMIM][OTf], respectively, with IL mole fraction of 0.05, 0.15, and 0.25 at 101.325 kPa. It was found that the addition of IL can enhance the relative volatility of propanol to water, and the separation ability follows the order of [MMIM] [MS] > [EMIM][MS] > [EMIM][ES] > [EMIM][BF4] > [EMIM][OTf], which was further explained at electronic level with quantum chemical calculation. Therefore, the azeotropic mixtures of {water + 1-propanol}, and {water + 2-propanol} might be separated effectively by the addition of the alkylsulfate-based ILs in extractive distillation. ã 2015 Elsevier B.V. All rights reserved.

Keywords: Vapor pressure Ionic liquid NRTL model Extractive distillation

1. Introduction In chemical industry, the separation of azeotropic or close boiling mixtures is usually realized with special distillation, such as extractive distillation and salt distillation, in which an entrainer is added into the mixtures for increasing relative volatility. However, the volatile organic solvents used in extractive distillation are likely to harm workers, pollute environment, and consume excessive energy in solvent recycle process, while the inorganic salts used in salt distillation may cause the problems of pipeline blockage and equipment corrosion. Ionic liquids (ILs) are prospective alternatives to conventional solvents in special distillation [1], since several ILs can combine the advantages of both organic solvents and inorganic salts, showing the favorable properties of negligible volatility, good stability, easy recyclability, non-flammability, and structural tenability [2].

* Corresponding author. Tel.: +86 10 64454730. E-mail address: [email protected] (Y. Dai). http://dx.doi.org/10.1016/j.fluid.2015.03.043 0378-3812/ ã 2015 Elsevier B.V. All rights reserved.

Vapor–liquid equilibria (VLE) data are the foundation of process design and equipment selection in distillation. Up to now, many researchers have studied the VLE behavior for systems containing ILs, aiming to screen suitable ILs as entrainers in extractive distillation to enhance separation efficiency [3]. Among those ILs, imidazoliumbased ILs are widely investigated, including the ILs with waterunstable fluorinated anions like tetrafluoroborate ([BF4]
) and hexafluorophosphate ([PF6]
), which can release toxic hydrogen fluoride under aqueous environment [4], and some expensive ILs with the anion of bis(trifluromethanesulfonyl)imide ([NTf2]
) or trifluoromethanesulfonate ([OTf]
), which seriously limited their technical application in large scale. Compared with those flawed ILs as entrainers, the hydrophilic ILs with alkylsulfate and dialkylphosphate anions have shown great potential in extractive distillation, owing to their low-viscosity, non-toxicity, and reasonable price [5,6]. Several works have been done to determine the VLE data for systems containing alkylsulfate- and dialkylphosphate-based ILs. Li et al. measured the vapor pressure for binary and ternary systems containing dialkylphosphate-based ILs, including 1-methyl-3methylimidazolium dimethylphosphate ([MMIM][DMP]),

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

59

Table 1 Specifications of chemical samples. Chemicals

Source

Water

Electrical conductivity < 5 mS cm
1 (298.15 K) Guangfu Chemical Regent Co., Tianjin, China Mass fraction > 0.997 Guangfu Chemical Regent Co., Tianjin, China Mass fraction > 0.997 Synthesized in the laboratory Mole fraction > 0.990

4A molecular sieves GCb 4A molecular sieves GCb[EMIM][MS] 1 Distillation H NMR, 13C NMRc

Synthesized in the laboratory

Distillation

1-Propanol 2-Propanol [EMIM] [MS] [EMIM][ES] a b c

Purity

Purification method Analysis method

Distilled in the laboratory

Mole fraction > 0.990

Electrical conductivity measurementa

1

H NMR, 13C NMRc

Measured at 298.15 K with a conductivity meter (type EC-215, Hanna Co., Italy). Checked with a gas chromatography (type GC2010, SHIMADZU Co., Japan) equipped with a FID detector and FFAP capillary column. Recorded on a 400.0 MHz NMR spectrometer (Bruker Co., Germany) at T = 300 K using deuterated water (D2O) as the external reference solvent.

1-ethyl-3-methylimidazolium dimethylphosphate ([EMIM][DMP]), 1-ethyl-3-ethylimidazolium diethylphosphate ([EEIM][DEP]), 1ethyl-3-methylimidazolium diethylphosphate ([EMIM][DEP]), and 1-butyl-3-emthylimidazolium dibutylphosphate ([BMIM][DBP]). Their results indicated that the addition of ILs can change the relative volatility of a solvent, and that ILs can exhibit “ionic” or “molecule” character in different solvents [7–9]. On the other hand, 1-ethyl-3-methylimidazolium ethylsulfate ([EMIM][ES]), one of the first commercial “bulk” ionic liquids (available on a ton-scale), was reported to have the ability to break the azeotrope of water–ethanol mixture with sufficient IL-content [10,11]. In our previous work, the effects of 1-methyl-3-methylimidazolium methylsulfate [MMIM][MS] on the vapor pressure of water, Table 2 Experimental and calculated vapor pressure data for the binary system {water (1) + [EMIM][MS] (2)}.a T/K

Pexp/kPa

Pcal/kPa

g 1exp

g 1cal

x1 = 0.9911 325.47 332.85 339.46 347.77 353.71 360.69 367.95 373.32

13.58 19.31 26.10 37.35 47.73 62.89 82.80 100.79

13.68 19.45 26.25 37.57 47.97 63.18 83.09 100.94

0.9910 0.9915 0.9929 0.9932 0.9940 0.9946 0.9957 0.9977

x1 = 0.9801 324.84 332.78 341.06 347.98 355.55 362.67 368.80 373.91

12.91 18.88 27.44 36.97 50.42 66.52 83.60 100.75

13.06 19.09 27.75 37.35 50.91 67.20 84.49 101.65

x1 = 0.9659 324.90 334.35 341.79 350.10 355.56 363.58 369.47 374.66

12.61 19.81 27.62 39.34 49.12 67.17 83.67 100.80

x1 = 0.9182 332.13 339.27 344.26 349.55 357.67 364.82 371.71 376.98

16.38 22.75 28.31 35.37 49.22 64.72 83.58 100.94

a

None

1-propanol, 2-propanol, {water + 1-propanol}, and {water + 2propanol} were studied [12]. In order to discuss the influence of alkylsulfate-based ILs with various anion structures on the VLE behavior of {water + propanol} mixtures, herein, we shifted the study object to 1-ethyl-3-methylimidazolium methylsulfate ([EMIM][MS]) and 1-ethyl-3-methylimidazolium ethylsulfate ([EMIM][ES]). Thus, the vapor pressure data for water, 1-propanol, and 2-propanol, as well as the mixtures of {water + 1-propanol} and {water + 2-propanol}, were measured in the presence of [EMIM][MS] or [EMIM][ES] at varying IL-content and temperature. The experimental data for binary systems were correlated by NRTL model, and the obtained binary model parameters were used to predict the vapor pressure of the ternary systems of {water + 1-

Table 3 Experimental and calculated vapor pressure data for the binary system {water (1) + [EMIM][ES] (2)}.a T/K

