Concentration by pervaporation of representative brown crab volatile compounds from dilute model solutions

2 Concentration by pervaporation of representative 1 brown crab volatile compounds from dilute 2 model solution. 3 Rodrigo Martínez, María Teresa Sanz ∗ , Sagrario Beltrán 4 Department of Chemical Engineering, University of Burgos, 09001 Burgos. Spain 5 In this work, the pervaporation technique is investigated in the separation of dilute 6 solutions of volatile compounds from brown crab effluent in order to obtain a valuable food 7 flavouring fraction. A systematic study of the pervaporation process has been carried out on 8 dilute model solutions of some of the compounds identified in the brown crab effluent as 9 typical volatile compounds. The membrane used in this work was a hydrophobic membrane 10 with a selective layer of POMS (polyoctylmethyl siloxane). The effect of some operating 11 variables, such as feed flow rate, feed concentration, feed temperature and permeate 12 pressure was analyzed on the pervaporation performance of the membrane. 13 Keywords: volatile compounds, concentration, pervaporation, POMS membrane 14 ∗ Corresponding author. Tel.: +34 947 258810. Fax: +34 947 258831. E-mail address [email protected] 3 1. Introduction 15 Brown crabs are found in the Eastern Atlantic and are heavily exploited commercially 16 being available throughout the year. The brown crab liquid effluent produced during 17 boiling is believed to contain important amounts of volatile flavour components (Cha et al. 18 1993). This work is part of a wider study to consider the conversion of this by-product into 19 valuable volatile concentrate. Concentrates of the volatile species have considerable 20 commercial utility, especially in the food industry due to longer shelf life, reduced 21 packaging and lower distribution and storage costs (She and Hwang, 2006). Additionally, 22 organic removal from water at low concentrations involves an important environmental 23 challenge. 24 Cha et al. (1993) studied the concentration of the liquid effluent produced during snow crab 25 boiling by steam distillation. However, this technique involves high energy consumption as 26 well as physical aroma losses (García et al. 2008). In this work, the pervaporation process 27 has been considered to recover the volatile fraction from brown crab effluent. 28 Pervaporation is a membrane process which has been developed rapidly in the last 20 years 29 for aroma concentration (She and Hwang, 2006) since the addition of chemical solvents is 30 avoided. Additionally moderate operating temperatures help to minimize degradation of 31 aroma character. 32 Brown crab effluent was supplied by IDOKI SCF Technologies S.L. (Spain). The first step 33 in this work was to determine the main volatile organic compounds present in the industrial 34 effluent. More than 150 compounds were identified in the brown crab effluent. These 35 included mainly aldehydes, ketones, alcohols, esters, aromatic compounds and sulphur and 36 nitrogen-containing compounds. To study the ability of pervaporation process to recover 37 4 the volatile fraction from the brown crab effluent seven of the identified compounds have 38 been selected for a model aqueous solution of brown crab effluent: 1-octen-3-ol, 1-penten-39 3-ol, 3-methylbutanal, hexanal, benzaldehyde, 2,3-pentadione and ethyl acetate. 40 A systematic study of the pervaporation process of the dilute model solution was performed 41 in order to analyze the influence of some operating variables on the pervaporation 42 performance. The permeation flux and enrichment factor of the selected volatile 43 compounds were analyzed at different operating conditions: feed flow rate, feed 44 temperature, feed concentration and permeate pressure. 45 2. Theory. 