Vapor-liquid equilibria for the ternary system of carbon dioxide + ethanol + ethyl acetate at elevated pressures

(1)

Vapor–liquid equilibria for the ternary system of carbon

dioxide + ethanol + ethyl acetate at elevated pressures

Chen-Hsi Cheng, Yan-Ping Chen

∗

Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan, ROC Received 8 September 2005; received in revised form 18 January 2006; accepted 25 January 2006

Abstract

Vapor–liquid equilibrium (VLE) data for the ternary system of carbon dioxide, ethanol and ethyl acetate were measured in this study at 303.2, 308.2, and 313.2 K, and at pressures from 4 to 7 MPa. A static type phase equilibrium apparatus with visual sapphire windows was used in the experimental measurements. New VLE data for CO2in the mixed solvent were presented. These ternary VLE data at elevated pressures were

also correlated using either the modified Soave–Redlich–Kwong or Peng–Robinson equation of state, with either the van der Waals one-fluid or Huron–Vidal mixing model. Satisfactory correlation results are reported with temperature-independent binary parameters. It is observed that at 313.2 K and 7 MPa, ethanol can be separated from ethyl acetate into the vapor phase at all concentrations in the presence of high pressure CO2.

1. Introduction

Supercritical carbon dioxide is widely used in extraction, reaction and separation to minimize the amount of tradition-ally required organic solvents. For example, supercritical CO2 was used for the enzymatic reaction and, at the same time, in the product recovery by extraction[1]. Phase equilibrium data are thus important for the fluid mixtures involving supercritical CO2. In recent literature, the esterification of acetic acid with ethanol to form ethyl acetate was investigated using supercritical CO2

[2,3].Table 1lists the data sources for vapor–liquid equilibrium (VLE) of ternary systems involving high pressure CO2, alcohol and ester compounds. It is indicated[4,5]that ternary VLE data for such systems are still inadequate, more experimental data are required for engineering applications. We have measured the VLE data for CO2and esters using either the semi-flow or static apparatus[6,7]at pressures up to 13 MPa. We extend our previous work to ternary systems of CO2with ethanol and ethyl acetate in this study. Using supercritical CO2, the esterification of acetic acid with ethanol can be a green process with less amount of acid catalyst[3]. VLE data for the binary, ternary and four-component mixtures in this reaction process are useful for

∗_{Corresponding author. Fax: +886 2 2362 3040.}

E-mail address:[email protected](Y.-P. Chen).

the chemical process design and separation of products due to the supercritical extraction effect. It is also shown inTable 1

that the ternary VLE data for supercritical CO2 with ethanol and ethyl acetate are not yet available in literature. We intend to measure these VLE data and investigate the appropriate condi-tions that the esterification reaction products can be separated due to the supercritical extraction effect without the formation of the azeotropic compound. The experimental measurements of this study were carried out at temperatures of 303.2, 308.2, and 313.2 K, with the pressures range from 4 to 7 MPa. The measured data were also correlated using the equation of state (EOS) method. The modified Soave–Redlich–Kwong EOS[8]

and the widely used Peng–Robinson EOS [9]were employed in the correlation. The van der Waals one-fluid (VDW1) mixing model and the Huron–Vidal mixing model[10]were applied in the correlation. The optimally fitted binary interaction parame-ters are presented and the accuracies of correlated results from these models are compared.

2. Experimental

2.1. Chemicals

(2)

Table 1

List of the VLE data for CO2(1) + X (2) + Y (3) ternary systems in recent literature

X Y P (MPa) T (K) Reference Ethanol Methanol 2.0–8.0 313.2 [11] Ethanol p-Cresol 10.0–30.0 373.2 [24] Ethanol o-Cresol 10.0–20.0 373.2 [24] 1-Dodecanol 1-Hexadecanol 25.0–30.0 393.2 [25] Ethanol α-Tocochromanol 17.0 343.2 [26] m-Cresol Phenol 8.0–40.0 308.2 [27] o-Cresol p-Cresol 15.0–31.0 323.2–373.2 [24] Methanol 2-Methyl-2-butanol 2.0–8.2 313.2 [28]