Pexp/kPa

Pcal/kPa

g 1exp

g 1cal

0.9987 0.9988 0.9989 0.9990 0.9990 0.9991 0.9991 0.9992

x1 = 0.9916 323.87 331.50 338.66 347.39 353.98 361.95 368.29 372.77

12.56 18.16 25.20 36.83 48.34 66.15 83.99 99.02

12.66 18.28 25.35 37.01 48.54 66.38 84.21 99.04

0.9918 0.9929 0.9934 0.9942 0.9952 0.9958 0.9968 0.9992

0.9991 0.9992 0.9992 0.9993 0.9993 0.9994 0.9994 0.9994

0.9831 0.9837 0.9837 0.9853 0.9861 0.9857 0.9854 0.9873

0.9939 0.9944 0.9948 0.9952 0.9955 0.9957 0.9960 0.9961

x1 = 0.9813 323.35 333.94 339.07 347.50 354.15 362.54 368.78 373.25

12.07 20.07 25.31 36.48 47.98 66.72 84.30 99.07

12.17 20.20 25.48 36.70 48.25 67.03 84.65 99.49

0.9872 0.9885 0.9880 0.9884 0.9881 0.9894 0.9913 0.9913

0.9959 0.9962 0.9964 0.9967 0.9968 0.9970 0.9972 0.9973

12.77 20.03 27.98 39.87 49.78 68.05 84.75 102.14

0.9713 0.9739 0.9732 0.9739 0.9744 0.9756 0.9762 0.9763

0.9834 0.9849 0.9859 0.9869 0.9875 0.9883 0.9888 0.9893

x1 = 0.9684 325.26 334.56 340.04 349.33 355.44 362.84 369.46 373.88

12.97 20.22 25.84 38.53 49.40 66.02 84.53 99.14

13.10 20.37 26.10 38.87 49.87 66.57 85.21 99.93

0.9789 0.9822 0.9805 0.9824 0.9821 0.9837 0.9836 0.9843

0.9889 0.9898 0.9903 0.9910 0.9914 0.9919 0.9923 0.9926

16.21 22.53 28.10 35.22 49.08 64.82 83.73 101.12

0.9390 0.9421 0.9428 0.9425 0.9449 0.9437 0.9461 0.9480

0.9291 0.9332 0.9358 0.9384 0.9421 0.9451 0.9478 0.9497

x1 = 0.9292 327.81 335.28 341.78 349.88 356.81 363.92 371.26 375.70

13.63 19.37 25.91 36.66 48.60 64.01 84.05 98.70

13.72 19.52 26.16 37.01 49.07 64.72 85.01 99.70

0.9438 0.9420 0.9433 0.9441 0.9447 0.9470 0.9511 0.9550

0.9537 0.9564 0.9585 0.9610 0.9629 0.9648 0.9665 0.9675

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

a

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

60

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

propanol + IL} and {water + 2-propanol + IL}. Ultimately, the isobaric VLE data for the ternary system containing [EMIM][MS] or [EMIM][ES] with IL mole fraction of 0.05, 0.15, and 0.25 at 101.325 kPa were calculated to assess the separation ability for the azeotropic systems of {water + 1-propanol} and {water + 2-propanol} by extractive distillation. 2. Materials and methods

described in the literature [5]. The purity of ILs was estimated as x  99% in terms of the 1H and 13C NMR spectra analysis (see Supplementary material). Before use, the ILs were treated by vacuum evaporation at 343.15 K over 12 h to remove the possible volatile impurities. The water content was less than 500 ppm in mass fraction as measured by a Karl–Fischer titrator (AKF-1B, Shanghai Scientific Instrument Co., Shanghai, China). The specifications and analysis methods for the chemicals used in VLE measurement are summarized in Table 1.

2.1. Materials 2.2. Apparatus and procedure Water was double distilled with an electrical conductivity less than 5 mS cm
1at 298.15 K. Analytical reagent grade 1-propanol and 2-propanol with a purity of w  99.7% were purchased from Guangfu Chemical Regent Co., Tianjin, China, and were both dried over 4A molecular sieves prior to use. Gas chromatography (GC) failed to show any significant impurities in the reagents of 1propanol and 2-propanol. 1-Methylimidazole and 1-ethylimidazole (Chengjie Chemical Co., Shanghai, China) with a purity of w  99% was redistilled for purification. Dimethyl sulfate (Xiya Chemical Reagent Co., Shandong, China) with a purity of w  99% and diethyl sulfate (Sinopharm Chemical Reagent Co. Ltd., Shanghai, China) with a purity of w  99% were used as provided. The alkylsulfate-based ILs, i.e., [EMIM][MS] and [EMIM][ES], were synthesized in our laboratory according to the procedures

The vapor pressure was measured by a quasi-static ebulliometric method. The details of the experimental apparatus and the measurement procedures were described elsewhere [7]. The apparatus consisted of two parallel inclined ebulliometers, which were filled with IL-containing solution and a reference material (water) separately. During the experiment, the bubble-point temperatures were measured under the same pressure, which could be calculated from the boiling temperature of water according to the corresponding pressure–temperature relation (e.g., Antoine equation). The temperatures inside the two ebulliometers were measured with two 25 V standard platinum resistance thermometers (type CST6501) connected to a twochannel standard digital indicator (type CST6502). The

Table 4 Experimental and calculated vapor pressure data for the binary system {1-propanol (1) + [EMIM][MS] (2)}.a

Table 5 Experimental and calculated vapor pressure data for the binary system {1-propanol (1) + [EMIM][ES] (2)}.a

