(water + propionic acid + methyl isoamyl ketone or

Fluid Phase Equilibria 250 (2006) 70–75

(Liquid + liquid) equilibria of (water + propionic acid + methyl isoamyl ketone or diisobutyl ketone or ethyl isoamyl keton) at T = 298.2 K ∗ ¨ Dilek Ozmen Istanbul University, Engineering Faculty, Department of Chemical Engineering, 34320 Istanbul, Turkey Received 21 February 2006; received in revised form 4 October 2006; accepted 7 October 2006 Available online 14 October 2006

Abstract (Liquid + liquid) equilibrium (LLE) data for (water + propionic acid + solvent) were measured at T = 298.2 K and atmospheric pressure. The solvents were methyl isoamyl ketone (5-methyl-2-hexanone), ethyl isoamyl ketone (5-methyl-3-heptanone) and diisobutyl ketone. The tie-line data were correlated by means of the NRTL and UNIQUAC equation, and compared with results predicted by the UNIFAC method. A comparison of the extracting capabilities of the solvents was made with respect to distribution coefficients, separation factors, and solvent free selectivity bases. © 2006 Elsevier B.V. All rights reserved. Keywords: LLE; Propionic acid; Methyl isoamyl ketone; Diisobutyl ketone; Ethyl isoamyl ketone

1. Introduction

2. Experimental

This study is part of an ongoing investigation to find an efficient solvent for extracting propionic acid from dilute aqueous solutions. Liquid extraction is widely used in industry as a cheaper alternative or precursor to distillation. Propionic acid has many applications in chemical and biochemical industries. It is used as a celluloic solvent in the pharmaceutical industry, and can also be used to provide propionates. Recovery of the organic acid from aqueous solutions that are resulted from fermentation processes is also of considerable economic importance. Many solvents have been used to enhance the extraction of propionic acid from aqueous solutions [1–11]. In this work, methyl isoamyl ketone (MIAK), diisobutyl ketone (DIBK), and ethyl isoamyl ketone were used as a solvent in the separation of propionic acid from water. LLE data have been determined for each solvent at 298.2 K. Separation factors (S) for solvents separation efficiency were determined from the tie-line data and the estimated plait points were indicated by Treybal et al. [12]. The experimental data have been correlated by means of the NRTL [13] and UNIQUAC [14] equations. A comparison between the values predicted by the UNIFAC [15] method is also presented.

2.1. Chemicals

∗

Fax: +90 212 473 7180. E-mail address: [email protected].

0378-3812/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.fluid.2006.10.004

All chemicals used in this work (mass fraction purity > 0.99) were supplied by Merck and were used without further purification. The purity of these materials was checked by gas chromatography. Deionized water was further distilled before use. The densities and refractive indices of pure components were measured and compared with the literature [16] at 293.15 K and atmospheric pressure. The densities were measured using a temperature controlled Anton Paar DMA 4500 density meter in an accuracy of ±1 × 10−4 g cm−3 and the refractive indices were measured with an Abb´e–Hilger refractometer with an accuracy of ±5 × 10−4 . The estimated temperature uncertainties in the density measurements were T = 0.01 K. The measured physical properties together with literature data are presented in Table 1. 2.2. Apparatus and procedure The LLE apparatus and experimental procedure were previously described in detailed [10,11]. Tie-line data determination was performed in an equilibrium cell equipped with a magnetic stirrer and isothermal fluid jacketed beaker to keep the temperature of the stock solution constant. The mixture temperature was regulated by a thermostatic certified Fischer thermometer with an accuracy of ±0.2 K. The cell, designed to contain

¨ D. Ozmen / Fluid Phase Equilibria 250 (2006) 70–75

71

Table 1 Densities (ρ) and refractive indexes (nD ) at T = 293.15 K and atmospheric pressure of the pure components [16] Component

Water Propionic Acid MIAK DIBK Ethyl isoamyl ketonea a

ρ (g cm−3 )

Formula

H2 O CH3 CH2 COOH CH3 COCH2 CH2 CH(CH3 )2 (CH3 )2 CHCH2 COCH2 CH(CH3 )2 CH3 CH2 CH (CH3 )CH2 COCH2 CH3

nD

Experimental

Literature

Experimental

Literature

0.9984 0.9927 0.8875 0.8048 0.8227

0.99823 0.99300 0.88800 0.80530 0.82300

1.3326 1.3811 1.4074 1.4155 1.419415

1.33299 1.38090 1.40620 1.41200 1.419515

Ref. [19].

