RESEARCH NOTES Chinese Journal of Chemical Engineering, 19(4) 703—708 (2011)

Vapor-Liquid Equilibrium Measurements and Modeling for Ternary System Water + Ethanol + 1-Butyl-3-methylimidazolium Acetate* DENG Dongshun (邓东顺), WANG Rufa (汪如发), ZHANG Lianzhong (章连众)**, GE Yun (葛筠) and JI Jianbing (计建炳) Zhejiang Province Key Laboratory of Biofuel, College of Chemical Engineering and Material Science, Zhejiang University of Technology, Hangzhou 310014, China

Abstract Vapor-liquid equilibrium (VLE) data were measured for ternary system water + ethanol + 1-butyl-3methylimidazolium acetate ([bmim][OAc]), in a relatively wide range of ionic liquid (IL) mass fractions up to 0.8. Six sets of complete T-x-y data were obtained, in which the mole fraction of ethanol on IL-free basis was fixed separately at 0.1, 0.2, 0.4, 0.6, 0.8, and approximate 0.98. The non-random-two-liquid (NRTL) and electrolyte non-random-two-liquid (eNRTL) equations were used for correlation, showing similar deviations. The ternary VLE was also modeled with the correlation from two data sets, with the mole fractions of ethanol on IL-free basis being 0.1 and approximate 0.98. The VLE data were also reproduced satisfactorily. With the eNRTL model, the root-mean-square deviation for temperature is 0.79 K and that for vapor-phase mole fraction is 0.0094. The calculations are in good agreement with experimental data. The effect of the IL on the VLE behavior of the volatile components is also illustrated. Keywords vapor-liquid equilibrium, eNRTL equation, ionic liquid, ethanol, activity coefficient, relative volatility

1

INTRODUCTION

In recent years, ionic liquids (ILs) have received considerable attention for their use in chemical industry and are considered to be an alternative to conventional entrainers in extractive distillation [1, 2]. When an IL is introduced into a mixture of volatile components to be separated, the components change their non-ideality and the activity coefficients are affected to a different extent. It is desirable that these changes result in an increase of relative volatility. The performance of IL depends mainly on the composition dependence of activity coefficients in the IL-containing mixture. It is necessary to understand the vapor-liquid phase behavior of a mixture containing IL, especially the relative volatility enhanced by the addition of IL. At the present stage, such information is mainly obtained from experimental measurement of vapor-liquid equilibrium (VLE) data and by correlation using a model of excess Gibbs energy [3-8]. In our previous work [9, 10], we presented a procedure for the experimental measurement and modeling of vapor-liquid equilibrium of ternary systems water + ethanol + 1-hexyl-3-methylimidazolium chloride ([hmim]Cl) and water + ethanol + 1-butyl-3methylimidazolium chloride ([bmim]Cl). Six sets of complete T, x, y data were measured, in which two data sets were used for correlation of the NRTL model [11] and the other four data sets were used for evaluating the model. It is recommended that the VLE behavior of a ternary system containing IL is modeled with the correlation of two ternary data sets covering a relatively wide range of IL mass fractions, while the

mole fractions of the volatile quasi-binary pair are distributed uniformly in the range. In this work, owing to the low viscosity of 1-butyl-3-methylimidazolium acetate ([bmim][OAc]), which is attractive for heat and mass transfer in extractive distillation, we will extend the experimental measurement and modeling to the system water (1) + ethanol (2) + [bmim][OAc] (3). The main goal is to obtain composition dependence of the activity coefficients of water and ethanol in the IL-containing mixture in a wide range of composition. The modeling is based on T, x, y data in a range of IL mass fractions up to 0.8 and in a relatively complete composition range for the volatile binary pair. In previous work [12], we also studied the same ternary system, but the measurements were performed only in the ethanol-rich region and the highest IL mass faction was 0.6. To the best of our knowledge, there are no other VLE data for the ternary system. 2 2.1

