High pressure vapor–liquid equilibria for the systems thiophene + nonane + CO2, and thiophene + decane + CO2

Fluid Phase Equilibria 236 (2005) 178–183

High pressure vapor–liquid equilibria for the systems thiophene + nonane + CO2, and thiophene + decane + CO2 Octavio Elizalde-Solis, Luis A. Galicia-Luna ∗ Instituto Polit´ecnico Nacional, ESIQIE, Laboratorio de Termodin´amica, Edif. Z-6, Secc. 6, 1er piso, UPALM, C.P. 07738, Lindavista, M´exico, D.F., Mexico Received 21 June 2005; accepted 21 June 2005 Available online 8 August 2005

Abstract The aim of this work was to obtain phase equilibrium data of thiophene in CO2 in the presence of an alkane as cosolute. These data are needed for the design of an extraction process of sulfur compounds from a model fuel. Isothermal vapor–liquid equilibria (VLE) for the thiophene (1) + nonane (2) + CO2 (3), and thiophene (1) + decane (2) + CO2 (3) ternary systems were obtained at temperatures from 333 to 383 K and pressures ranging from 1.9 to 15.3 MPa. The initial loading for the solutes in the equilibrium cell was on a solvent-free basis at volume ratios of 1:4.52 and 1:1.25, for thiophene/nonane and thiophene/decane, respectively. Experimental solubilities of thiophene obtained in these systems were compared with those previously measured for the binary system thiophene + CO2 . The distribution coefficients and separation factors for these systems are also reported. © 2005 Elsevier B.V. All rights reserved. Keywords: Vapor–liquid equilibria; Equation of state; Separation factor

1. Introduction The development of a supercritical fluid process that could be applied in the extraction of sulfur compounds from fuels is an interesting option. However, in the process design, it is important to have accurate data on some properties such as phase behavior and physical properties of the compounds involved in the studied system. The presence of thiophene, benzothiophene, dibenzothiophene and their alkyl-derivative compounds in a commercial Mexican gasoline has been already identified by gas chromatograph analysis [1]. The phase behavior for dibenzothiophene + naphthalene + CO2 reported by Mitra et al. [2] has been the unique system that relates the solubility of this sulfur compound in CO2 in the presence of a hydrocarbon. Some studies have started and are focused onto getting solubilities of thiophene in supercritical CO2 , and in the presence of a polar cosol∗

Corresponding author. Tel.: +52 55 5729 6000x55133; fax: +52 55 5586 2728. E-mail addresses: [email protected], [email protected] (Luis A. Galicia-Luna). 0378-3812/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.fluid.2005.06.029

vent [1]. Following this study, experimental solubilities of thiophene + nonane, and thiophene + decane binary mixtures in CO2 from 333 to 383 K are reported in this work. The solute mixtures were used as model fuels in order to analyze the separation of thiophene with CO2 as solvent. Therefore, it is necessary to study how nonane and decane modify the distribution of thiophene in the phase equilibrium. The experimental solubilities were compared with those predicted with the Peng-Robinson equation of state (EoS) [3] and the classical mixing rules for the P − x ternary data. The interaction parameters (kij ) used for the calculations of ternary systems were correlated from the VLE for binary systems reported in literature [1,4,5]. The calculated solubilities were in agreement with those experimental data.

2. Experimental 2.1. Materials CO2 with a minimum purity of 99.995 mol% and helium with a minimum purity of 99.998 mol% were purchased from

Octavio Elizalde-Solis, Luis A. Galicia-Luna / Fluid Phase Equilibria 236 (2005) 178–183

179

Fig. 2. TCD calibration curve for decane.

Fig. 1. Flow diagram of the equilibrium cell: AB air bath; CT titanium cap; CNi connecting nuts; ECT titanium equilibrium cell; MCS movable capillary sampler; MR magnetic rod; MSD magnetic stirring device; OR Oring; PT pressure transducer; PTPi platinum temperature probe; TRi thermal regulator i; and Vi shut-off valve i.