a solution from 50 to 200 cm3 , was filled with heterogeneous (water + propionic acid + solvent) mixtures prepared by mass. The tie-line data were prepared ternary mixtures of known overall compositions lying within the two-phase region, and after being allowed to reach equilibrium, samples were carefully taken from each phase and analysed. The liquid samples were analysed using a gas chromatography (HP 6890), equipped with flame ionization (FI) and thermal conductivity (TC) detectors. Ethanol was used as an internal standard. A 15 m long HP Plot Q column (0.32 mm i.d., 0.2 ␮m film thickness) for TCD, and HP-Innowax polyethylene glycol capillary column (30 m long, 0.32 mm i.d., 0.5 ␮m film thickness) for FID were utilized to separate organic components of samples at tailorized oven programs. The detector temperature was kept T = 473.15 K, while injection port temperature was held at T = 423.15 K. Injections were performed on the split 1/100 mode. Nitrogen was used as a carrier at a rate of 0.1 cm3 min−1 . The composition of water was determined by means of thermal conductivity detector (TCD).

3. Results and discussion The experimental tie-line data of (water + propionic acid + MIAK), (water + propionic acid + DIBK), (water + propionic acid + ethyl isoamyl ketone) at T = 298.2 K were reported in Table 2 with the estimated plait points [12], in which xiI and xiII refer to mole fraction of the ith component in the aqueous and solvent phase, respectively. From the LLE phase diagrams (water + solvent) mixture is the only pair that is partially miscible and two liquid pairs (propionic acid + water) and (propionic acid + solvent) are completely miscible. The correlation of the experimental data was made with NRTL [13] and UNIQUAC [14] equations. The value of the nonrandomness parameter of the NRTL equation, α, was previously assigned as 0.2. The structural parameters for UNIQUAC, van der Waals volume r and surface area q, were taken from the UNIFAC group contribution of Poling et al. [17], listed in Table 3.

Table 2 Experimental tie-lines, distribution coefficients (Di ) and separation factors (S) of the water (1) + propionic acid (2) + solvent (3) system at T = 298.2 K Water-rich phase x1I

Solvent-rich phase x2I

x3I

x1II

x2II

x3II

Water (1) + propionic acid (2) + MIAK (3) 0.9972 0.0000 0.0028 0.9808 0.0179 0.0013 0.9612 0.0384 0.0004 0.9423 0.0575 0.0002 0.9199 0.0792 0.0009 0.906 8 0.0928 0.0004

0.0236 0.1395 0.2795 0.4287 0.5479 0.6407

0.0000 0.2024 0.2844 0.3051 0.2910 0.2665

0.9764 0.6581 0.4361 0.2662 0.1611 0.0928

Estimated plait point:

0.7821

0.2020

0.0159

Water (1) + propionic acid (2) + DIBK (3) 0.9997 0.0000 0.0003 0.9722 0.0257 0.0021 0.9499 0.0485 0.0016 0.9215 0.0778 0.0007 0.8790 0.1209 0.0001 0.8238 0.1739 0.0023

0.0063 0.1516 0.1756 0.2652 0.3962 0.5417

0.0000 0.1765 0.3308 0.4082 0.3967 0.3464

0.9937 0.6719 0.4936 0.3266 0.2071 0.1119


0.6500

0.3080

0.0420

Water (1) + propionic acid (2) + ethyl isoamyl ketone (3) 0.9994 0.0000 0.0006 0.0432 0.9768 0.0230 0.0001 0.0754 0.9583 0.0416 0.0002 0.1666 0.9337 0.0654 0.0009 0.3087 0.8953 0.1009 0.0038 0.4564

0.0000 0.1779 0.3000 0.3594 0.3519

0.9568 0.7467 0.5334 0.3319 0.1918


0.7352

0.0357

0.2291

D1 (x1II /x1I )

D2 (x2II /x2I )

S (D2 /D1 )

– 0.1422 0.2908 0.4550 0.5956 0.7066

– 11.3073 7.4063 5.3061 3.6742 2.8718

– 79.4994 25.4701 11.6630 6.1689 4.0645

– 0.1559 0.1849 0.2878 0.4507 0.6576

– 6.8677 6.8206 5.2468 3.2812 1.9919

– 44.0421 36.8958 18.2312 7.2796 3.0293

– 0.0772 0.1739 0.3306 0.5099

– 7.7348 7.2115 5.4954 3.4866

– 100.2136 41.4772 16.6215 6.8381


72 Table 3 The UNIQUAC structural parameters [17] Component

r

q

Water Propionic acid MIAK DIBK Ethyl isoamyl ketone

0.9200 2.8768 5.2703 6.6183 5.9447

1.4000 2.6120 4.4920 5.5680 5.0320

The effective binary interaction parameter aij , is defined by aij = (uij − uji )/R, where uij is the UNIQUAC interaction parameter between molecules i and j and R is the gas constant. The effective binary interaction parameters for both NRTL and UNIQUAC equations aij and aji of binary systems water (1) + propionic acid (2) and water (1) + solvent (3) can be numerically solved by using experimental binary compositions as input data according to the iso-activity criterion: xiI γiI = xiII γiII ,

i=1−3

(1)

where γ i is the activity coefficient of component i. For a ternary system, there are six group-interaction parameters, aij . In this work, these effective binary interaction parameters, a12 , a13 , a21 , a23 , a31 , and a32 were determined by a numerical method. First, we defined an objective function [18]: Fa =