2-Butanol Vinyl acetate 0.9–14.0 278.2–510.2 [5]

α-Cyano-m-phenoxybenzyl alcohol α-Cyano-m-phenoxybenzyl acetate 20.0 313.2 [1]

1-Octanol α-Cyano-m-phenoxybenzyl acetate 9.4–16.2 313.2 [1]

Methyl oleate Methyl linoleate 4.7–21.0 313.2–333.2 [29]

Ethyl acetate Isoamyl acetate 6.0 333.2 [18]

Acetic acid Ethanol 7.4–17.4 304.5–547.7 [4]

Acetic acid Water 7.0–17.0 313.0–433.0 [30,31]

Ethanol Water 8.1–18.5 313.2–343.2 [14,21]

Ethyl acetate Water 7.1–14.9 304.5–546.2 [4]

Absolute GR grade ethanol was purchased from Fisher Scien-tific Co. and its purity was better than 99.99 mass% from the gas chromatograph (GC) test. Ethyl acetate was purchased from Merck Co. and its purity was better than 99.8 mass% from the GC test. No water content was detected from the GC analyses. These chemicals were used without further purification. The properties of pure ethanol and ethyl acetate were measured in this study and the results are shown inTable 2. The refractive indices were measured at 293.2± 0.1 K using an Abbe refractometer, Atago 3T, with an accuracy of±0.0001. The densities of pure com-pounds were measured at 293.2± 0.1 K using the Anton Paar DMA 58 density meter with an accuracy of±0.1 kg/m3. From the comparison with literature values shown inTable 2, the puri-ties of pure chemicals are acceptable for VLE measurements. 2.2. Apparatus

The phase equilibrium apparatus has been described in our previous work[7]. It was a static-type apparatus in which the coexisting phases were recirculated, sampled, and analyzed. The apparatus mainly consists of three sections for the input of sam-ples, the high pressure equilibrium cell, and the analyses of the compositions in the equilibrium phases. The equilibrium cell has an internal volume of 320 mL, with three pairs of visual sapphire windows (Sitec, Switzerland) for visual observation of the phase behavior. The pressure in the equilibrium cell was measured by a Druck gauge (PDCR-4031, up to 700 bar) with a digital indicator (DPI-281). The temperature of the equilibrium cell was mea-sured by a K-type thermocouple with the resolution of 0.01 K.

The accuracies for the temperature and pressure measurements in our experimental ranges are±0.1 K and ±0.01 MPa, respec-tively. The liquid phase in the cell was recirculated using a magnetic pump (Micropump, series 180-HP) to reach phase equilibrium. The vapor sampling valve (Valco, 6UW) and the liquid sampling valve (Rheodyne, 7010) were used in this study. The GC (Shimadzu, 14B) column was packed with Porapak P. It was equipped with a thermal conductivity detector (TCD) for on-line analyses. Helium was used as the carrier gas with a flow rate of 45 mL/min. The temperatures for the TCD and the GC column were kept at 483.2 and 413.2 K, respectively. The equi-librium cell and sampling valves were all immersed in a water bath where the temperature was controlled at the desired value ±0.1 K. Heating tapes were used in the sampling line to avoid any condensation.

2.3. Calibration procedures

The volumes of the sampling loops were calibrated in this study using distilled water controlled at 298.2 K by following the similar procedures shown in literature[11,12]. The volumes were determined as 21.83 and 4.52␮L for the vapor and liquid sampling loops, respectively.