T/K

Pexp/kPa

Pcal/kPa

g 1exp

g 1cal

T/K

x1 = 0.9703 323.16 332.53 339.79 347.06 352.84 359.75 366.83 370.46

11.85 19.23 27.34 38.16 49.14 65.60 87.02 100.07

11.89 19.27 27.37 38.18 49.16 65.59 86.98 100.00

1.0052 1.0065 1.0074 1.0078 1.0079 1.0082 1.0085 1.0087

1.0087 1.0086 1.0085 1.0084 1.0083 1.0082 1.0080 1.0080

x1 = 0.9357 326.23 334.17 339.00 346.95 353.61 360.52 366.87 371.05

13.67 20.45 25.82 37.20 49.74 66.30 85.31 100.18

13.66 20.42 25.77 37.15 49.68 66.20 85.20 100.02

1.0228 1.0237 1.0243 1.0239 1.0239 1.0240 1.0238 1.0242

x1 = 0.8946 327.00 333.46 339.78 348.00 354.18 361.07 367.43 371.91

13.77 19.13 25.92 37.78 49.41 65.67 85.01 100.32

13.78 19.12 25.95 37.78 49.43 65.77 84.64 100.44

x1 = 0.7845 328.26 337.59 342.65 350.47 356.78 364.51 370.21 374.40

13.23 21.09 26.78 38.08 49.90 68.34 85.37 100.08

13.21 21.06 26.75 38.08 49.93 68.48 85.55 100.23

a

Pexp/kPa

Pcal/kPa

x1 = 0.9721 324.98 333.05 338.57 346.76 352.95 359.90 366.25 370.39

13.01 19.64 25.68 37.44 49.10 65.67 84.68 99.38

1.0220 1.0222 1.0223 1.0225 1.0225 1.0226 1.0226 1.0225

x1 = 0.9393 327.82 334.68 339.56 347.15 354.93 360.05 366.75 371.16

1.0355 1.0374 1.0364 1.0380 1.0381 1.0376 1.0441 1.0386

1.0360 1.0368 1.0375 1.0382 1.0387 1.0392 1.0396 1.0398

1.0633 1.0668 1.0681 1.0703 1.0718 1.0729 1.0745 1.0765

1.0609 1.0651 1.0672 1.0702 1.0725 1.0751 1.0768 1.0781

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

g 1exp

g 1cal

13.06 19.72 25.77 37.59 49.29 65.92 84.92 99.62

1.0004 1.0001 1.0009 1.0005 1.0008 1.0011 1.0022 1.0028

1.0038 1.0041 1.0042 1.0045 1.0047 1.0049 1.0051 1.0052

14.74 20.81 26.31 37.28 52.27 64.62 84.44 100.09

14.68 20.72 26.19 37.12 52.07 64.40 84.13 99.67

1.0111 1.0123 1.0125 1.0131 1.0133 1.0132 1.0140 1.0150

1.0071 1.0077 1.0081 1.0088 1.0094 1.0098 1.0103 1.0107

x1 = 0.9002 326.77 335.68 340.87 349.00 354.47 361.89 367.74 372.37

13.34 20.88 26.74 38.70 49.02 66.51 83.84 100.20

13.34 20.90 26.77 38.75 49.12 66.77 84.16 100.45

1.0084 1.0090 1.0098 1.0111 1.0111 1.0101 1.0112 1.0130

1.0088 1.0102 1.0110 1.0123 1.0131 1.0142 1.0150 1.0156

x1 = 0.7946 328.42 337.14 343.15 351.19 357.43 364.46 371.09 375.29

12.80 19.77 26.26 37.63 49.15 65.49 84.89 99.46

12.77 19.75 26.24 37.67 49.18 65.53 84.89 99.39

1.0069 1.0091 1.0109 1.0121 1.0149 1.0174 1.0202 1.0223

1.0047 1.0080 1.0103 1.0132 1.0155 1.0179 1.0203 1.0217

a

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

temperature measurement system was pre-calibrated by the National Institute of Metrology, China, with an uncertainty of 0.02 K. The uncertainty of pressure originating from temperature measurement was estimated within 0.04 kPa. The vapor phase condensers were cooled with glycol aqueous solution below 275 K to minimize the vapor loss to avoid the concentration variation of the solution. The reliability of the apparatus was examined in our previous work [12]. The average absolute relative deviation for pressure measurement was 0.04% and the maximum absolute relative deviation was 0.06%. All sample solutions with different compositions were prepared gravimetrically with a digital balance (type CP224S, Sartorius Co., Germany), having a readability of 0.0001 g. The uncertainty of the mole fraction of mixtures was estimated as 0.0001. 2.3. Quantum chemical calculations The geometries for the anions of ILs were optimized in gas phase with the density functional theory (DFT) method at B3LYP/631++G** level by Gaussion 09 package [13]. All the optimized geometries were verified as minima with no imaginary frequency by full calculation of the Hessian and a harmonic frequency analysis. The detailed coordinates for the optimized geometries can be found in Supplementary material. The natural population analysis (NPA) of the atomic charges on the anions was calculated by NBO 3.1 program as implemented in the Gaussion 09 program.

Table 6 Experimental and calculated vapor pressure data for the binary system {2-propanol (1) + [EMIM][MS] (2)}.a

3. Results and discussion 3.1. Binary vapor pressure Vapor pressure data for six binary systems with IL-content w2 = 0.10, 0.20, 0.30, and 0.50 were carried out within the temperature range from 312.65 to 376.98 K. The binary systems here were {water (1) + [EMIM][MS] (2)}, {water (1) + [EMIM][ES] (2)}, {1-propanol (1) + [EMIM][MS] (2)}, {1-propanol (1) + [EMIM] [ES] (2)}, {2-propanol (1) + [EMIM][MS] (2)}, and {2-propanol (1) + [EMIM][ES] (2)}. The measured data for each system are listed in Tables 2–7, respectively, where the IL-contents are expressed in mole fraction. When a system has achieved vapor–liquid equilibrium, the vapor phase composition of component i can be calculated by the following equation: " # l s ^ v ¼ Ps fs g x exp V i ðP
Pi Þ Pyi f (1) i i i i i RT where P is the vapor pressure at system temperature T, and Psi represents the corresponding saturated vapor pressure of component i. yi and xi denote the mole fraction of component i in vapor phase and liquid phase, respectively. gi represents the activity ^ v is the fugacity coefficient of coefficient of component i. f i

Table 7 Experimental and calculated vapor pressure data for the binary system {2-propanol (1) + [EMIM][ES] (2)}.a T/K