Infra, M´exico. Thiophene (99+% purity), nonane (99+%, purity) and anhydrous decane (99+% purity) were supplied by Aldrich Chemical Co. Inc, Wisconsin, USA. These substances were degassed and vigorously stirred under vacuum. 2.2. Apparatus description The solubility measurements were performed in an apparatus based on the static-analytical method. It contains an equilibrium cell (ECT) with a volume of ∼100 cm3 , which can be operated to a maximum temperature of 673.15 K and pressures up to 60 MPa. The ECT showed in Fig. 1 is connected on-line to a gas chromatograph (GC) for the composition analyses. About 1 ␮L of sample is taken from the ECT using a ROLSITM sampler-injector [6] and sent to the GC through a thermo-regulated transferring line. This apparatus has been described by Elizalde-Solis et al. [7]. Temperature was measured using two platinum probes Pt100 (Specitec, France) immersed in two thermometric wells at the top and bottom of the cell. The temperature probes were connected to a digital indicator (Model F250, Automatic Systems Laboratories, USA). A thermoregulated pressure transducer (Model PDCR 910-1756, Druck, UK) with a digital display (Model DPI 145, Druck, UK) was used to measure the operating pressure in the ECT.

series II) equipped with a thermal conductivity detector (TCD). A packed column (Porapak Q, Alltech) with a 1.2m by 0.32 cm diameter tubing allowed the chromatographic separation of the mixture constituents. Helium was the carrier gas, and was set to 30 ml/min. An area–concentration calibration curve for the TCD detector was previously made by injecting external standards of each compound. The calibration curves obtained were used to quantify the composition of the phases. The TCD calibration for decane is showed in Fig. 2, and the experimental uncertainty for this component did not exceed ±0.8% as it is shown in Fig. 3. 2.4. Experimental procedure After the pressure transducer, temperature sensors and the TCD from the GC were calibrated, the ECT was loaded with a sufficient amount of the solute mixture (∼50 cm3 ) on a solvent-free basis. The first system was originally loaded at a volume ratio of 1:4.52 for thiophene and nonane. For the second system, the liquid mixture consisted of thiophene and decane with an initial volume ratio of 1:1.25. The liquid mixture was degassed and vigorously stirred under vacuum with the magnetic rod (MR). After degassing for 20 min, the system was isolated. Then, CO2 was pumped into the equilib-

2.3. Composition analysis Samples of liquid and vapor phase were analyzed by means of a gas chromatograph (GC, Hewlett-Packard 5890

Fig. 3. Experimental decane mole number uncertainty in the TCD analysis.

180

Octavio Elizalde-Solis, Luis A. Galicia-Luna / Fluid Phase Equilibria 236 (2005) 178–183

rium cell from the supply tank using a syringe pump (Model 100DM, Isco, USA) up to a required pressure. Afterward, the temperature in the cell was regulated by air bath (AB). Phase equilibrium measurements started when the pressure and temperature were kept constant. It usually takes 120 min to reach the equilibrium. Isothermal VLE data were measured with continuous increments in pressure by adding CO2 into the ECT. After the pressure was increased, the apparatus takes about 20 min to achieve the new equilibrium conditions. At each temperature and pressure, at least five consecutive samples of both liquid or vapor phase were taken in order to check for the reproducibility of the measurements, and to perform the mole fraction (x, y) error analysis. The phase composition uncertainties were about ± 1%. Estimated uncertainties for temperature (T) and pressure (P) were within ± 0.03 K, and ± 0.04%, respectively.

3. Results and discussion 3.1. Experimental data The reliability of the apparatus to measure phase equilibrium data was previously tested and described [1,7]. Isothermal VLE for the thiophene (1) + nonane (2) + CO2 (3) system were obtained at 334.31, 363.72, and 383.32 K. These data are presented in Table 1. The experimental VLE for the thiophene (1) + decane (2) + CO2 (3) system were performed at 334.40, 363.78, and 383.27 K, and are reported in Table 2. In both ternary systems, the mole fraction solubilities increased with temperature increasing. The isothermal solubilities had the same trend and were dependant on the pressure conditions. The solubility of thiophene (y1 ) and the alkanes (y2 ) in CO2 had a high increase when this solvent was at supercritical conditions. As shown in Table 1, the solubilities of thiophene were higher than those obtained for nonane (y2 ) in CO2 up to its critical pressure (7.374 MPa), and above this pressure, nonane had better solubility in CO2 than thiophene. In the case of decane, a different behavior was observed with the pressure settings. The solubilities of thiophene were higher than decane in all the studied range of pressure as can be seen in Table 2. The solubilities of thiophene (y1 ) obtained in these ternary systems were compared with those obtained for the binary system thiophene + CO2 reported in literature [1]. The Fig. 4 shows a comparison of this behavior at ∼383 K. It can be observed that the solubility of thiophene in CO2 decreases in the ternary systems due to the presence of the alkane.