EXPERIMENTAL Materials

Water was doubly distilled. Ethanol (analytical reagent grade, Sinopharm Chemical reagent Co. Ltd, its purity was 99.9% checked by gas chromatography (GC), while Karl-Fischer analysis indicated a water mass fraction of 3.8×10−4), 1-methylimidazole (99.5%, Yancheng Medical Chemical Factory), 1-chlorobutane (chemical pure grade, Sinopharm Chemical reagent Co. Ltd) and potassium acetate (ACS grade, 99.0%, Aladdin-reagent) were used without further purification. For preparation of [bmim][OAc], [bmim][Cl]

Received 2010-12-06, accepted 2011-04-11. * Supported by the National Natural Science Foundation of China (20776132). ** To whom correspondence should be addressed. E-mail: [email protected]

704

Chin. J. Chem. Eng., Vol. 19, No. 4, August 2011

was synthesized with the method in [13], by reacting 1-methylimidazole with 1-chlorobutane under a nitrogen atmosphere. [bmim][Cl] was purified by recrystallization in a mixture of ethyl acetate and acetonitrile. Consequently, [bmim][OAc] was prepared by a patented anion metathesis procedure, in which ethanol was used as solvent and potassium acetate was used as the source of anion [14]. Before use, the IL was dried for 24 h under vacuum at 340 K to remove volatile impurities. 2.2

Experimental apparatus

VLE data were measured by an ebulliometer, which was described in detail previously [15, 16]. In the measurements, liquid-phase circulation was enhanced by a pump-like stirrer and had a value of approximate 200 cm3·min−1. Vapor-phase circulation was maintained at about 1 cm3·min−1, while the vapor-toliquid circulation ratio calculated on mass basis was 0.005. Vapor condensate was cooled to 275 K. The volume of vapor material is estimated to be 1.7 cm3, very small compared to the total sample volume of 270 cm3 in the ebulliometer. Pressure was measured by a precision pressure gauge with an uncertainty of ±0.04 kPa. A 20 dm3 glass container was used as pressure buffer connected to the ebulliometer. Temperature was measured by a standard platinum thermometer and a 6-1/2-digit multimeter. Uncertainty of the resistance measurement was ±8 mΩ, which is equivalent to ±0.08 K for temperature measurement. 2.3

Experimental procedures

The VLE measurements were performed in such a way that the IL mass fraction, w3, changed from high to low, while the mole fraction of ethanol on IL-free basis, x2′ , remained approximately unchanged. At higher IL mass fractions, namely at w3 = 0.6, 0.7, and 0.8, the measurements were performed at subatmospheric pressures. It is commonly recognized that the activity coefficients in a liquid mixture depend strongly on composition. Often, they depend only weakly on temperature and very weakly on pressure. Therefore, the present measurements will provide mainly the dependence of activity coefficients on composition. At the beginning of measurement, samples of water, ethanol, and the IL were introduced into the ebulliometer. The water contents of ethanol and the IL were determined by Karl-Fischer analysis. Every sample added in or taken out of the ebulliometer was weighed with an electronic balance (Mettler-Toledo AL204) with an uncertainty of ±0.0002 g. Masses of water and other components added in the ebulliometer were calculated, so we had the total amount for the first measurement. When the equilibrium was established, the vapor condensate was sampled and ana-

lyzed. As the IL is nonvolatile, the vapor phase is composed of water and ethanol. Vapor-phase composition was determined by analyzing the water content using the Karl-Fischer method. When water mole fraction is greater than 0.1, the sample was first mixed with ethanol quantitatively and the water content of the mixture was analyzed. Vapor-phase composition was then calculated from the ratio of mixing and the water content. The uncertainty of the vapor-phase composition was estimated to be 0.0001 in water mole fraction or 1%. Liquid-phase compositions were calculated on the basis of mass balance, using a procedure presented in previous work [15, 17]. The next measurement was carried out by replacement of certain amount of the mixture in the boiler with IL-free mixture of water and ethanol to keep x2′ approximately unchanged. The measurement was repeated until w3 was close to 0.1. 3