The calibration of GC was made by plotting the peak areas against the number of moles of the pure gas or liquid sample, as shown in previous literature [13,14]. For the liquid sam-ples, a digital balance (Mettler Toledo AX105, with accuracy of 2± 10−5g) was used to evaluate the weight and number of moles. The GC peak areas for ethanol and ethyl acetate were Table 2

Comparison of the measured refractive indices and densities for pure compounds with literature data

Compound nD_{(T = 293.2 K)} _{ρ (kg m}−3_{, T = 293.2 K)} _{GC purity (mass%)}

Experimental Literature Experimental Literature

Ethanol 1.3600 1.3594[32] 789.5 789.4[33] >99.99

(3)

plotted against their number of moles, and the third order poly-nomial curves were found satisfactory for the calibration of these pure liquid samples. Calibrations for the gas samples were per-formed when carbon dioxide was pressurized to a desired value at a given temperature. With the gas sampling valves connected to GC, the peak areas were obtained at various T and P condi-tions. The molar volumes for carbon dioxide at various T and P were obtained using the data base of NIST [15]. The third order polynomial curves were again found satisfactory at vari-ous isotherms for gas phase calibrations.

2.4. Experimental procedures

The experimental apparatus was firstly evacuated using a vac-uum pump. The liquid mixture of ethanol and ethyl acetate at a specific composition (ethanol mole fractions of 0.7143, 0.6667, 0.6, 0.5 and 0.2857) was degassed for 20 min before charging into the equilibrium cell that was immersed in a water bath. Car-bon dioxide was then fed into the system through a high pressure pump until the system reached a desired pressure. The liquid phase mixture was recirculated using a magnetic pump at a flow rate of 320 mL/min in order to enhance rapid phase equilibrium. At all experimental conditions, the equilibrium vapor and liquid phases were visually observed through the windows of the cell after 0.5 h of recirculation. The system was then settled for 3.5 h to reach the final equilibrium state. The final pressure was then recorded and the equilibrium compositions of the two phases were analyzed using GC. The equilibrium compositions were the averaged values of at least five repeated measurements. The sampling loops were then evacuated to remove any residual. The reproducibility for the measured vapor and liquid mole fractions were estimated to be±0.0001 and ±0.0002, respectively. 2.5. VLE calculation models

The experimentally measured VLE data were further cor-related using the equation of state (EOS) models. The equal fugacity criterion was employed where the fugacity for each component in the equilibrium phases was calculated using the cubic type EOS. It has been presented by Chrisochoou et al.

[1]that the modified Soave–Redlich–Kwong (MSRK) EOS[8]

can satisfactorily correlate the ternary and multicomponent VLE data for systems of the transesterification process involving supercritical CO2using only the binary interaction parameters determined from binary VLE data. It is intended to justify their conclusion for the ternary VLE data measured in this study. Besides the MSRK EOS, the widely used Peng–Robinson (PR) EOS [9] was also employed in correlating our experimental results. The MSRK EOS has the following form:

P = RT v − b − a v(v + b) (1) a = 0.42748R2Tc2 Pc α(T ) (2) b = 0.08664RTc Pc (3) Table 3

Pure component properties used in this study

Component Tc[34](K) Pc[34](bar) ω[34] m[18] n[18] Carbon dioxide 304.19 73.82 0.228 0.6605 0.2054 Ethanol 516.25 63.84 0.637 1.1639 0.3923 Ethyl acetate 523.20 38.30 0.361 0.7460 0.3189 α(T ) = 1 + (1 − Tr) m + n Tr (4) where the parameters m and n are available for many pure fluids

[8]. The PR EOS has the following form: p = RT v − b − a v(v + b) + b(v − b) (5) a = 0.45724R2Tc2 Pc α(T ) (6) b = 0.07780RTc Pc (7) α(T ) = 1+ (0.37464 + 1.54226ω − 0.26992ω2) 1− T Tc 2 (8) The EOS parameters were determined from the constants of pure fluids that are listed inTable 3. For mixture calculations, the following mixing models were employed. For the van der Waals one-fluid (VDW1) mixing model, the EOS parameters were determined by:

am= xixj(aiaj)0.5(1− kij) (9) bm= xibi (10)

where kijis the binary interaction parameter and was determined by regressing the experimental VLE data. Another mixing model presented by Huron and Vidal[10]was also used in this study. In this mixing model, the excess Gibbs energy calculated from an EOS is set equal to that from an activity coefficient model at an infinite pressure limit. Applying the NRTL activity coefficient model[16], the EOS parameters were evaluated by the following equations: am= bm n i=1 xi × ai bi − 1 ln 2 _n j=1xjCjibj exp(−αjiCji/(RT )) _n j=1xjbj exp(−αjiCji/(RT )) (11) bm= n i=1 xibi (12)

where the non-randomness factorα was taken as 0.3 in our cor-relation. The binary parameters Cijand Cjiin the NRTL model were determined by regressing the experimental VLE data.