46 On the basis on the solution/diffusion model the flux of component i through the membrane 47 is proportional to the difference in partial vapor pressure at both sides of the membrane 48 (Blume et al. 1990): 49 ( ) pi s iii i ,OVipyp xQJ −γ= (1) 50 where Ji is the partial permeation flux, QOVi the pressure-normalized permeation flux 51 (permeance), xi the mole fraction of component i in the feed, γ i the activity coefficient and 52 s i p the saturation vapor pressure at the temperature of the feed, yi, the mole fraction in the 53 permeate and pp the permeate pressure. In case of pervaporation of dilute aqueous solutions 54 activity coefficients at infinite dilution in water ( ∞ γi ) are used as feed-side activity 55 coefficients due to the very low concentrations of aroma compounds in the feed (Trifunovic 56 and Trägardh, 2006). In this work, the activity coefficients at infinite dilution in water were 57 estimated with the help of the software Aspen plus (2008) by using UNIQUAC equation 58 5 when binary interaction parameters were available, otherwise the predictive method 59 UNIFAC-Dortmund was used. 60 According to the resistance-in-series model, the two main mass transfer resistances that 61 affect the pervaporation process are the liquid boundary layer resistance and the membrane 62 resistance. At steady state the flux through the different mass transfer layers is equal: 63 ( ) ( ) p i m i i, m m i i i,blpi s iiii,ov i p y pQ )x x(kpypxQ J− = −ρ=−γ= (2) 64 where kbl is the liquid boundary layer mass transfer coefficient, ρ the total mass volume 65 concentration of the feed, m i x mol fraction of i at the membrane-fluid interface, Qm,i the 66 pressure normalized permeation flux across the membrane and m i p the partial vapor 67 pressure of i at the membrane-fluid interface. The rest of the symbols are the same as in 68 Equation 1. The overall mass transfer coefficient in the steady state can be expressed as the 69 sum of these two resistances: 70 i ,ml,b s i i i,ov Q 1 k p Q 1+ ρ γ = (3) 71 The term ρ γs i ip is the conversion factor from a concentration driving force to a partial 72 vapour pressure driving force. The overall mass transfer coefficient QOV,i of Eq. 1 can be 73 obtained from experimental measurements of the permeate flux and feed concentration of 74 the permeating component i. The liquid boundary layer mass transfer coefficient, kbL, is 75 related to the feed hydrodynamic conditions and it can be estimated from the Sherwood 76 correlation in terms of Reynolds (Re) and Schmidt (Sc) numbers for a plate-and-frame 77 module (Dotremont et al. 1994): 78 6 3 / 1 h 3/13/1 water ,i h bl L d ScRe86.1 D dk Sh       == (4) 79 where dh is the hydraulic diameter, L a characteristic measure of the module defined by 80 Dotremont et al. (1994) for a similar plate and frame module and Di,water the diffusion 81 coefficient of i in water estimated using the Wilke-Chang correlation (Poling et al. 2001). 82 For pervaporation of dilute organic solutions, the boundary layer mass transfer resistance 83 for water transport is assumed to be negligible (Ji et al. 1994): 84 ( ) pwww s w m ww pyxpQJ −γ= (5) 85 for dilute aqueous solutions activity coefficient and molar fraction of water are 86 approximately equal to 1. 87 The separation performance of a pervaporation membrane can be described in terms of the 88 permeation flux and the separation factor of the membrane (Huang and Rhim, 1991). The 89 enrichment factor of a given component is the relationship between the concentration in the 90 permeate and the feed: 91 f,ip,ii ww=β (6) 92 In dilute systems, as aroma recovery systems, the solvent enrichment factor is close to one, 93 so aroma enrichment factors can be considered equal to the corresponding separation 94 factors. 95 3. Experimental section 96 3.1. Materials 97 Pervaporation membrane 98 7 The membrane used in this work was a hydrophobic membrane kindly supplied by GKKS 99 Research Center (Germany). This membrane has a selective layer of POMS 100 (polyoctylmethyl siloxane) on a PEI (poly ether imide) support (batch 03/011). 