Table 1 Experimental VLE for thiophene (1) + nonane (2) + CO2 (3) system P (MPa)

x1

x2

y1

y2

T = 334.31 K 4.172 5.083 6.102 7.045 8.084 9.003 10.066 10.464

0.1453 0.1299 0.1143 0.0959 0.0777 0.0586 0.0345 0.0227

0.5091 0.4588 0.4063 0.3502 0.2886 0.2239 0.1196 0.0746

0.0063 0.0060 0.0057 0.0057 0.0060 0.0064 0.0086 0.0107

0.0028 0.0028 0.0029 0.0032 0.0041 0.0060 0.0168 0.0230

T = 363.72 K 4.035 5.000 6.048 7.023 7.992 8.962 10.011 10.987 12.017 13.036 13.675

0.1581 0.1491 0.1372 0.1284 0.1168 0.1067 0.0950 0.0815 0.0695 0.0537 0.0370

0.5706 0.5299 0.4820 0.4433 0.4040 0.3659 0.3219 0.2757 0.2314 0.1558 0.0962

0.0138 0.0124 0.0116 0.0115 0.0112 0.0114 0.0123 0.0135 0.0152 0.0186 0.0284

0.0067 0.0065 0.0065 0.0068 0.0076 0.0084 0.0108 0.0141 0.0195 0.0306 0.0637

0.0419 0.0300 0.0247 0.0220 0.0203 0.0195 0.0187 0.0185 0.0180 0.0190 0.0193 0.0210 0.0243 0.0282 0.0332 0.0362

0.0159 0.0126 0.0114 0.0110 0.0109 0.0116 0.0124 0.0137 0.0152 0.0186 0.0216 0.0278 0.0398 0.0559 0.0771 0.0889

T = 383.32 K 1.980 3.145 4.176 5.220 6.066 7.032 8.093 9.168 10.147 11.197 12.115 13.034 14.062 14.753 15.114 15.184

0.1991 0.1851 0.1682 0.1583 0.1433 0.1292 0.1157 0.1042 0.0939 0.0830 0.0688 0.0568 0.0448

0.5584 0.5250 0.4874 0.4596 0.4246 0.3876 0.3554 0.3210 0.2940 0.2617 0.2194 0.1802 0.1221

3.2. Modeling The Peng-Robinson equation of state (P-R EoS) [3] with classical mixing rules was used to predict the phase behavior of these ternary systems. The explicit form of this equation

Fig. 4. Solubility of thiophene in CO2 at ∼383 K for the systems: (䊉) thiophene + CO2 [1], (
) thiophene + nonane + CO2 , and () thiophene + decane + CO2 .

Octavio Elizalde-Solis, Luis A. Galicia-Luna / Fluid Phase Equilibria 236 (2005) 178–183 Table 2 Experimental VLE for thiophene (1) + decane (2) + CO2 (3) system

Table 3 Physical properties of pure compounds [8,9]

P (MPa)

x1

x2

y1

y2

T = 334.40 K 2.564 4.025 5.088 6.100 7.005 8.096 9.013 10.011 10.538

0.4125 0.3673 0.3287 0.2796 0.2433 0.1925 0.1445 0.0868 0.0509

0.3634 0.3178 0.2861 0.2517 0.2234 0.1804 0.1399 0.0783 0.0428

0.0169 0.0140 0.0129 0.0124 0.0127 0.0137 0.0148 0.0199 0.0276

0.0012 0.0011 0.0011 0.0011 0.0012 0.0018 0.0028 0.0078 0.0158

T = 363.78 K 3.050 3.959 4.989 6.181 6.964 8.113 9.011 9.993 11.061 12.027 13.112 13.746

0.4270 0.4058 0.3745 0.3441 0.3239 0.2922 0.2691 0.2466 0.2081 0.1724 0.1306 0.0928

0.3792 0.3520 0.3250 0.2946 0.2757 0.2462 0.2259 0.1994 0.1729 0.1438 0.0986 0.0621

0.0352 0.0301 0.0275 0.0252 0.0246 0.0252 0.0255 0.0276 0.0314 0.0346 0.0473 0.0648

0.0025 0.0024 0.0025 0.0026 0.0027 0.0031 0.0036 0.0047 0.0069 0.0098 0.0193 0.0351