(4)

Table 4

The correlation results for the VLE data of binary systems using the PR and the MSRK EOS with the VDW1 mixing model

EOS T (K) P (MPa) k12 AADx1(%) AADy1(%) Total points Reference

CO2(1) + ethanol (2) PR 304.2–308.2 3.75–7.67 0.084 0.43 0.10 11 [17] MSRK 304.2–308.2 3.75–7.67 0.080 0.40 0.11 11 [17] CO2(1) + ethyl acetate (2) PR 313.2 0.92–7.88 −0.016 0.87 0.15 10 [18] MSRK 313.2 0.92–7.88 −0.019 0.88 0.17 10 [18]

Ethyl acetate (1) + ethanol (2)

PR 313.2–333.2 0.02–0.64 0.021 0.02 5.43 37 [19] MSRK 313.2–333.2 0.02–0.64 0.024 0.02 5.55 37 [19] AADx1= (100/N) |(xcal 1 − x exp 1 )/x exp 1 |; AADy1= (100/N) |(ycal 1 − y exp 1 )/y exp

1 |, where N is the number of data points.

We have correlated the VLE data shown in literature[17–19]

for three binary systems of CO2+ ethanol, CO2+ ethyl acetate and ethyl acetate + ethanol. The optimally fitted temperature-independent binary interaction parameters in either the VDW1 or Huron–Vidal mixing model were evaluated through flash cal-culation by minimizing the following objective function: obj=

n

i=1

[|(ycal_i − yexp_i )| + |(xcal_i − xexp_i )|] (13)

The correlation results are shown inTables 4 and 5for the VDW1 or Huron–Vidal mixing model, respectively. Generally, both the MSRK and PR EOS yield the comparable accuracy in correlating the binary VLE data. The Huron–Vidal mixing model with the NRTL activity coefficient model has two adjustable parameters, and presents relatively superior results to those from the simple VDW1 mixing model with only a single parame-ter. These optimally fitted binary interaction parameters were directly applied in predicting the ternary VLE results and com-paring with the experimental measured data.

3. Results and discussion

The measured VLE data in this study for the ternary sys-tem CO2 (1) + ethanol (2) + ethyl acetate (3) at 303.2, 308.2, and 313.2 K are presented in Tables 6–8, respectively. Five compositions for ethanol in the feed of ethanol + ethyl acetate were included for each isotherm in the experiments. For each

Table 6

Vapor–liquid equilibrium data for the ternary system of CO2 (1) + ethanol

(2) + ethyl acetate (3) at T = 303.2 K P (MPa) x1 x2 x3 y1 y2 y3 4.14 0.3200 0.5100 0.1700 0.9940 0.0050 0.0010 5.02 0.4261 0.4218 0.1521 0.9950 0.0044 0.0006 6.00 0.5374 0.3191 0.1435 0.9955 0.0041 0.0004 6.92 0.7261 0.2124 0.0615 0.9959 0.0039 0.0002 4.12 0.3262 0.4681 0.2057 0.9938 0.0042 0.0020 5.03 0.4431 0.3720 0.1849 0.9941 0.0041 0.0018 6.04 0.5450 0.2765 0.1785 0.9950 0.0034 0.0016 6.92 0.7593 0.1691 0.0716 0.9958 0.0033 0.0009 4.11 0.4190 0.3481 0.2329 0.9934 0.0039 0.0027 5.02 0.4948 0.2952 0.2100 0.9940 0.0035 0.0025 6.04 0.5651 0.2433 0.1916 0.9945 0.0033 0.0022 6.97 0.7840 0.1276 0.0884 0.9951 0.0029 0.0020 4.19 0.4931 0.2750 0.2319 0.9936 0.0027 0.0037 5.03 0.6063 0.2067 0.1870 0.9938 0.0025 0.0037 6.03 0.6240 0.1960 0.1800 0.9943 0.0023 0.0034 7.03 0.8166 0.1040 0.0794 0.9948 0.0018 0.0034 4.16 0.6883 0.1000 0.2117 0.9939 0.0010 0.0051 5.09 0.7264 0.0970 0.1766 0.9942 0.0010 0.0048 6.09 0.8250 0.0800 0.0950 0.9947 0.0008 0.0045 7.03 0.8806 0.0450 0.0744 0.9954 0.0006 0.0040 feed composition, VLE data were measured at the given tem-perature and at four pressures ranging from 4 to 7 MPa. It is observed that the vapor phase is mainly CO2with mole frac-tion >0.99. The mole fracfrac-tions of ethanol and ethyl acetate in Table 5