101 Volatile compounds 102 The identification of the main volatile components present in the brown crab effluent was 103 performed by using a headspace-solid phase dynamic extraction-gas chromatography/mass 104 spectrometry (HS-SPDE-GC/MS). More than 150 compounds were identified in the brown 105 crab effluent. Among them, seven compounds have been selected for a model aqueous 106 solution of brown crab effluent. The selected volatile compounds belong to different 107 chemical classes: 1-octen-3-ol (Sigma Aldrich, 98 %), 1-penten-3-ol (Sigma Aldrich, 108 99 %), 3-methylbutanal (Sigma Aldrich, 97 %), hexanal (Sigma Aldrich, 98 %), 109 benzaldehyde (Sigma Aldrich, >=99 %), 2,3-pentadione (Sigma Aldrich, 98 %) and ethyl 110 acetate (Sigma Aldrich, HPLC grade). These compounds are characteristic of seafood 111 flavour: 1-octen-3-ol has been reported to be one of the volatile components widely 112 distributed in fresh and saltwater fish, 1-penten-3-ol contributes to a butter-like odor 113 (although its aroma treshold value is rather high), 3-methylbutanal is one of the most 114 abundant volatile compound in boiled and pasteurized crabmeat, hexanal is one of the most 115 abundant volatiles generated during lipid oxidation at moderate temperatures, benzaldehyde 116 contributes to characteristic cooked crab flavour and ketones such as 2,3 pentadione 117 contribute to the sweet floral, fruity flavour of many crustacean (Cha et al., 1993; 118 Josephson, 1990; Matiella and Hsieh, 1990). Ethyl acetate was also found in the brown 119 crab effluent and it was included in the model solution since could be considered as model 120 molecule (Baudot et al., 1999). 121 8 Table 1 summarizes the organoleptic characteristics of the selected volatile compounds 122 including the aroma threshold values (ATV), defined as the lowest concentration in a water 123 solution at which an aroma compound is perceptible. Table 2 lists some thermodynamic 124 properties of the selected compounds, including activity coefficients at infinite dilution and 125 vapor pressure of the volatile compounds. Vapor pressure correlations were obtained or 126 predicted by using Aspen Plus (2008) except for 2,3-pentadione which Antoine constants 127 were obtained from the literature (Soni et al., 2008). Figure 1 shows the vapor pressure of 128 the volatile compounds including water vapor pressure as a function of temperature. 129 3.2. Feed solutions 130 Different feed solutions were used in this work. First, pervaporation experiments were 131 performed using pure water as feed solution to check the performance of the POMS 132 membrane. Further, separations of binary mixtures (water/1-octen-3-ol) and 133 multicomponent mixtures were carried out in order to evaluate the influence of some 134 operating variables such as: feed flow rate, feed concentration, feed temperature and 135 permeate pressure on pervaporation performance. 136 3.3. Pervaporation experiments 137 The pervaporation experiments were performed under steady state with a plate and frame 138 laboratory stainless steel permeation cell (Sulzer Chemtech) with an effective membrane 139 area in contact with the feed mixture of 170 cm2 (Delgado et al., 2009). The temperature of 140 the feed liquid mixture was kept constant (± 0.5 ºC) by using a thermostat to heat the stirred 141 tank feed reactor of 5 L capacity. Permeate pressure was regulated with an air-inlet located 142 between the condensers and the vacuum pump. The chemical stability of the membrane 143 9 was checked between each experiment, measuring pure water flux at reference operating 144 conditions. 145 3.4. Sample analysis 146 Permeate and feed concentrations were measured off-line using a Hewlett Packard (6890) 147 gas chromatograph (GC) equipped with series connected thermal conductivity (TCD) and 148 flame ionization (FID) detectors. Helium, 99.999 % pure, was used as carrier gas. The GC 149 column was a 007 FFAP 25 m × 0.25 mm bonded phase fused silica capillary column. The 150 injector and detectors were at 200 ºC and 250ºC respectively. The oven was operated at 151 programmed temperature, from 40ºC to 220ºC. 1-hexanol was used as internal standard for 152 analysis of the samples. 