RESULTS and DISCUSSION

The experimental VLE data for the ternary system water (1) + ethanol (2) + [bmim][OAc] (3) are listed in Table 1. The six data sets are for x2′ = 0.1, 0.2, 0.4, 0.6, 0.8, and approximate 0.98, while x2′ is almost unchanged. The pressure is about 30 kPa for w3 = 0.8, 0.7, and 0.6, and 100 kPa for w3 = 0.5 to 0.1. Therefore the pressure is almost the same for a given w3 at different x2′ . In the complete T, x, y data, the liquidphase compositions are reported in x2′ and w3. Activity coefficients of water ( γ 1 ) and ethanol ( γ 2 ) and the relative volatility of ethanol to water (r2,1) are also reported. In the calculation of the activity coefficients the vapor phase is regarded as an ideal gas and the vapor pressures are calculated by parameters in literature [18]. In the measurements, temperature uncertainty was less than 0.04 K. As discussed in previous work [9, 12], the uncertainty for w3 was estimated to be ±0.003. The uncertainty for molar composition of the volatile binary pair, x2′ or x1′ , was estimated to be less than 1%. Modeling of the ternary VLE is performed by correlating the six data sets, using the NRTL [11] and eNRTL [19] equations. For the eNRTL equation, Vercher et al. have presented expressions for the liquid-phase activity coefficients of volatile components in a ternary system containing a salt [20]. In the correlations, the binary parameters for water + ethanol are from literature [21]. The suitability of the binary parameters has been discussed in previous work [9]. With the non-randomness factors α13 and α23 at the most common value of 0.3 for the NRTL equation and 0.2 for the eNRTL equation, the binary parameters for the ternary system are obtained by minimization of the following objective function F=

∑ n=1 (γ 1,cal / γ 1,exp − 1) N

∑ n=1 (γ 2,cal / γ 2,exp − 1) N

2

2

/N +

/N

(1)

705

Chin. J. Chem. Eng., Vol. 19, No. 4, August 2011 Table 1 Vapor-liquid equilibrium data for the ternary system water (1) + ethanol (2) + [bmim][OAc] (3) x′2

w3

y2′

p/kPa

T/K

γ1

γ2

r2,1

0.0991 0.8008 0.3785

30.04

372.26

0.30

0.74

5.54

0.0998 0.6997 0.3946

29.69

354.48

0.50

1.28

5.88

0.0998 0.6000 0.4092

30.02

344.54

0.69

1.86

6.25

0.1000 0.4998 0.4203 100.52 369.40

0.81

2.35

6.52

0.0998 0.4002 0.4301 100.52 365.36

0.89

2.70

6.81

0.1001 0.3001 0.4388 100.53 362.67

0.95

2.96

7.03

0.0999 0.2000 0.4448 100.52 360.91

0.99

3.15

7.22

0.0999 0.1000 0.4467 100.52 359.71

1.02

3.26

7.27

0.1999 0.7998 0.5606

29.94

370.98

0.26

0.59

5.11

0.2000 0.6999 0.5493

29.94

352.83

0.46

0.98

4.87

0.1998 0.5999 0.5416

29.88

342.19

0.67

1.37

4.73

0.1996 0.5000 0.5359 100.32 366.09

0.83

1.71

4.63

0.1998 0.4001 0.5308 100.36 361.69

0.96

1.92

4.53

0.1999 0.2998 0.5381 100.32 358.94

1.02

2.09

4.66

0.2001 0.2000 0.5410 100.34 357.34

1.06

2.19

4.71

0.2001 0.1000 0.5451 100.32 356.36

1.07

2.25

4.79

in which N is the number of data points. The obtained parameters are used for calculation of the ternary VLE data. The comparisons of calculations with experimental values are given in Table 2, in which δT and δy are respectively the root mean square deviations of temperature and vapor-phase mole fractions. The NRTL and eNRTL equations, with the most common values for the non-randomness factors, provide good correlation for the IL-containing mixture in the experimental composition range. Deviations with the two equations are similar. Table 2 Root mean square deviations δT and δy in calculation of VLE data of water (1) + ethanol (2) + [bmim][OAc] (3) based on the NRTL and eNRTL equations ①