The correlation results for the VLE data of binary systems using the PR and the MSRK EOS with the Huron–Vidal mixing model

EOS T (K) P (MPa) α C12(J/mol) C21(J/mol) AADx1(%) AADy1(%) Total points Reference

CO2(1) + ethanol (2) PR 304.2–308.2 3.75–7.67 0.3 4759.95 −495.42 0.36 0.10 11 [17] MSRK 304.2–308.2 3.75–7.67 0.3 3641.27 −4536.05 0.33 0.11 11 [17] CO2(1) + ethyl acetate (2) PR 313.2 0.92–7.88 0.3 −1469.44 1387.55 0.43 0.15 10 [18] MSRK 313.2 0.92–7.88 0.3 391.98 72.96 0.59 0.17 10 [18]

Ethyl acetate (1) + ethanol (2)

PR 313.2–333.2 0.02–0.64 0.3 2049.93 4336.29 0.01 2.18 37 [19] MSRK 313.2–333.2 0.02–0.64 0.3 2053.90 4334.81 0.01 2.17 37 [19] AADx1= (100/N) |(xcal 1 − x exp 1 )/x exp 1 |; AADy1= (100/N) |(ycal 1 − y exp 1 )/y exp

(5)

Table 7

(2) + ethyl acetate (3) at T = 308.2 K P (MPa) x1 x2 x3 y1 y2 y3 4.16 0.2793 0.5726 0.1481 0.9934 0.0055 0.0011 5.02 0.3828 0.4860 0.1312 0.9940 0.0050 0.0010 6.08 0.4815 0.4140 0.1045 0.9949 0.0042 0.0009 6.97 0.6400 0.3000 0.0600 0.9953 0.0039 0.0008 4.15 0.2977 0.5010 0.2013 0.9930 0.0049 0.0021 5.04 0.4204 0.4054 0.1742 0.9938 0.0042 0.0020 6.04 0.5366 0.3004 0.1630 0.9945 0.0039 0.0016 6.96 0.6338 0.2434 0.1228 0.9949 0.0037 0.0014 4.15 0.3110 0.4134 0.2756 0.9927 0.0040 0.0033 5.02 0.4558 0.2953 0.2489 0.9933 0.0036 0.0031 6.09 0.5863 0.2168 0.1969 0.9940 0.0031 0.0029 6.98 0.6876 0.1703 0.1421 0.9945 0.0028 0.0027 4.19 0.5446 0.1706 0.2848 0.9925 0.0030 0.0045 5.09 0.6628 0.1295 0.2077 0.9930 0.0028 0.0042 6.10 0.7948 0.0792 0.1260 0.9936 0.0024 0.0040 6.98 0.8582 0.0532 0.0886 0.9943 0.0019 0.0038 4.16 0.7100 0.0500 0.2400 0.9920 0.0010 0.0070 5.09 0.7900 0.0400 0.1700 0.9930 0.0010 0.0060 6.09 0.8500 0.0400 0.1100 0.9932 0.0009 0.0059 6.99 0.8966 0.0300 0.0734 0.9942 0.0008 0.0050

the vapor phase increase with temperature and decrease with pressure.