2 N 3 (xijI γijI − xijII γijII ) j=1 i=1

(2)

2

(xijI γijI + xijII γijII )

where xijI and xijII stand for the experimental mole fraction of component i of water-rich and solvent-rich phase, respectively, along tie-line j, γijI and γijII are the corresponding activity coefficient calculated from the UNIQUAC model and N is the total number of tie-lines. Then the effective binary interaction parameters of the UNIQUAC model were determined numerically by minimizing the objective function Fa . Next, we used the effective binary interaction parameters obtained above as the initial guesses of the following objective function [18]: Fb =

k

j

2

expl

calc (xijk − xijk ) ,

(3)

i

expl

calc are the experimental and calculated comwhere xijk and xijk position of component i in phase j along a tie-line k, respectively. The experimental data were compared with predicted value by UNIFAC method [15]. The interaction and structural parameters were taken from Poling et al. [17]. The root-mean-square deviation (RMSD) was a measure of the agreement between the experimental data and the calculated values. The RMSD value was defined as follows (Figs. 1–3):

⎡ ⎛ ⎞⎤1/2 2 expl N 2 3 calc ) (xijk − xijk ⎝ ⎠⎦ , RMSD = ⎣ 6N k

j

i

(4)

Fig. 1. Solubility curve and tie-lines of (water + propionic acid + MIAK) at T = 298.2 K: (♦) experimental tie-line data, ( ) UNIFAC, () UNIQUAC, () NRTL tie-line data, () the estimated plait points.


73

where N is the number of tie-lines, k = 1, 2, 3, . . ., N (tie-lines). The RMSD values and the NRTL and UNIQUAC parameters obtained for the studied systems were listed in Table 4. In the phase diagrams (Figs. 1–3) the experimental and calculated mole

Fig. 2. Solubility curve and tie-lines of (water + propionic acid + DIBK) at T = 298.2 K: (♦) experimental tie-line data, ( ) UNIFAC, () UNIQUAC, () NRTL tie-line data, () the estimated plait points.

Fig. 3. Solubility curve and tie-lines of (water + propionic acid + ethyl isoamyl ketone) at T = 298.2 K: (♦) experimental tie-line data, ( ) UNIFAC, () UNIQUAC, () NRTL tie-line data, () the estimated plait points.


74

Table 4 Binary interaction parameters for the UNIQUAC and NRTL models and root mean square deviations (RMSD) of the models at T = 298.2 K Systemsa

NRTLb

RMSD UNIFAC

NRTL (α = 0.2)

UNIQUAC

i–j

System 1

0.0528

0.0583

0.1326

1–2 1–3 2–3

System 2

0.0404

0.0821

0.0684

1–2 1–3 2–3

System 3

0.0369

0.0100

0.0338

1–2 1–3 2–3

aij [K] 90.6 2122.7 1887.6 −3270 1840 1970 417.0 3035.6 −575.6

UNIQUACc aji [K]

bij [K]

bji [K]

10.1 −123.8 −1494.6

−6142.5 251.0 5934.3

677.0 123.4 −6407.5

130 180010 −4620

−536.0 110.8 761.4

1056.3 258.1 −689.1

122.7 808.0 898.7

95.4 253.1 1491.9

−174.2 941.8 −468.9

a System 1: water (1) + propionic acid (2) + MIAK (3); system 2: water (1) + propionic acid (2) + DIBK (3) system 3: water (1) + propionic acid (2) + ethyl isoamyl ketone (3). b a = gij −gjj . ij R c

bij =

uij −ujj R

.

fractions of two phases at equilibrium and the corresponding tie-lines were represented. The effectiveness of extraction of propionic acid (2) by these solvents was given by separation factors (S), which was a measure of the ability of the solvents to separate the propionic acid (2) from water (1): xI xII S = 1II 2I x1 x2

(5)

This quantity is found to be greater than 1 (separating factors varying between 3.0293 and 100.2136) for the systems reported here, which means that extraction of propionic acid by MIAK, DIBK or ethyl isoamyl ketone is feasible. The distribution coefficients (Di ) and separation factors (S) for the each system were given with experimental tie-line data in Table 2. To compare the selectivity advantages of DIBK, MIAK and ethyl isoamyl ketone, we plotted solvent-free based selectivity diagrams in Fig. 4. The selectivity diagram indicated that the performance