Pexp/kPa

Pcal/kPa

g 1exp

g 1cal

1.0098 1.0097 1.0096 1.0095 1.0094 1.0092 1.0091 1.0089

x1 = 0.9725 313.61 323.19 327.63 333.75 339.19 345.60 351.50 355.22

13.76 22.91 28.66 38.50 49.49 65.71 84.26 97.98

13.83 23.01 28.76 38.61 49.58 65.74 84.25 97.95

0.9959 0.9968 0.9979 0.9983 0.9995 1.0008 1.0013 1.0015

1.0013 1.0013 1.0013 1.0013 1.0013 1.0012 1.0012 1.0012

1.0261 1.0274 1.0276 1.0284 1.0288 1.0266 1.0298 1.0290

1.0252 1.0254 1.0256 1.0257 1.0257 1.0257 1.0257 1.0257

x1 = 0.9402 312.65 320.18 325.91 333.73 339.44 345.82 351.90 355.83

12.66 19.02 25.53 37.31 48.55 64.25 82.95 97.30

12.73 19.11 25.62 37.42 48.66 64.39 83.08 97.37

0.9996 0.9999 1.0014 1.0019 1.0024 1.0025 1.0030 1.0040

1.0046 1.0047 1.0047 1.0047 1.0047 1.0047 1.0046 1.0046

16.26 21.22 27.85 37.27 47.79 62.36 82.77 101.23

1.0422 1.0416 1.0431 1.0441 1.0450 1.0454 1.0458 1.0463

1.0416 1.0423 1.0430 1.0436 1.0441 1.0446 1.0451 1.0454

x1 = 0.9019 316.55 322.37 326.62 334.48 339.56 346.62 352.17 356.47

15.17 20.62 25.56 37.42 47.27 64.46 81.34 96.82

15.18 20.62 25.56 37.35 47.13 64.21 80.93 96.25

1.0083 1.0091 1.0089 1.0111 1.0124 1.0134 1.0143 1.0153

1.0085 1.0088 1.0090 1.0092 1.0093 1.0095 1.0092 1.0093

13.17 17.30 25.65 35.23 47.21 61.65 85.76 102.00

1.0736 1.0746 1.0759 1.0768 1.0784 1.0803 1.0815 1.0837

1.0677 1.0706 1.0747 1.0779 1.0809 1.0834 1.0865 1.0881

x1 = 0.7973 314.65 322.92 329.30 336.98 343.40 349.90 355.69 360.16

12.10 18.76 25.82 37.13 49.61 65.61 83.22 99.43

12.07 18.76 25.85 37.23 49.69 65.67 83.24 99.29

1.0086 1.0097 1.0103 1.0111 1.0139 1.0162 1.0180 1.0206

1.0061 1.0093 1.0114 1.0138 1.0155 1.0170 1.0183 1.0192

T/K

Pexp/kPa

Pcal/kPa

g 1exp

g 1cal

x1 = 0.9704 314.00 318.95 326.09 331.17 337.55 343.43 350.99 355.77

14.16 18.50 26.71 34.31 46.30 60.27 83.12 100.94

14.22 18.58 26.81 34.36 46.30 60.20 82.94 100.71

1.0050 1.0052 1.0057 1.0081 1.0094 1.0103 1.0112 1.0112

x1 = 0.9357 314.88 319.84 326.70 333.08 338.22 344.76 350.52 356.22

14.63 19.11 27.12 36.97 46.92 62.54 80.05 100.81

14.62 19.07 27.06 36.87 46.78 62.49 79.73 100.48

x1 = 0.8946 317.39 322.45 327.85 333.89 339.29 345.33 352.06 357.06

16.27 21.21 27.86 37.28 47.83 62.40 82.82 101.32

x1 = 0.7844 315.46 320.52 328.23 334.81 341.19 347.28 355.22 359.58

13.24 17.36 25.68 35.20 47.11 61.47 85.36 101.59

a

61

a

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa.

62

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

[(Fig._1)TD$IG]

Table 8 Antoine constans Ai,Bi, and Ci for compounds involved in this work [14]. Compound

Antoine constant Ai

Bi

Ci

Water 1-Propanol 2-Propanol

7.074056 6.97878 6.86634

1657.459 1497.734 1360.183


46.13
69.056
75.557

s

component i in the vapor phase, and fi is the fugacity coefficient of pure component i in its saturated state. The vapor pressure PSi of pure water, 1-propanol, and 2propanol can be calculated with Antoine equation: logðPsi =kPaÞ ¼ Ai


Bi T=K þ C i

(2)

where the Antoine constants, Ai, Bi, and Ci, for the concerning solvents are taken from reference [14] and listed in Table 8. For the binary systems {solvent (1) + IL (2)} in this work, the vapor pressure of IL, Ps2 , is assumed to be zero because of the negligible volatility of IL; consequently, the vapor phase composition is (y1 = 1, y2 = 0). Since the experimental pressures are low

Fig. 1. Experimental and correlated vapor pressure of binary system {water (1) + [EMIM][MS] (2)} at varying mass fraction of IL: &, w2 = 0.1; , w2 = 0.2; 4, w2 = 0.3; ^, w2 = 0.5. Legends: dot line (  ), pure water; solid line (—), calculated by NRTL equation.

[(Fig._2)TD$IG]

enough, the Poynting factor, exp½V l1 ðP1
Ps1 Þ=RT, is approximately equal to 1. Meanwhile, with the assumption that the vapor phase ^ v and fs , can be regarded as ideal gas, the fugacity coefficients, f 1

1

are treated as 1. Therefore, Eq. (1) can be simplified as P ¼ Ps1 g 1 x1

(3)

According to the above equations, the activity coefficients of the solvents could be calculated from binary vapor pressure data. The obtained results are also given in Tables 2–7, respectively. The activity coefficients of the solvents were calculated from Eq. (3); thus, the uncertainty of activity coefficient is arisen from the uncertainties of pressure, temperature, and liquid mole fraction, which are assumed as 0.04 kPa, 0.02 K, and 0.0001, respectively. Then, the deviations for activity coefficients d(g 10
g1) can be estimated as following:

dð g 1

0

P þ 0:04 P P
jþj
g1Þ ¼ j x1 Ps1 ðTÞ x1 Ps1 ðTÞ x1 Ps1 ðT þ 0:02Þ P P P jþj
j
x1 Ps1 ðTÞ ðx1 þ 0:0001ÞPs1 ðTÞ x1 Ps1 ðTÞ

(4)

where Ps1 ðT Þ refers to the corresponding Antoine equation for calculating the saturated vapor pressure of the solvents at Table 9 Binary NRTL parameters fitted here and taken from literature. System

a12

Water (1) + [EMIM][MS] (2)a Water (1) + [EMIM][ES] (2)a 1-Propanol (1) + [EMIM] [MS] (2)a 1-Propanol (1) + [EMIM] [ES] (2)a 2-Propanol (1) + [EMIM] [MS] (2)a 2-Propanol (1) + [EMIM] [ES] (2)a Water (1) + 1-propanol (2)b Water (1) + 2-propanol (2)b

0.8946 0.7424 0.5218

a b

Fitted in this work. Taken from Ref. [16].

g12
g22 (J mol
1)

g21
g11 (J mol
1)

rRMSD(P)

36.00


3771.69

0.0092

11.30 10401.13


3924.00
1670.89

0.0076 0.0016

0.6439

14753.34


1221.57

0.0030

0.4458

11404.65


2017.06

0.0026

0.2334

8889.00


5223.00

0.0029

0.4770 0.3000

7896.70 6900.81

1648.80 77.49

Fig. 2. Predicted activity coefficients of solvents (g1) at T = 343.15 K and varying IL mole fractions (x2) for binary systems {solvent (1) + IL (2)} with the IL: &, [MMIM] [MS]; , [EMIM][MS]; 4, [EMIM][ES]. Legends: solid line (—), water; dash line ( [TD$INLE]), 1-propanol; dot line (  ), 2-propanol.

temperature T. After calculating the deviations for each data points in binary systems, it was found that the maximum deviation and the average deviation of the activity coefficients were 0.0046 and 0.0022, respectively. Therefore, the uncertainty of activity coefficient was within 0.0022. The NRTL model for nonelectrolyte solution [15] was used to correlate the experimental vapor pressure data for the binary systems containing IL. The NRTL parameters were obtained by optimizing the following objective function: exp n X Pcal i
Pi

OFðPÞ ¼ min

i¼1

Pexp i

!2 (5)

where n is the number of data points. Results for the fitted parameters are summarized in Table 9, together with the correlation accuracy expressed in terms of relative root mean square deviation (rRMSD), which is defined as following: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! u exp 2 n u1X Pcal t i
Pi (6) rRMSDðPÞ ¼ n i¼1 Pexp i

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

63

The correlated results from NRTL model are in good agreement with the experimental ones with an overall rRMSD of 0.0053 for the binary systems studied. The comparison between experimental and correlated vapor pressure for {water (1) + [EMIM][MS] (2)} system is presented in Fig. 1. The plot exhibits a linear relation between the lg(P/kPa) and the 1000(T/K
46.13) over the temperature and pressure range studied at a specified IL concentration, which is in line with Antoine-type relation as observed for pure water. Additionally, the vapor pressure decreases with increasing IL-content at a fixed temperature. For the other binary systems, similar tendency are obtained as well and presented in Supplementary material (Figs. S5–S9). As proved earlier, ILs can decrease the vapor pressure of a solvent. Nevertheless, the degree of non-ideality, which is reflected by activity coefficient, varies with the nature of the volatile solvents. From the obtained activity coefficients of solvents in