153 4. Results and discussion 154 4.1 Pure water as feed solution 155 The effect of feed temperature and permeate pressure on membrane performance was 156 studied using pure water as feed solution to check the behaviour of the POMS membrane. 157 Feed temperature was varied in the range 26 ºC to 45 ºC. By increasing feed temperature, 158 water permeation flux also increases mainly due to the increase of saturated water pressure 159 on the feed side of the membrane (Eq. 5). The temperature dependence of water permeation 160 flux, Jwater, can be expressed by an Arrhenius-type relation: 161 ( ) RTEexpJ J water, ao ,waterwater − = (7) 162 where Ea,water is the apparent activation energy of permeation, Jwater,o the preexponential 163 factor and T the absolute temperature. An apparent activation energy of 46.65 kJ/mol 164 10 (Figure 2) was found by fitting water permeation flux obtained in this section (pure water 165 as feed solution) as well as water permeation flux obtained in the pervaporation of volatile 166 compounds dilute aqueous solutions (section 4.3). 167 Permeate pressure was varied in the range 100 Pa to 1200 Pa. Figure 3 shows the water 168 permeation flux dependence on permeate pressure. This Figure shows experimental data 169 obtained using pure water as feed solution and the results obtained in subsequent studies 170 (section 4.3). Figure 3 shows that by increasing the permeate pressure, water permeation 171 flux decreases as consequence of a decrease in the driving force (Equation 5). Water 172 permeances for the POMS membrane, calculated as the ratio of the permeation flux to the 173 permeant driving force, were constant whatever the feed temperature and permeate pressure 174 considered (1.95·10-7 ± 1.23·10-8 mols-1m-2Pa-1 for all the experiments performed in this 175 work). 176 4.2 Binary feed solution 177 4.2.1 Boundary layer effect 178 First, the boundary layer effect was studied in the pervaporation of the binary system 179 water/1-octen-3-ol by varying the feed flow rate between 25 kg/h to 92 kg/h. According to 180 resistance-in-series model when boundary layer is dominant resistance, mass transfer 181 across the membrane increases with feed flow rate due to a decrease of the boundary layer 182 thickness. Figure 4 shows the effect of increasing feed flow rate on partial (water and 1-183 octen-3-ol) permeation flux. Water and organic permeation fluxes were approximately 184 constant inferring that little concentration polarization takes place. The mass transfer 185 coefficient kbl was calculated according to Sherwood correlation (Eq. 4). The relative 186 significance of the boundary layer mass transfer resistance was estimated less than 2% of 187 17 D = diffusion coefficient, m2·s-1 325 Ea = apparent activation energy of permeation, kJ·mol-1 326 J = mass permeation flux, g·s-1·m-2 327 kbl = liquid boundary layer mass transfer coefficient, m·s-1 328 L = characteristics of the module, m 329 p = pressure, Pa 330 QOV = pressure-normalized permeation flux, g·s-1·m-2·Pa-1 331 Qm = pressure-normalized permeation flux across the membrane, g·s-1·m-2·Pa-1 332 R = gas constant, kJ·mol-1K-1 333 T = absolute temperature, K 334 x, y = mol fraction 335 ρ = total mass volume concentration of the feed, g·m-3 336 γ = activity coefficient 337 β = enrichment factor 338 upperscripts 339 i = component 340 m = membrane 341 p = permeate 342 s = saturation 343 18 Acknowledgments 344 Financial support from the MICINN through CTQ2008-04999-PPQ is gratefully 345 acknowledged. R. Martinez acknowledges the JCyL Education Ministry. 346 References 347 Aroujalian, A., & Raisi A. (2007). Recovery of volatile aroma components from orange 348 juice by pervaporation. J. Membr. Sci., 303, 154-161. 349 Aspen Plus V7.1 (2008) Aspen Technology, Inc., 2008. 350 Baudot, A., Souchon, I., & Marin, M. (1999). Total permeate pressure influence on the 351 selectivity of the pervaporation of aroma compounds. J. Membr. Sci., 158, 167-185. 352 Blume, I., Wijmans, J.G., & Baker, R. W. (1990). The separation of dissolved organics 353 from water by pervaporation. J. Membr. Sci., 49, 253-286. 