The optimally fitted binary interaction parameters obtained from correlating the binary mixtures were directly used to predict the VLE results for the ternary system. The calculation results, however, are not satisfactory as shown inTable 9. One possi-ble reason is that the VLE data for the binary mixture of ethyl acetate + ethanol were at a low pressure range. If the binary inter-action parameter for ethyl acetate + ethanol was taken as the only adjustable parameter in fitting the ternary VLE data, acceptable results are obtained as shown inTable 9.Table 9also shows that satisfactory correlation results are presented by fitting the temperature-independent binary interaction parameters directly to the ternary VLE data. It is not always possible to obtain a set of binary interaction parameters that could yield accurate predic-tion for both binary and multicomponent systems. It is indicated in literature[20]that fitting the EOS to the ternary VLE data gives better correlation.

Table 8

(2) + ethyl acetate (3) at T = 313.2 K P (MPa) x1 x2 x3 y1 y2 y3 4.11 0.2230 0.6280 0.1490 0.9890 0.0092 0.0018 5.05 0.3145 0.5450 0.1405 0.9900 0.0088 0.0012 6.07 0.4224 0.4641 0.1135 0.9903 0.0086 0.0011 7.06 0.5396 0.3514 0.1090 0.9904 0.0085 0.0011 4.18 0.2300 0.5590 0.2110 0.9874 0.0089 0.0037 5.08 0.3682 0.4562 0.1756 0.9880 0.0085 0.0035 6.04 0.4634 0.3689 0.1677 0.9890 0.0080 0.0030 7.07 0.5600 0.2902 0.1498 0.9893 0.0079 0.0028 4.16 0.3000 0.4230 0.2770 0.9873 0.0075 0.0052 5.08 0.4000 0.3500 0.2500 0.9874 0.0075 0.0051 6.02 0.5500 0.2500 0.2000 0.9877 0.0075 0.0048 7.07 0.6006 0.2160 0.1834 0.9878 0.0074 0.0048 4.13 0.4718 0.2773 0.2509 0.9876 0.0050 0.0074 5.03 0.5456 0.2428 0.2116 0.9885 0.0050 0.0065 6.04 0.6600 0.1600 0.1800 0.9890 0.0049 0.0061 6.98 0.7400 0.1000 0.1600 0.9895 0.0047 0.0058 4.15 0.5957 0.1619 0.2424 0.9878 0.0037 0.0085 5.09 0.6892 0.1279 0.1829 0.9900 0.0021 0.0079 6.06 0.7915 0.0786 0.1299 0.9905 0.0019 0.0076 6.98 0.8969 0.0212 0.0819 0.9910 0.0019 0.0071

Graphical presentations of the ternary VLE data are shown in Figs. 1 and 2at 303.2 K and two isobars of 4 and 7 MPa.

Fig. 3shows the VLE data at 7 MPa but at the other temperature of 313.2 K. These three figures show that the liquid composi-tions of ethanol and ethyl acetate increase with temperature but decrease with pressure. The solid tie lines represent the calcu-lated results using the PR EOS and the VDW1 mixing model with the binary interaction parameters fitted to the ternary VLE data. Satisfactory agreement between the experimental and cor-related results is demonstrated.

To illustrate the effect of separation for ethanol (2) and ethyl acetate (3) in the presence of high pressure CO2, the separation factorsα23= (y2/x2)/(y3/x3) were examined from all experimen-tal results. As discussed in previous literature[21], component 2 can be separated into the vapor phase withα23greater than unity. In our experimental range, theα23values are greater than unity at 313.2 K and 7 MPa over the entire concentration range. As shown in literature[22], a solvent-free y*− x*plot was used to Table 9

Correlation results for the ternary VLE data of CO2(1) + ethanol (2) + ethyl acetate (3)

EOS Mixing model Predicted results Adjustment of BIP for (2) + (3)a Fitting BIP to ternary VLE datab AADx1(%) AADy1(%) AADx1(%) AADy1(%) AADx1(%) AADy1(%)