Tables 2–7, all propanol-containing binary systems show positive deviations from Raoult’s law; however, the aqueous binary system shows negative deviations. In order to discuss the interactions between ILs of various structures and volatile solvents concerned in this work, the activity coefficients of water, 1-propanol, and 2-propanol in their corresponding mixtures with [EMIM][MS] or [EMIM][ES] were predicted at 343.15 K separately by the obtained binary NRTL parameters. The relevant results are shown in Fig. 2. The predictions for [MMIM][MS] from our previous work are also included here. From the figure, it is apparent that both the negative deviations for water and the positive deviations for propanol become increasingly significant for all the ILs as the IL-content increases. In other words, the non-ideality becomes more and more obvious with the increasing IL-concentration. At a fixed ILcontent, moreover, the non-ideality, whether the negative deviations for water or the positive deviations for propanol, shows the

Table 10 Experimental and predicted vapor pressure data for the ternary system {water (1) + 1-propanol (2) + [EMIM][MS] (3)} with a fixed IL mass fraction of w3 = 0.3.a,b

Table 11 Experimental and predicted vapor pressure data for the ternary system {water (1) + 1-propanol (2) + [EMIM][ES] (3)} with a fixed IL mass fraction of w3 = 0.3.a,b

Pcal/kPa

g 1cal

g 2cal

T/K

x1 = 0.9334, x2 = 0.0306 318.90 12.13 327.51 18.80 333.75 25.36 341.73 36.37 348.56 48.91 355.38 65.10 362.46 85.45 366.87 99.53

12.52 19.22 25.80 36.87 49.28 64.97 85.43 100.66

0.9991 0.9998 1.0003 1.0007 1.0011 1.0014 1.0016 1.0018

11.0014 10.6787 10.4473 10.1565 9.9126 9.6745 9.4337 9.2871

x1 = 0.8503, x2 = 0.1075 316.42 12.61 323.94 18.60 330.31 25.36 338.44 36.92 344.82 48.87 352.12 66.32 358.61 85.59 362.51 99.33

12.31 18.18 24.87 36.33 48.18 65.55 85.07 98.96

1.0670 1.0665 1.0660 1.0652 1.0645 1.0636 1.0628 1.0623

x1 = 0.7330, x2 = 0.2162 317.57 13.49 324.88 19.66 330.61 26.03 338.65 37.82 345.08 50.26 352.56 68.94 358.29 86.01 362.33 100.49

13.15 19.22 25.51 37.20 49.55 68.03 85.85 100.57

x1 = 0.5543, x2 = 0.3815 317.81 13.16 325.92 19.97 331.39 25.65 339.37 37.81 345.73 50.09 352.52 66.80 358.76 85.57 362.86 100.01 x1 = 0.2498, x2 = 0.6633 321.46 13.52 329.86 20.73 334.76 26.26 342.47 37.51 348.99 49.92 356.00 67.04 362.16 85.97 366.33 100.21

T/K

a b

Pexp/kPa

Pcal/kPa

g 1cal

g 2cal

x1 = 0.9354, x2 = 0.0307 321.48 13.69 328.35 19.31 334.08 25.36 342.32 36.89 348.81 48.87 355.99 65.23 362.30 83.01 367.09 98.39

14.51 20.32 26.58 38.34 50.44 67.36 85.90 102.58

1.0049 1.0050 1.0051 1.0051 1.0052 1.0052 1.0052 1.0051

11.3089 11.0168 10.7776 10.4428 10.1864 9.9112 9.6767 9.5038

4.8437 4.8135 4.7848 4.7448 4.7110 4.6702 4.6324 4.6090

x1 = 0.8525, x2 = 0.1077 318.53 13.80 326.99 21.18 331.84 26.78 339.63 38.31 346.56 51.77 352.67 66.49 359.45 86.53 363.49 100.67

13.93 21.41 27.04 38.73 52.41 67.64 88.67 103.57

1.0766 1.0751 1.0742 1.0728 1.0715 1.0703 1.0690 1.0683

4.8903 4.8486 4.8229 4.7791 4.7380 4.7004 4.6573 4.6312

1.1769 1.1770 1.1769 1.1762 1.1755 1.1743 1.1733 1.1725

2.6554 2.6521 2.6488 2.6430 2.6376 2.6304 2.6244 2.6199

x1 = 0.7352, x2 = 0.2168 318.45 13.91 326.26 20.75 331.97 27.38 339.68 39.06 345.25 49.92 352.38 67.39 358.43 85.27 362.77 100.90

13.92 20.79 27.47 39.30 50.27 67.98 86.84 102.92

1.1987 1.1974 1.1962 1.1944 1.1929 1.1909 1.1891 1.1878

2.6302 2.6256 2.6216 2.6153 2.6103 2.6031 2.5965 2.5916

12.84 19.59 25.68 37.40 49.71 66.43 85.73 100.74

1.3122 1.3180 1.3213 1.3252 1.3276 1.3295 1.3309 1.3315

1.7317 1.7278 1.7251 1.7211 1.7179 1.7144 1.7111 1.7089

x1 = 0.5564, x2 = 0.3829 317.47 12.74 326.16 20.12 331.92 26.66 339.28 37.54 346.48 51.56 352.42 66.37 358.83 85.56 363.09 100.73

12.79 20.10 26.68 37.67 51.90 66.81 86.74 102.50

1.3712 1.3747 1.3763 1.3776 1.3783 1.3784 1.3781 1.3777

1.6844 1.6804 1.6777 1.6743 1.6711 1.6684 1.6655 1.6636

13.13 20.27 25.78 37.01 49.48 66.63 85.53 100.71

1.3320 1.3532 1.3646 1.3813 1.3943 1.4072 1.4176 1.4243

1.2726 1.2692 1.2672 1.2641 1.2615 1.2588 1.2563 1.2547

x1 = 0.2514, x2 = 0.6665 323.00 14.83 328.94 20.04 334.88 26.72 342.98 38.79 349.04 50.59 355.40 66.27 362.38 86.05 366.47 100.63

14.50 19.67 26.34 38.41 50.25 65.82 87.28 102.39

1.5481 1.5579 1.5664 1.5764 1.5828 1.5884 1.5936 1.5961

1.2129 1.2118 1.2109 1.2097 1.2088 1.2080 1.2071 1.2066

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa. Note: The overall predicted deviation in terms of rRMSD(P) is 0.0156.

a b

Pexp/kPa

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa. Note: The overall predicted deviation in terms of rRMSD(P) is 0.0228.

64

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

order of [MMIM][MS] > [EMIM][MS] > [EMIM][ES]. Obviously, the non-ideality of solvents is basically related to the ion sizes of ILs. The smaller the ILs are, the stronger the interactions between ILs and water are and the weaker the interactions between ILs and propanol are. From the comparison of the activity coefficients, the strength of the molecular interactions of the alkylsulfate-based ILs with three solvents follows the order of water > 1-propanol > 2propanol. Therefore, the strong affinity of these ILs to water enables them to be potential entrainers for separating {water + 1propanol/2-propanol} azeotropic mixtures by extractive distillation.