T = 383.27 K 2.772 4.145 5.075 6.170 7.213 8.090 9.110 10.075 11.087 12.035 13.070 14.058 15.011 15.366

0.5131 0.4571 0.4325 0.4060 0.3776 0.3522 0.3223 0.2969 0.2738 0.2367 0.2069 0.1751 0.1369 0.1118

0.3362 0.3203 0.3044 0.2864 0.2698 0.2542 0.2349 0.2188 0.1984 0.1798 0.1609 0.1383 0.1026 0.0778

0.0044 0.0040 0.0040 0.0041 0.0043 0.0050 0.0057 0.0067 0.0082 0.0100 0.0137 0.0191 0.0331 0.0470

0.9301 0.9448 0.9500 0.9541 0.9557 0.9548 0.9539 0.9525 0.9497 0.9465 0.9376 0.9270 0.8985 0.8715

can be written as follows: a RT − P= v − b v(v + b) + b(v − b)

(1)

where a and b are related to: a(T ) = 0.45724 b = 0.07780

181

R2 Tc2 α(Tr ) Pc

RTc Pc

(2) (3)

and α(Tr ) is expressed in terms of the acentric factor (ω): α(Tr )
2  . = 1 + (0.37464 + 1.54226ω−0.26992ω2 ) 1−Tr1/2 (4) Physical properties of pure compounds taken from literature [8,9] are presented in Table 3. In the case of mixtures, the a

Compound

MW

Tc (K)

Pc (MPa)

ω

Carbon dioxide Thiophene Nonane Decane

44.010 84.142 128.258 142.285

304.12 580.00 594.60 617.70

7.374 5.660 2.290 2.110

0.225 0.193 0.445 0.490

and b parameters are given by  xi xj aij am = i

bm =



(5)

j

xi bi

(6)

i

and aij is defined as √ aij = (1 − kij ) aii ajj

(i = j)

(7)

where kij is the binary interaction parameter which is related to molecular interactions between each pair of components i and j involved into the mixture. For the prediction of the VLE of ternary mixtures, the kij values obtained by correlating the VLE of binary systems CO2 + thiophene, CO2 + nonane, and CO2 + decane reported in literature [1,4,5] were used. Since there is no phase equilibrium data reported in literature for the binary mixtures nonane + thiophene, and decane + thiophene, the kij values for these mixtures were set equal to zero. Bubble point calculations were used to evaluate the vapor phase compositions and pressure. The Marquardt-Levenberg method was employed to optimize the kij values with the following objective function     cal exp 2 exp 2 Nd Nc   yijcal − yij Pj − P j   (8) + F= exp exp yij Pj j=1 i=1 where Nd is the number of data points, and Nc is the number compounds in the mixture. No temperature dependency was used for the interaction parameters. Afterward, the optimized kij parameters reported in Table 4 were used to predict solubility behavior of the ternary mixtures. The absolute average deviation (%AAD) between experimental and calculated values for the system thiophene (1) + nonane (2) + CO2 (3) were found within 7.03 and 3.42% for y1 and P, respectively. For the system thiophene (1) + decane (2) + CO2 (3), the %AAD was 6.08 and Table 4 kij values calculated from binary systems kij

Mixtures Carbon dioxide Carbon dioxide Carbon dioxide Nonane Decane

Thiophene Nonane Decane Thiophene Thiophene

0.0751 0.1016 0.0962 0.0 0.0

Octavio Elizalde-Solis, Luis A. Galicia-Luna / Fluid Phase Equilibria 236 (2005) 178–183

182

Fig. 5. K1 values for the system thiophene + nonane + CO2 at (䊉) 334.31 K, (
) 363.72 K, () 383.32 K, and (–) P-R EoS.

5.29% for y1 and P, respectively. They were calculated as follows: Nd exp exp |(B − Bical )/Bi | %AAD = i=1 i × 100 (9) Nd where B = P or y. Calculated isothermal solubilities from ternary systems were in agreement with the experimental data. In Figs. 5 and 6 similar trends are shown between the calculated distribution coefficient (K1 ) for thiophene and the experimental values. 3.3. Separation factor for thiophene The distribution coefficient for a certain component can be calculated from the relation of vapor and liquid mole fractions, for the case of thiophene (1), it is written as K=

y1 . x1

(10)

Fig. 6. K1 values for the system thiophene + decane + CO2 at (䊉) 334.40 K, (
) 363.78 K, () 383.27 K, and (–) P-R EoS.