Data sets used in correlation Model

Root mean square deviations ②

δT/K

NRTL

0.60

0.0095

x′2 = 0.1 and x′2 ≈ 0.98

NRTL

0.75

0.0101

six data sets in Table 1

eNRTL

0.75

0.0080

x2′ = 0.1 and x2′ ≈ 0.98

eNRTL

0.79

0.0094

30.21

372.26

0.22

0.40

4.11

0.3997 0.6999 0.7041

29.79

353.73

0.41

0.63

3.57

① Deviations are calculated for the six data sets.

0.3995 0.5998 0.7004

30.02

342.14

0.61

0.92

3.52

② δT =

∑ (Tcal − Texp )

0.4001 0.5002 0.6716 100.37 364.86

0.84

1.15

3.07

0.4001 0.4001 0.6621 100.41 359.61

1.01

1.31

2.94

③ δy =

∑ ( y2,cal − y2,exp )

0.4000 0.3000 0.6483 100.43 356.53

1.15

1.40

2.76

0.4001 0.2001 0.6368 100.46 354.77

1.25

1.43

2.63

0.3999 0.1000 0.6288 100.47 353.86

1.30

1.44

2.54

0.5997 0.7997 0.8513

30.25

376.02

0.17

0.29

3.82

0.5991 0.6998 0.8343

29.85

355.89

0.33

0.49

3.37

0.5991 0.5999 0.8156

30.14

343.50

0.55

0.70

2.96

0.6003 0.5000 0.7903 100.34 365.52

0.81

0.90

2.51

0.6001 0.4000 0.7739 100.38 359.51

1.04

1.04

2.28

0.5997 0.2998 0.7585 100.00 355.89

1.23

1.13

2.10

0.5998 0.2000 0.7409 100.43 353.75

1.40

1.16

1.91

0.5998 0.1000 0.7257 100.00 352.58

1.51

1.16

1.76

0.7979 0.7997 0.9336

30.41

379.29

0.14

0.23

3.56

0.8001 0.6999 0.9249

30.34

358.94

0.28

0.38

3.08

0.8001 0.5998 0.9134

30.22

345.75

0.49

0.56

2.64

0.8002 0.5002 0.8955 100.03 366.73

0.79

0.75

2.14

0.8000 0.4001 0.8845 100.06 360.12

1.05

0.89

1.91

0.8003 0.3002 0.8699 100.00 355.80

1.34

0.98

1.67

0.8001 0.1999 0.8547

1.60

1.02

1.47

water ethanol

353.37

0.8003 0.0999 0.8378

99.76

351.81

1.85

1.04

1.29

0.9265 0.7995 0.9766

30.19

382.32

0.13

0.20

3.31

0.9553 0.6999 0.9829

30.00

362.18

0.26

0.31

2.69

0.9703 0.5998 0.9864

30.00

349.39

0.46

0.44

2.21

0.9795 0.4996 0.9886 100.49 369.60

0.77

0.62

1.82

0.9800 0.4002 0.9875 100.53 361.68

1.09

0.78

1.61

0.9799 0.3000 0.9859 100.55 356.66

1.42

0.89

1.43

0.9799 0.2001 0.9840 100.56 353.75

1.74

0.96

1.26

0.9800 0.1001 0.9818 100.57 352.15

2.06

0.99

1.10

③

six data sets in Table 1

0.3986 0.7996 0.7313

99.87

δy

2

/N . 2

/N .