(3)}, and {water (1) + 2-propanol (2)+ [EMIM][ES] (3)}, were measured at a fixed IL mass fraction w3 = 0.3 and component 2 (1-propanol or 2-propanol) mass fraction w20 = 0.1, 0.3, 0.5, 0.7, and 0.9. Here w20 denotes to the mass fraction of 1-propanol or 2propanol for the ternary system on IL-free basis. The measured vapor pressure data are listed in Tables 10–13, where the ILcontents are expressed in mole fraction. For the ternary systems {solvent (1) + solvent (2) + IL (3)} studied here, the vapor phase is regarded as ideal gas under low pressure. Also, the vapor pressure of IL is ðPs3 ¼ 0Þ. Therefore, the system pressure can be deduced as the following equation:

3.2. Ternary vapor pressure

P ¼ Ps1 g 1 x1 þ Ps2 g 2 x2

Vapor pressure data for four ternary systems, {water (1) + 1propanol (2) + [EMIM][MS] (3)}, {water (1) + 1-propanol (2) + [EMIM][ES] (3)}, {water (1) + 2-propanol (2) + [EMIM][MS]

where the activity coefficients (g 1 and g 2) can be calculated by NRTL model. Using the newly correlated NRTL binary parameters in this work and the known parameters for volatile components taken from literature [16], the vapor pressure of the four ternary

Table 12 Experimental and predicted vapor pressure data for the ternary system {water (1) + 2-propanol (2) + [EMIM][MS] (3)} with a fixed IL mass fraction of w3 = 0.3.a,b

Table 13 Experimental and predicted vapor pressure data for the ternary system {water (1) + 2-propanol (2) + [EMIM][ES] (3)} with a fixed IL mass fraction of w3 = 0.3.a,b

T/K

Pexp/kPa

Pcal/kPa

g 1cal

g 2cal

T/K

Pexp/kPa

(7)

Pcal/kPa

g 1cal

g 2cal

x1 = 0.9334, x2 = 0.0306 321.07 14.91 326.85 19.99 333.34 27.38 340.45 37.93 346.17 48.67 352.26 62.45 359.87 84.11 364.29 100.04

15.88 20.97 28.26 38.55 48.97 62.53 83.67 98.46

0.9889 0.9897 0.9906 0.9914 0.9920 0.9925 0.9932 0.9935

8.6891 8.4181 8.1318 7.8391 7.6175 7.3944 7.1332 6.9896

x1 = 0.9354, x2 = 0.0307 318.57 12.89 327.01 19.97 333.40 27.24 339.73 36.58 346.34 49.12 353.20 65.18 360.09 85.07 363.94 99.37

13.43 20.30 27.29 36.12 47.75 62.96 82.07 94.66

0.9993 0.9997 0.9999 1.0001 1.0003 1.0004 1.0006 1.0006

7.5220 7.2528 7.0594 6.8762 6.6936 6.5130 6.3403 6.2476

x1 = 0.8503, x2 = 0.1075 312.69 13.15 318.79 18.14 325.15 25.00 332.79 36.29 339.48 48.96 346.52 66.58 352.54 84.09 356.64 99.79

13.63 18.68 25.53 36.46 48.97 65.82 83.74 98.13

1.0286 1.0288 1.0290 1.0291 1.0291 1.0290 1.0289 1.0288

5.0645 4.9624 4.8600 4.7424 4.6442 4.5451 4.4640 4.4104

x1 = 0.8524, x2 = 0.1078 311.84 12.34 320.23 19.22 326.47 26.27 333.26 36.27 339.29 47.94 347.34 67.46 352.07 82.33 356.85 98.81

12.28 18.99 25.81 35.42 46.32 65.09 78.78 94.95

1.0534 1.0520 1.0510 1.0498 1.0489 1.0476 1.0469 1.0461

4.4087 4.3233 4.2606 4.1937 4.1353 4.0590 4.0151 3.9715

x1 = 0.7329, x2 = 0.2161 310.95 12.84 318.59 19.26 324.90 26.46 332.33 37.79 339.62 52.56 345.97 69.04 351.26 86.20 355.04 100.12

13.12 19.66 26.96 38.36 53.19 69.75 86.60 100.55

1.1235 1.1226 1.1218 1.1206 1.1193 1.1181 1.1170 1.1163

2.9243 2.8867 2.8564 2.8216 2.7884 2.7601 2.7371 2.7209

x1 = 0.7352, x2 = 0.2166 311.38 13.00 318.53 19.06 323.71 24.79 331.79 36.60 338.21 49.10 344.66 64.72 350.81 83.20 355.20 99.16

12.86 18.80 24.43 36.01 48.20 63.73 82.18 97.91

1.1647 1.1617 1.1595 1.1561 1.1534 1.1507 1.1482 1.1465

2.6251 2.6061 2.5921 2.5698 2.5519 2.5339 2.5166 2.5042

x1 = 0.5543, x2 = 0.3815 313.75 15.26 318.79 19.90 324.72 26.83 331.31 36.85 336.69 47.13 343.23 62.65 349.63 81.82 354.62 99.96

15.28 19.94 26.93 37.01 47.43 63.26 82.79 101.27

1.2902 1.2898 1.2891 1.2880 1.2868 1.2851 1.2832 1.2816

1.8010 1.7943 1.7865 1.7780 1.7711 1.7629 1.7549 1.7488

x1 = 0.5564, x2 = 0.3829 311.50 13.29 318.62 19.42 324.01 25.57 332.04 37.65 338.08 49.55 344.58 65.70 350.68 84.32 354.83 99.57

13.22 19.36 25.47 37.54 49.50 65.73 84.81 100.21

1.3472 1.3447 1.3426 1.3391 1.3363 1.3331 1.3300 1.3278

1.6907 1.6850 1.6806 1.6740 1.6689 1.6634 1.6581 1.6546

x1 = 0.2499, x2 = 0.6632 313.00 14.13 319.26 19.65 324.80 26.10 331.99 36.95 337.74 48.18 344.02 63.61 350.20 82.68 355.03 100.40

14.08 19.71 26.17 37.13 48.44 63.97 83.06 101.01

1.3901 1.3974 1.4032 1.4097 1.4142 1.4185 1.4222 1.4246

1.2667 1.2643 1.2622 1.2595 1.2573 1.2549 1.2525 1.2506

x1 = 0.2513, x2 = 0.6667 313.21 14.15 321.01 21.41 326.64 28.37 332.54 37.50 338.33 49.22 344.71 65.10 350.98 84.54 355.26 99.97

13.97 21.16 28.11 37.38 48.81 64.64 84.12 100.03

1.4740 1.4873 1.4959 1.5039 1.5109 1.5178 1.5237 1.5272

1.2127 1.2090 1.2064 1.2037 1.2012 1.1984 1.1958 1.1940

a b

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa. Note: The overall predicted deviation in terms of rRMSD(P) is 0.0186.

a b

Standard uncertainties u are u(T) = 0.02 K, u(x) = 0.0001, and u(P) = 0.04 kPa. Note: The overall predicted deviation in terms of rRMSD(P) is 0.0205.