Fig. 4. Selectivity diagram (solvent-free basis) of (water + propionic acid + solvent) at T = 298.2 K: () DIBK, ( ) MIAK, () ethyl isoamyl ketone.

of the solvents increases in the order of MIAK, ethyl isoamyl ketone, and DIBK. 4. Conclusions The experimental tie-line data of (water + propionic acid + MIAK), (water + propionic acid + DIBK), and (water + propionic acid + ethyl isoamyl ketone) were measured at T = 298.2 K and atmospheric pressure. Each ternary system exhibits type-1 behaviour of LLE. The separation factors (S) and slopes of tie-lines indicated that all solvents used in this study, may serve as adequate solvents for the extraction of propionic acid from dilute aqueous solutions. Figs. 1–3 show that all models fitted the experimental data quite-well for (water + propionic acid + ethyl isoamyl ketone) system. List of symbols aij , bij binary energy parameters for i–j pair (K) Di distribution coefficient DIBK diiso butyl ketone Fa , Fb objective functions given by Eqs. (2) and (3), respectively gij NRTL parameter (J mol−1 ) MIAK methyl isoamyl ketone N number of tie-lines nD refractive index q the relative van der Waals surface area r the relative van der Waals volume R the gas constant RMSD root mean square deviation calculated by Eq. (4) S separation factors calculated by Eq. (5) uij UNIQUAC parameter (J mol−1 ) xi mole fraction of the ith component the experimental mole fraction of component i along a xij tie-line j xijk composition of component i in phase j along a tie-line k


Greek letters ρ density (g cm−3 ) α NRTL parameter Activity coefficient of component i along a tie-line j γ ij Superscripts calc calculated values expl experimental values I aqueous phase II solvent phase Acknowledgements This work was supported by the Research Found of Istanbul University, Project Number: T-912/06112000. The author would ¨ like to thank to Assist. Prof. Atilla Ozmen. References [1] A. Badakhshan, A.I. Chowdhury, R. Leung, J. Chem. Eng. Data 30 (1985) 416–421. [2] A. Arce, A. Blanco, P. Souza, I. Vidal, J. Chem. Eng. Data 38 (1993) 201–203.

75

[3] A. Arce, A. Blanco, P. Souza, I. Vidal, J. Chem. Eng. Data 40 (1995) 225–229. [4] S. C ¸ ehreli, B. Tatlı, U. Dramur, Chim. Acta Turcica 27 (2) (1999) 53–61. [5] V. Taghikhani, G.R. Vakili-Nezhaad, M.K. Khoshkbarchi, M. ShariatyNiassar, J. Chem. Eng. Data 46 (2001) 1107–1109. [6] T.M. Letcher, G.G. Redhi, Fluid Phase Equilib. 193 (2002) 123–133. ¨ [7] D. Ozmen, U. Dramur, B. Tatlı, Braz. J. Chem. Eng. 21 (2004) 647–657. [8] Sh. Hashemi, C. Ghotbi, V. Taghikhani, B. Behzadi, Fluid Phase Equilib. 226 (2005) 251–259. [9] S. C ¸ ehreli, B. Tatlı, P. Ba˘gman, J. Chem. Thermodyn. 37 (2005) 1288–1293. ¨ [10] D. Ozmen, S. C ¸ ehreli, U. Dramur, J. Chem. Thermodyn. 37 (2005) 837–842. ¨ [11] S. C ¸ ehreli, D. Ozmen, B. Tatlı, J. Chem. Thermodyn. 37 (2005) 1144– 1150. [12] R.E. Treybal, L.D. Weber, J.F. Daley, Ind. Eng. Chem. 38 (1946) 817–821. [13] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135–144. [14] D.S. Abrams, J.M. Prausnitz, AIChE J. 21 (1975) 116–128. [15] A. Fredenslund, R.L. Jones, J.M. Prausnitz, AIChE J. 21 (1975) 1086–1099. [16] R.C. Weast, Handbook of Chemistry and Physics, seventieth ed., CRC Press, Boca Raton, FL, 1989–1990. [17] B.E. Poling, J.M. Prausnitz, J.P. O’Connel, The Properties of Gases and Liquids, fifth ed., McGraw-Hill Companies Inc., USA, 2001. [18] J.M. Sorensen, Correlation of liquid–liquid equilibrium data, Ph.D. Thesis, Technical University of Denmark, Lyngby, Denmark, 1980. [19] S. Budavari (Ed.), The Merck Index: An Encyclopedia of Chemicals, Drugs, and Biologicals, eleventh ed., Merck & Co. Inc., Rahway, NJ, USA, 1989.