Following the procedure recommended in [9] for modeling of ternary VLE behavior, two data sets at x2′ = 0.1 and x2′ ≈ 0.98 are used for correlation. Results are also shown in Table 2. The deviations for the ternary VLE data are close to those from the six data sets. Therefore, the two data sets at x2′ = 0.1 and x2′ ≈ 0.98 appear to be adequate for modeling the VLE behavior in the experimental composition range. The optimized binary parameters for the eNRTL equation are given in Table 3. Table 3 Energy parameters Δgij and Δgji and nonrandomness factors αij for the eNRTL model obtained from correlation of ternary VLE data of water (1) + ethanol (2) + [bmim][OAc] (3) using data sets at x2′ = 0.1 and x2′ ≈ 0.98 Component i Component j

Binary parameters

①

Δgij/J·mol−1

Δgji/J·mol−1

αij

[bmim][OAc]

−5125.2

−6756.9

0.2

[bmim][OAc]

−5203.6

−4275.6

0.2

Binary parameters for water (1) + ethanol (2) are taken from reference [20] and fixed at Δg12 = 4458.8 + 8.4420T, Δg21 = −3791.4 + 4.1451T, α12 = 0.1448. ①

Different values of the non-randomness factors have also been tested in the correlations. Results indicate that only slight improvement is achieved by using quite different values. Therefore, only the results with the most common values are presented here. VLE results calculated by using parameters in

706

Chin. J. Chem. Eng., Vol. 19, No. 4, August 2011

(a) Water activity coefficient

(b) Ethanol activity coefficient

Figure 1 Experimental and calculated activity coefficients of (a) water γ1 and (b) ethanol γ2 in relation with IL mole fraction x3 for the saturated mixture water (1) + ethanol (2) + [bmim][OAc] (3) (with pressures given in Table 1; lines: calculated by eNRTL parameters in Table 3 at corresponding x2′ and relevant pressures) ○ x2′ = 0.1; ● x2′ = 0.2; □ x2′ = 0.4; ■ x2′ = 0.6; ◇ x2′ = 0.8; ◆ x2′ ≈ 0.98

(a) Water activity coefficient (b) Ethanol activity coefficient Figure 2 Experimental and calculated activity coefficients of (a) water γ1 and (b) ethanol γ2 in relation with mole fraction of ethanol on IL-free basis x2′ for the saturated mixture water (1) + ethanol (2) + [bmim][OAc] (3) ○ w3 = 0.1, p ≈ 100 kPa; □ w3 = 0.3, p ≈ 100 kPa; △ w3 = 0.5, p ≈ 100 kPa; ● w3 = 0.6, p ≈ 30 kPa; ■ w3 = 0.7, p ≈ 30 kPa; ▲ w3 = 0.8, p ≈ 30 kPa; lines: calculated by the eNRTL equation using parameters in Table 3; calculated values at w3 = 0.1, 0.3, 0.5, 0.6, 0.7, and 0.8 at relevant pressures; calculated values for the system water (1) + ethanol (2) at p = 100 kPa

Table 3 are shown in Figs.1-3, in which good agreement between experiment and calculation is graphically presented. Effects of the IL on the phase behavior are also illustrated in these figures. Figure 1 shows the activity coefficients of water (γ1) and ethanol (γ2) in relation with IL mole fraction x3. Both γ1 and γ2 decrease with increasing w3. It should be noted that, at a given x2′ the pressure varies from 30 kPa to 100 kPa and the temperature is in a narrow range of 351 K to 383 K. Because the activity

coefficients are independent of pressure and depend on temperature weakly, Fig. 1 illustrates the dependence of the activity coefficients on IL mole fraction. With the increase of w3, the activity coefficients at different x2′ approach a similar value. It can be expected that, at a given temperature, the activity coefficient of water (or ethanol) at different x2′ will reach a certain value as x3 approaches unity. This value is the activity coefficient of water (or ethanol) in the IL at infinite dilution at given temperature.