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

[(Fig._3)TD$IG]

65

model parameters are feasible to describe the VLE behavior of multi-component systems containing IL. 3.3. Isobaric vapor–liquid equilibrium Isobaric x–y phase diagram is an important tool to provide intuitive image describing the salt effect of ILs on the VLE behavior of azeotropic mixtures. Thus, the isobaric VLE data for these mixtures with the IL-content x3 = 0.05, 0.15, and 0.25 were predicted at 101.325 kPa with NRTL model to evaluate the effect of [EMIM][MS] and [EMIM][ES] on the relative volatility for the azeotropic mixtures of {water + 1-propanol} and {water + 2-propanol}. According to Eq. (7), the vapor phase mole fraction of component i can be calculated as the following equation yi ¼ exp

Ps1

Psi g i xi ; g 1 x1 þ Ps2 g 2 x2

i ¼ 1; 2:

(8)

cal

Fig. 3. Experimental (P ) versus predicted (P ) vapor pressure for the four ternary systems studied containing an alkylsulfate-based IL.

systems concerned here can be predicted. The predictions for vapor pressure and the activity coefficients are also given in Tables 10–13. The values of predicted and experimentally measured vapor pressure agree quite well with the overall rRMSD(P) of 0.0196. Fig. 3 compares the experimental and predicted vapor pressure data. The results indicate that the NRTL

The relative volatility (a21) of component 1 to component 2 is defined as the following:

a21 ¼

y2 =x0 2 y1 =x0 1

(9)

The calculated isobaric phase diagrams (x0
y) for the ternary system {water (1) + 1-propanol (2) + [EMIM][MS] (3)}, {water (1) + 1-propanol (2) + [EMIM][ES] (3)}, {water (1) + 2-propanol

[(Fig._4)TD$IG]

Fig. 4. Predicted isobaric VLE diagram (y2
x20 ) for the ternary system (a) {water (1) + 1-propanol (2) + [EMIM][MS] (3)}, (b) {water (1) + 1-propanol (2) + [EMIM][ES] (3)}, (c) {water (1) + 2-propanol (2) + [EMIM][MS] (3)}, and (d) {water (1) + 2-propanol (2) + [EMIM][ES] (3)} at 101.325 kPa, with varying IL mole fractions: dash dot line ( [TD$INLE]), x3 = 0.05; dash line ( [TD$INLE]), x3 = 0.15; solid line (—), x3 = 0.25, respectively. The isobaric VLE data for the corresponding IL-free systems are indicated as the legend circle (). Here x20 denotes the mole fraction of component 2 for the concerning ternary system on an IL-free basis.

66

[(Fig._5)TD$IG]

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

Fig. 5. Predicted isobaric VLE diagram (y2
x20 ) for the ternary system (a) {water (1) + 1-propanol (2) + IL (3)} and (b) {water (1) + 2-propanol (2) + IL (3)} at 101.325 kPa with fixed IL mole fraction x3 = 0.15 of varying ILs: solid line (—), [MMIM][MS], taken from ref. [12]; dash line ( [TD$INLE]), [EMIM][MS]; dash dot line ( [TD$INLE]), [EMIM][ES]; dash dot dot line ( [TD$INLE]), [EMIM][BF4], taken from ref. [17]; short dot line (  ), [EMIM][OTf], taken from [18], respectively. The isobaric VLE data for the corresponding IL-free systems are indicated as the legend circle (). Here x20 denotes the mole fraction of component 2 for the concerning ternary system on an IL-free basis.

(2) + [EMIM][MS] (3)}, and {water (1) + 2-propanol (2) + [EMIM] [ES] (3)} are shown in Fig. 4, in which the liquid phase mole fraction x0 is provided on IL-free basis. The detailed isobaric VLE data for the ternary systems can be found in Supplementary material. For the purpose of comparison, the VLE data in the absence of IL are also plotted in these figures. As observed in all the systems here, the addition of ILs indeed shift the azeotropic points upward to the region of higher propanol concentration. The results suggest that the relative volatility of propanol to water is improved. The azeotropic phenomenon can be completely eliminated at the sufficient IL-content. Moreover, we find that both ILs exhibit a saltin effect on 1-propanol or 2-propanol in water-rich region, but show a salt-out effect in water-lean region. In this section, the comparison of separation ability on the azeotropic systems among various ILs is tried as well. For comparing the separation ability of [EMIM][MS] and [EMIM][ES] to that of other ILs, the isobaric VLE data at 101.325 kPa for the corresponding ternary systems were calculated by the NRTL parameters obtained in this work and reference, with the IL liquid mole fraction x3 = 0.15. 1,3Dimethylimidazolium methylsulfate ([MMIM][MS]) [12], 1-ethyl3-methylimidazolium tetrafluoroborate ([EMIM][BF4]) [17], and

1-ethyl-3-methylimidazolium trifluoromethanesulfonate [EMIM] [OTf] [18] are selected here to consider the diversity of cation and anion. The isobaric phase diagrams for the ternary systems with different ILs are presented in Fig. 5. From the figure, ILs can bring in a remarkable improvement of relative volatility for all azeotropic systems. For {water + 1-propanol} mixture, further, the enhancement of relative volatility shows the order of [MMIM][MS] > [EMIM] [MS] > [EMIM][ES] > [EMIM][BF4] > [EMIM][OTf]. The result indicates that the alkylsulfate-based ILs exhibit greater separation ability than the other two ILs; in addition, the alkylsulfate-based ILs with smaller ion size are more favorable. Actually, the phenomenon that ion size can influence the separation efficiency has also been observed among other imidazolium-based ILs [19]. For {water + 2propanol} mixture, the separation ability follows the trend of [MMIM][MS] > [EMIM][MS] > [EMIM][ES] > [EMIM][BF4], which is in line with the qualitatively predicted result using Conductor-like Screening Model for Real Solvents (COSMO-RS) approach [20]. Generally, the salt effect of an IL on azeotropic mixtures is associated with its the anion structure. Here, this paragraph is specially written to explain the separation ability of ILs with various anions under an electronic view. Charge distribution can provide useful information of electronic property to discuss the potential interaction mechanism [21]. The NPA charges were calculated at the B3LYP/6-31++G** level for each anion of interest, and the results are shown in Fig. 6. According to previous studies [22,23], the active H atom on water or alcohol molecule probably interacts with O or F atom on IL anion to form the corresponding hydrogen bond. As manifested from the NPA charges, negative charge mainly distributes on the O and F atoms which are hydrogen acceptor. However, the interaction between IL and water/alcohol in solution phase is not simple hydrogen bonding interaction in short distance. ILs tend to be dissociated in low IL-content solution, as discussed in our earlier study [22]. From the view of electrolyte solution theory [24], the more negative charge and smaller size of ions, the stronger electrostatic interaction between ions and water, and accordingly the greater salt-out effect on alcohol. Herein, for the anions of alkylsulfate-based IL, the lengthening alkyl chain slightly reduces the negative charge on the terminal O atom and meanwhile depolarizes the anion. As a result, the smaller [EMIM] [MS] shows the greater salt effect on the mixtures of water and propanol than [EMIM][ES]. The highly symmetric anion, [BF4]
, has a structural framework similar to the electronegative segment of alkylsulfate-based anions; however, the negative charge on each F atom of [BF4]
is much smaller than that on O atom of [MS]
or [ES]
. Thereby, the ionic solvation ability of [BF4]
is weaker than that of [MS]
or [ES]
. With regard to [OTf]
, the electronwithdrawing F atoms can absorb negative charge from the O atoms. As a result, the O atoms of [OTf]
are less negative than those of [MS]
, and the F atoms of [OTf]
also carry some negative charge. Among the four anions, [OTf]
has the least localized distribution of negative charge. Charge localization can facilitate stronger Coulombic interactions between the ions, creating larger cationanion affinity, and causing that the IL-containing aqueous mixtures become more similar to electrolyte solution [25]. Because of these reasons, the salt effect of ILs from anion structure on VLE of the studied azeotropes follow the sequence of [MS]
> [ES]
> [BF4]
> [OTf]
. In summary, the alkylsulfate-based ILs, [EMIM][MS] and [EMIM][ES], exhibit satisfactory ability to enhance the separation of {water + 1-propanol} and {water + 2-propanol} mixtures. From the commercial point of view, these ILs are more economically feasible than many other imidazolium-based ILs in industrial scale production. Thus, the alkylsulfate-based ILs here might be applied as a promising entrainer to separate of 1-propanol and 2propanol from water by extractive distillation in industrial applications.