PR VDW1 8.95 0.51 1.60 0.19 1.50 0.19

Huron–Vidal 6.68 0.21 1.55 0.18 1.47 0.19

MSRK VDW1 9.09 0.53 1.75 0.20 1.53 0.20

Huron–Vidal 6.69 0.22 1.65 0.19 1.48 0.18

BIP, binary interaction parameters.

a_{PR EOS: k}

23=−0.180, C23= 0.584, C32= 3.000; MSRK EOS: k23=−0.166, C23=−0.038, C32= 1.374, Cij[=] kJ/mol. b _{PR EOS: k}

12= 0.096, k13=−0.009, k23=−0.164, C12= 12.761, C21= 1.050, C13=−0.359, C31= 2.356, C23= 3.412, C32=−0.242; MSRK EOS: k12= 0.099,

(6)

Fig. 1. Vapor–liquid equilibria for carbon dioxide (1) + ethanol (2) + ethyl acetate (3) at T = 303.2 K and P = 4 MPa (, : mole fractions of the liquid and vapor phases measured in this study;䊉: mole fraction of the liquid phase from literature[35,36]).

Fig. 4. Plot of the CO2-freey∗2− x∗2 curves at 313.2 K for the ternary

mix-ture CO2(1) + ethanol (2) + ethyl acetate (3) (experimental data:, P = 5 MPa;

䊉, P = 7 MPa; –, calculated results using the PR EOS with the VDW1 mixing model).

illustrate the separation effect. The solvent-free mole fractions are defined as:

z∗_i = ni j=CO2nj

(14) The plot of the CO2-free basis (y∗₂− x∗₂) curves is shown inFig. 4at 313.2 K and at two pressures of 5 and 7 MPa. At atmospheric pressure, the binary mixture of ethanol and ethyl acetate has an azeotrope with ethanol mole fraction of 0.46[23]. At 313.2 K and 7 MPa, the relative volatility of ethanol to ethyl acetate is greater than unity in the entire composition range and no azeotrope is found. No such separation effect is observed at all concentration range for other temperatures and pressures investigated in this study.

4. Conclusions

Experimental VLE data for the ternary system of car-bon dioxide with ethanol and ethyl acetate are reported at 303.2, 308.2, and 313.2 K and pressures from 4 to 7 MPa. The modified Soave–Redlich–Kwong or Peng–Robinson EOS with either the van der Waals one-fluid or Huron–Vidal mix-ing model were used to correlate the ternary experimental VLE data. With the temperature-independent binary parameters fit-ted to the ternary VLE data, satisfactory correlation results are obtained for both EOS and mixing models. The experimen-tal data are analyzed on the CO2-free basis. It is shown that at 313.2 K and 7 MPa, ethanol can be separated from ethyl acetate at all concentrations in the presence of high pressure CO2.

(7)

List of symbols

a, b parameters in the equation of state

C binary interaction parameter in the NRTL model f fugacity

k binary interaction parameters in the mixing model n number of mole

P pressure R gas constant T temperature

x mole fraction in the liquid phase y mole fraction in the vapor phase Greek letter

α non-randomness factor in the NRTL model Subscripts c critical properties i, j component i or j m mixture r reduced properties 1, 2 component 1 or 2 Superscripts

cal calculated value exp experimental data * _{CO2-free basis}

Acknowledgement

The authors are grateful to the National Science Council, ROC for supporting this research.

References

[1] A.A. Chrisochoou, K. Schaber, K. Stephen, J. Chem. Eng. Data 42 (1997) 558–561.

[2] L.A. Blanchard, J.F. Brennecke, Green Chem. 3 (2001) 17–19. [3] Z. Hou, B. Han, X. Zhang, H. Zhang, Z. Liu, J. Phys. Chem. B 105

(2001) 4510–4513.

[4] T. Hu, Z. Qin, G. Wang, X. Hou, J. Wang, J. Chem. Eng. Data 49 (2004) 1809–1814.

[5] R.M.M. Stevens, A. Bakx, P.R.H. van der Neut, T.W. de Loos, J. de Swaan Arons, J. Chem. Eng. Data 42 (1997) 1280–1284.