Chin. J. Chem. Eng., Vol. 19, No. 4, August 2011

707

NOMENCLATURE Δg p r2,1 T w x y α12, α13, α23 γ δ

NRTL parameters pressure, kPa relative volatility of component 2 to component 1 temperature, K mass fraction mole fraction of liquid phase mole fraction of vapor phase NRTL parameters activity coefficient root mean square deviations

Superscripts sat ′

saturated vapor pressure of a pure component prime symbol, indicating the quantity on IL-free basis

1, 2, 3

volatile (1, 2) or nonvolatile (3) component

Subscripts Figure 3 Experimental and calculated relative volatility of ethanol to water r2,1 in relation with mole fraction of ethanol on IL-free basis x2′ for the saturated mixture water (1) + ethanol (2) + [bmim][OAc] (3) ○ w3 = 0.1, p ≈ 100 kPa; □ w3 = 0.3, p ≈ 100 kPa; △ w3 = 0.5, p ≈ 100 kPa; ● w3 = 0.6, p ≈ 30 kPa; ■ w3 = 0.7, p ≈ 30 kPa; ▲ w3 = 0.8, p ≈ 30 kPa; lines: calculated by the eNRTL equation using parameters in Table 3; calculated values at w3 = 0.1, 0.3, 0.5, 0.6, 0.7, and 0.8 at relevant pressures; calculated values for the system water (1) + ethanol (2) at p = 100 kPa

REFERENCES 1

2

3

Figure 2 also shows γ1 and γ2 in relation with x2′ , comparing the calculated activity coefficients with the experimental values. At all w3, γ2 decreases as x2′ increases, while γ1 increases with x2′ at w3 = 0.1 and 0.3 and decreases with increasing x2′ at higher IL mass fractions, namely at w3 = 0.6, 0.7 and 0.8. These trends are similar to those presented for the systems water + ethanol + [hmim]Cl [9] and water + ethanol + [bmim]Cl [10]. The relative volatility of ethanol to water, r2,1,is related with the ratio of activity coefficients and the ratio of vapor pressures by the expression r2,1 =

4

5

6

7

(γ 2 / γ 1 ) ⋅ ( p2sat / p1sat ) , in which the ratio p2sat / p1sat is a weak function of temperature and has a small change from 2.19 to 2.30 in the present measurement. Therefore, the effect of IL on the relative volatility is mainly determined by the effect of IL on the activity coefficients. Although the decrease in γ1 is beneficial for enhancement of the relative volatility, it is not desirable. The effect of IL on the relative volatility r2,1 is shown in Fig. 3. While the addition of IL increases the relative volatility in the ethanol-rich region, showing a salting-out effect, there is a salting-in effect in the water-rich region. This should be related with the fact that γ2 decreases with increasing IL mass fraction more quickly than γ1 in the water-rich region, as shown in Fig. 1.