Y. Dai et al. / Fluid Phase Equilibria 397 (2015) 58–67

[(Fig._6)TD$IG]

67

Fig. 6. NPA charges of the IL anions for [MS]
, [ES]
, [BF4]
, and [OTf]
calculated at B3LYP/6-31++G** level.

4. Conclusions Vapor pressure data were measured by a quasi-static ebulliometric method for water, 1-propanol, and 2-propanol, as well as the mixtures of {water + 1-propanol} and {water + 2-propanol}, in the presence of [EMIM][MS] or [EMIM][ES]. The data for binary systems were correlated by NRTL model with a rRMSD of 0.0053, and the obtained binary NRTL parameters were employed to predict the vapor pressure of two ternary systems with an overall rRMSD of 0.0196. From the results of binary systems, the vapor pressure of solvents decreases with increasing IL-content at a fixed temperature. In the range of IL-contents and temperatures investigated in this work, 1-propanol and 2-propanol show positive deviations from Raoult’s law, while water shows negative deviations from Raoult’s law. Besides, the deviations becomes increasingly significant in the order of [MMIM][MS] > [EMIM][MS] > [EMIM][ES], which is consistent with the corresponding ion sizes. The predicted activity coefficients of solvents for binary systems indicate that the interaction strength between the alkylsulfate-based ILs and the volatile solvents follows the order of water > 1-propanol > 2propanol. According to the isobaric VLE prediction for ternary systems, the addition of [EMIM][MS] or [EMIM][ES] can enhance the relative volatility of propanol to water. By studying comparatively, the imidazolium-based ILs with various anion structures can generate a significant salt effect on the VLE for the mixtures of water and propanol, following the order of [MMIM][MS] > [EMIM] [MS] > [EMIM][ES] > [EMIM][BF4] > [EMIM][OTf]. A theoretical insight at electronic level was also presented to explain the underlying reason for varying separation ability of ILs from differing anion structures. To sum up, the alkylsulfate-based ILs might be a promising entrainer to separate the azeotropic mixtures of {water + 1-propanol} and {water + 2-propanol} by extractive distillation. Acknowledgements The authors are grateful for the apparatus for vapor pressure measurement from Prof. Chunxi Li (Beijing University of Chemical Technology, Beijing, China) and the experimental guidance from

Dr. Chong Shen (Liming Research Institute of Chemical Industry, Luoyang, China). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.fluid.2015.03.043. References [1] M. Seiler, C. Jork, A. Kavarnou, W. Arlt, R. Hirsch, AIChE J. 50 (2004) 2439–2454. [2] N.V. Plechkova, K.R. Seddon, Chem. Soc. Rev. 37 (2008) 123–150. [3] A.B. Pereiro, J.M.M. Araújo, J.M.S.S. Esperança, I.M. Marrucho, L.P.N. Rebelo, J. Chem. Thermodyn. 46 (2012) 2–28. [4] M.G. Freire, C.M. Neves, I.M. Marrucho, J.A. Coutinho, A.M. Fernandes, J. Phys. Chem. A 114 (2009) 3744–3749. [5] J.D. Holbrey, W.M. Reichert, R.P. Swatloski, G.A. Broker, W.R. Pitner, K.R. Seddon, R.D. Rogers, Green Chem. 4 (2002) 407–413. [6] E. Kuhlmann, S. Himmler, H. Giebelhaus, P. Wasserscheid, Green Chem. 9 (2007) 233–242. [7] J. Zhao, X. Jiang, C. Li, Z. Wang, Fluid Phase Equilib. 247 (2006) 190–198. [8] J. Wang, C. Li, Z. Wang, Z. Li, Y. Jiang, Fluid Phase Equilib. 255 (2007) 186–192. [9] X. Jiang, J. Wang, C. Li, L. Wang, Z. Wang, J. Chem. Thermodyn. 39 (2007) 841–846. [10] N. Calvar, B. González, E. Gómez, Á. Domínguez, J. Chem. Eng. Data 53 (2008) 820–825. [11] J.F. Wang, C.X. Li, Z.H. Wang, J. Chem. Eng. Data 52 (2007) 1307–1312. [12] Y. Dai, Y. Qu, S. Wang, J. Wang, Fluid Phase Equilib. 385 (2015) 219–226. [13] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, Gaussian 09, Gaussian, Inc., Wallingford, CT, 2009. [14] P. Ma, H. Chang, S. Xia, M. Chen, Chemical Engineering Thermodynamics, first ed., Chemical Industry Press, Beijing, 2005. [15] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135–144. [16] J. Gmehling, U. Onken, W. Arlt, Vapor–Liquid Equilibrium Data Collection, DECHEMA, Frankfurt, 1977. [17] L. Zhang, J. Han, R. Wang, X. Qiu, J. Ji, J. Chem. Eng. Data 52 (2007) 1401–1407. [18] A.V. Orchillés, P.J. Miguel, E. Vercher, A. Marti’nez-Andreu, J. Chem. Eng. Data 53 (2008) 2426–2431. [19] C. Shen, X. Li, Y. Lu, C. Li, J. Chem. Thermodyn. 43 (2011) 1748–1753. [20] V.K. Verma, T. Banerjee, J. Chem. Thermodyn. 42 (2010) 909–919. [21] L.E. Ficke, J.F. Brennecke, J. Phys. Chem. B 114 (2010) 10496–10501. [22] Y. Dai, Y. Qu, S. Wang, J. Wang, Chem. Phys. Lett. 608 (2014) 366–372. [23] Q. Zhang, N. Wang, S. Wang, Z. Yu, J. Phys. Chem. B 115 (2011) 11127–11136. [24] P. Debye, J. McAulay, Phys. Z. 26 (1925) 22–29. [25] A.A.H. Pádua, M.F. Costa Gomes, J.N.A. Canongia Lopes, Acc. Chem. Res. 40 (2007) 1087–1096.