[6] W.H. Hwu, C.H. Cheng, M. Tang, Y.P. Chen, Fluid Phase Equilib. 215 (2004) 237–244.

[7] C.H. Cheng, Y.P. Chen, Fluid Phase Equilib. 234 (2005) 77–83. [8] J.A. Sandarusi, A.J. Kidnay, V.F. Yesavage, Ind. Eng. Chem. Process

Des. Dev. 25 (1986) 957–963.

[9] D.Y. Peng, D.B. Robinson, Ind. Eng. Chem. Fundam. 15 (1976) 59–64. [10] M.J. Huron, J. Vidal, Fluid Phase Equilib. 3 (1979) 255–271. [11] J.H. Yoon, H.S. Lee, H. Lee, J. Chem. Eng. Data 38 (1993) 53–55. [12] T. Laursen, P. Rasmussen, S.I. Andersen, J. Chem. Eng. Data 47 (2002)

198–202.

[13] K. Suzuki, H. Sue, M. Itou, R.L. Smith, H. Inomata, K. Arai, S. Saito, J. Chem. Eng. Data 35 (1990) 63–66.

[14] J.H. Yoon, H. Lee, B.H. Chung, Fluid Phase Equilib. 102 (1994) 287–292.

[15] http://webbook.nist.gov/chemistry.

[16] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135–144.

[17] S. Takishima, K. Saiki, K. Arai, S. Saito, J. Chem. Eng. Jpn. 19 (1986) 48–56.

[18] A. Chrisochoou, K. Schaber, U. Bolz, Fluid Phase Equilib. 108 (1995) 1–14.

[19] P.S. Murti, M. Van Winkle, Chem. Eng. Data Ser. 3 (1958) 72–81. [20] R. Ruivo, A. Paiva, P. Simoes, J. Supercrit. Fluids 29 (2004) 77–85. [21] J.S. Lim, Y.Y. Lee, H.S. Chun, J. Supercrit. Fluids 7 (1994) 219–230. [22] P. Traub, K. Stephan, Chem. Eng. Sci. 45 (1990) 751–758.

[23] M. Kato, H. Konishi, M. Hirata, J. Chem. Eng. Data 15 (1970) 435–439. [24] O. Pfohl, A. Pagel, Fluid Phase Equilib. 157 (1999) 53–79.

[25] I.F. Holscher, M. Spee, G.M. Schneider, Fluid Phase Equilib. 49 (1989) 103–113.

[26] A. Birtigh, J. Stoldt, G. Brunner, J. Supercrit. Fluids 8 (1995) 162–166. [27] G. Di Giacomo, S. Brandani, V. Brandani, G. Del Re, Fluid Phase

Equilib. 95 (1994) 313–327.

[28] H.S. Lee, S.Y. Mun, H. Lee, Fluid Phase Equilib. 167 (2000) 131–144. [29] M. Zou, Z.R. Yu, S.S.H. Rizvi, J.A. Zollweg, J. Supercrit. Fluids 3

(1990) 85–90.

[30] K.M. Dooley, A.W. Cain, F.C. Knopf, J. Supercrit. Fluids 11 (1997) 81–89.

[31] B. Rumpf, J. Xia, G. Maurer, Ind. Eng. Chem. Res. 37 (1998) 2012–2019.

[32] T.E. Daubert, R.P. Danner, Physical and Thermodynamic Properties of Pure Chemicals: Data Compilation, Hemisphere, New York, 1989. [33] D.R. Lide, H.P.R. Frederikse, CRC Handbook of Chemistry and Physics,

80th ed., CRC Press Inc., Boca Raton, FL, 1999.

[34] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gas and Liquids, 5th ed., McGraw Hill Inc., New York, 2001.

[35] C.J. Chang, K.L. Chiu, C.Y. Day, J. Supercrit. Fluids 12 (1998) 223–237. [36] Z. Wanger, J. Pavlicek, Fluid Phase Equilib. 97 (1994) 119–126.