8

9

10

11

Seiler, M., Jork, C., Karvanou, A., Arlt, W., Hirsch, R., “Separation of azeotropic mixtures using hyperbranched polymers or ionic liquids”, AIChE J., 50 (10), 2439-2454 (2004). Marsh, K.N., Boxall, J.A., Lichtenthaler, R., “Room temperature ionic liquids and their mixtures—a review”, Fluid Phase Equilib., 219 (1), 93-98 (2004). Kato, R., Gmehling, J., “Systems with ionic liquids: Measurement of VLE and γ∞ data and prediction of their thermodynamic behavior using original UNIFAC, mod. UNIFAC(Do) and COSMO-RS(Ol)”, J. Chem. Thermodyn., 37 (6), 603-619 (2005). Lei, Z.G., Arlt, W., Wasserscheid, P., “Separation of 1-hexene and n-hexane with ionic liquids”, Fluid Phase Equilib., 241 (1-2), 290-299 (2006). Wang, J.F., Li, C.X., Wang, Z.H., Li, Z.J., Jiang, Y.B., “Vapor pressure measurement for water, methanol, ethanol, and their binary mixtures in the presence of an ionic liquid 1-ethyl-3-methylimidazolium dimethylphosphate”, Fluid Phase Equilib., 255 (2), 186-192 (2007). Orchilles, A.V., Miguel, P.J., Vercher, E., Martínez-Andreu, A., “Isobaric vapor-liquid equilibria for ethyl acetate + ethanol + 1-ethyl-3-methylimidazolium trifluoromethanesulfonate at 100 kPa”, J. Chem. Eng. Data, 52 (6), 2325-2330 (2007). Li, Q.S., Xing, F.Y., Lei, Z.G., Wang, B.H., Chang, Q.L., “Isobaric vapor-liquid equilibrium for isopropanol + water + 1-ethyl-3methylimidazolium tetrafluoroborate”, J. Chem. Eng. Data, 53 (1), 275-279 (2008). Westerholt, A., Liebert, V., Gmehling, J., “Influence of ionic liquids on the separation factor of three standard separation problems”, Fluid Phase Equilib., 280 (1), 56-60 (2009). Zhang, L.Z., Ge, Y., Ji, D. X., Ji, J. B., “Experimental measurement and modeling of vapor-liquid equilibrium for ternary systems containing ionic liquids: A case study for the system water + ethanol + 1-hexyl-3-methylimidazolium chloride”, J. Chem. Eng. Data, 54 (8), 2322-2329 (2009). Geng, W., Zhang, L.Z., Deng, D.S., Ge, Y., Ji, J.B., “Experimental measurement and modeling of vapor-liquid equilibrium for the ternary system water + ethanol + 1-butyl-3-methylimidazolium chloride”, J. Chem. Eng. Data, 55 (4), 1679-1683 (2010). Renon, H., Prausnitz, J. M., “Local compositions in thermodynamic excess functions for liquid mixtures”, AIChE J., 14 (1), 135-144 (1968).

708

12

13

14

15

16

Chin. J. Chem. Eng., Vol. 19, No. 4, August 2011

Ge, Y., Zhang, L.Z., Yuan, X.C., Geng, W., Ji, J.B., “Selection of ionic liquids as entrainers for separation of (water + ethanol)”, J. Chem. Thermodyn., 40 (8), 1248-1252 (2008). Gordon, C.M., “Synthesis of ionic liquids”, In: Ionic Liquids in Synthesis, Wasserscheid, P., Welton, T., eds., Wiley-VCH, Weiheim, 7-21 (2003). Zhang, L.Z., Ji, J.B., Deng, D.S., Shu, Z.J., Chen, X., “Method for manufacturing acetic acid type ionic liquids”, CN Pat., 200710070383.X (2007). Zhang, L.Z., Deng, D.S., Han, J.Z., Ji, D.X., Ji, J.B., “Isobaric vaporliquid equilibria for water + 2-propanol + 1-butyl-3-methylimidazolium tetrafluoroborate”, J. Chem. Eng. Data, 52 (1), 199-205 (2007). Zhang, L.Z., Han, J.Z., Deng, D.S., Ji, J.B., “Selection of ionic liquids as entrainers for separation of water and 2-propanol”, Fluid Phase Equilib., 255 (2), 179-185 (2007).

17

18 19

20

21

Zhang, L.Z., Yuan, X.C., Qiao, B.B., Qi, R.Z., Ji, J.B., “Isobaric vapor-liquid equilibria for water + ethanol + ethyl acetate + 1-butyl-3-methylimidazolium acetate at low water mole fractions”, J. Chem. Eng. Data, 53 (7), 1595-1601 (2008). Reid, R., Prausnitz, J.M., Poling, B., The Properties of Gases and Liquids, McGraw-Hill, New York, Appendix A (1987). Mock, B., Evans, L.B., Chen, C.C., “Thermodynamic representation of phase equilibria of mixed-solvent electrolyte”, AIChE J., 32 (10), 1655-1664 (1986). Vercher, E., Rojo, F.J., Martínez-Andreu, A., “Isobaric vapor-liquid equilibria for 1-propanol + water + calcium nitrate”, J. Chem. Eng. Data, 44 (6), 1216-1221 (1999). Gmehling, J., Onken, U., Rearey-Nies, J.R., Vapor-Liquid Equilibrium Data Collection, DECHEMA, Frankfurt, Vol. I, Part Ib (1988).