Fig. 7. System thiophene + nonane + CO2 : separation factor between thiophene (1) and nonane (2) in CO2 at (䊉) 334.31 K, (
) 363.72 K, and () 383.32 K.

The separation factor is an important parameter for the development of a supercritical fluid process. In the ternary systems, it is calculated from the relation between the distribution coefficient of the key solute (K1 ) and the second solute. For the separation of thiophene (1) from nonane or decane (2) using CO2 as solvent, it can be written as follows: S12 =

K1 y1 /x1 = . K2 y2 /x2

(11)

Experimental VLE was used to calculate the separation factor between thiophene and the alkane with Eqs. (10) and (11). These results are illustrated in Figs. 7 and 8. At constant temperature, a decrease on the separation factor value was observed with increasing pressure. It occurred due to an increase on the solvating capacity of supercritical CO2 (solvent higher density effect). Meanwhile, with a low-pressure gradient (from 7 to 10 MPa), high changes in the separation values were observed at ∼333 K. This change can

Fig. 8. System thiophene + decane + CO2 : separation factor between thiophene (1) and decane (2) in CO2 at (䊉) 334.40 K, (
) 363.78 K, and () 383.27 K.

Octavio Elizalde-Solis, Luis A. Galicia-Luna / Fluid Phase Equilibria 236 (2005) 178–183

gas constant, J mol−1 K−1 separation factor between i and j temperature, K molar volume, cm3 mol−1 liquid mole fraction vapor mole fraction

be also obtained at 363 or 383 K, however it is needed to apply high-pressure gradients (from 3 to 15 MPa). At constant pressure above 9.0 MPa, the separation was enhanced when temperature was increased. The opposite trend occurred below 9.0 MPa, where the separation was not enhanced with increasing temperature. The separation values above 5 at supercritical solvent conditions indicate the feasibility of a selective separation of thiophene with CO2 from the studied systems.

R Sij T v x y

4. Conclusions

Superscripts exp experimental value cal calculated value

Experimental VLE data were obtained for two different ternary systems. These data were compared to the data for thiophene + CO2 system reported in literature [1], a decrease on the solubility of thiophene in CO2 was observed when nonane or decane was added to the systems. In the system where nonane was added, the solubility of thiophene was higher than that for nonane up to the vicinity of the critical point of the solvent; and above this range the solubility of nonane was higher than the obtained for thiophene. In the system where decane was included, thiophene was more soluble in CO2 than decane in the whole range. The solubility data and the distribution coefficients predicted with the Peng-Robinson EoS for the ternary systems were in agreement with the experimental data. Although the solubility of thiophene decreased in the presence of these alkanes using CO2 as supercritical solvent, experimental separation factors confirm the possibility to separate thiophene from nonane and decane, and the optimal conditions to achieve this separation could be at 2 MPa and 334 K. However, additional studies will be necessary to validate these results. List of symbols a energy parameter b covolume parameter B function for pressure or vapor mole fraction in Eq. (9) Ki distribution coefficient of i kij binary interaction parameter between i and j MW molecular weight, g/mol P pressure, MPa

183

Greek letters α temperature dependence ω acentric factor

Subscripts c critical property i, j species of the mixture m mixture property r reduced property Acknowledgment The authors are grateful to CONACYT and IPN for their financial support. References [1] O. Elizalde-Solis, L.A. Galicia-Luna, Fluid Phase Equilib. 230 (2005) 51–57. [2] S. Mitra, J.W. Chen, D.S. Viswanath, J. Chem. Eng. Data 33 (1988) 35–37. [3] D.Y. Peng, D.B. Robinson, Ind. Eng. Chem. Fundam. 15 (1976) 59–64. [4] D.W. Jennings, R.C. Schucker, J. Chem. Eng. Data 41 (1996) 831–838. [5] N. Nagarajan, R.L. Robinson, J. Chem. Eng. Data 31 (1986) 168–171. [6] P. Guilbot, A. Valtz, H. Legendre, D. Richon, Analusis 28 (2000) 426–431. [7] O. Elizalde-Solis, L.A. Galicia-Luna, S.I. Sandler, J.G. SampayoHern´andez, Fluid Phase Equilib. 210 (2003) 215–227. [8] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases and Liquids, fifth ed., McGraw-Hill, New York, 2001, pp. 6–16. [9] L.C. Yaws, Chemical Properties Handbook, McGraw-Hill, New York, 1999, p. 10.