354 Cha, Y.J., Cadwallader, K.R., & Baek, H.H. (1993). Volatile flavor components in snow 355 crab cooker effluent and effluent concentrate. J. Food Sci., 58, 525-530. 356 Delgado, P., Sanz, M.T., & Beltrán, S. (2009). Pervaporation of the quaternary mixture 357 present during the esterification of lactic acid with ethanol. J. Membr. Sci., 332, 113-120. 358 Dotremont, C., Van den Ende, S., Vandommele, H., & Vandecasteele, C. (1994). 359 Concentration polarization and other boundary layer effects in the pervaporation of 360 chlorinated hydrocarbons. Desalination, 95, 91-113. 361 Feng, X.; & Huang, R.Y.M. (1996). Estimation of activation energy for permeation in 362 pervaporation processes. J. Membr. Sci., 118, 127-131. 363 Flavor-Base Professional (2007), Lefingwell & Associates. 364 19 García, V., Diban, N., Gorri, D., Keiski, R., Urtiaga, A., &. Ortiz, I. (2008). Separation and 365 concentration of bilberry impact aroma compound from dilute model solution by 366 pervaporation. J. Chem. Technol. Biotechnol., 83, 973-982. 367 Huang, R.Y.M, & Rhim, J.W. (1991) Separation characteristics of pervaporation membrane 368 separation processes. In Elsevier (Ed.), Pervaporation Membrane Separation Processes 369 (pp. 111-180). Amsterdam. 370 Isci, A., Sahin, S., & Sumnu, G. (2006). Recovery of strawberry aroma compounds by 371 pervaporation. J. Food Eng, 75, 36-42. 372 Ji, W., Hilalyb, A., Sikdarb, S.K., & Hwang, S.T. (1994). Optimization of multicomponent 373 pervaporation for removal of volatile organic compounds from water. J. Membr. Sci., 97, 374 109-125. 375 Josephson, D.B. (1991). Seafood. In Marcel Dekker (Ed.), Volatile compounds in Foods 376 and Beverages (pp. 179-202). New York. 377 Kanani, D.M., Nikhade, B.P., Balakrishnan, P., Singh, G., & Pangarkar, W.G. (2003). 378 Recovery of valuable tea aroma components by pervaporation. Ind. Eng. Chem. Res., 42, 379 6924-6932. 380 Matiella, J., & Hsieh, T.C. (1990). Analysis of crabmeat volatile compounds. J. Food Sci., 381 55, 962-966. 382 Olsson, J., & Tragardh, G. (1999a). Influence of feed flow velocity on pervaporative aroma 383 recovery from a model solution of apple juice aroma compounds. J. Food Eng., 39, 107-384 115. 385 Olsson, J., & Tragardh, G. (1999b). Influence of temperature on membrane permeabilibity 386 during pervaporative aroma recovery. Sep. Sci. Technol., 34, 1643-1659. 387 20 Poling, B.E., Prausnitz, J.M., & O’Connell, J.P. (2001). Difussion Coefficients. In 388 McGraw-Hill (Ed.), The Properties of Gases and Liquids (pp. 11.1-11.55). New York. 389 Raisi, A., Aroujalina, A., & Kaghazchi, T. (2008). Multicomponent pervaporation process 390 for volatile aroma compounds recovery form pomegranate juice. J. Membr. Sci., 322, 339-391 348. 392 Sampranpiboon, P., Jiraratananon, R., Uttapap, D., Feng, X., & Huang, R.Y.M. (2000). 393 Separation of aroma compounds from aqueous solution by pervaporation using 394 polyoctylmethylsiloxane (POMS) and polydimethylsiloxane (PDMS) membranes. J. 395 Membr. Sci., 174, 55-65. 396 She, M., & Hwang, S.-T. (2006). Recovery of key components from real flavour 397 concentrates by pervaporation. J. Membr. Sci., 279, 86-93. 398 Soni, M., Ramjugernath, D., & Raal, J.D. (2008). Vapor-Liquid Equilibrium for binary 399 systems of 2,3-pentanedione with diacetyl and acetone. J. Chem. Eng. Data, 53, 745-749. 400 Trifunovic, O., & Trägardh, G. (2006). Mass transport of aliphatic alcohols and esters 401 through hydrophobic pervaporation membranes. Sep. Purif. Technol., 50, 51-61. 402 Trifunovic, O., Lipnizki, F., & Tragardh, G. (2006). The influence of process parameters on 403 aroma recovery by hydrophobic pervaporation. Desalination, 189, 1-12. 404 Wijmans, J.G., Athayde, A.L., Danield, R., Ly, J.H., Kamaruddin, H.D., & Pinnau, I. 405 (1996). The role of boundary layers in the removal of volatile orgnacis compounds from 406 water by pervaporation. J. Membr. Sci., 109, 135-146. 407 408 21 Table 1. Aroma compounds used in the model solution (Flavor-Base Professional, 2007). 409 Aroma compound Organoleptic characteristics ATV, ppb 1-Octen-3-ol Very strong, sweet, earthy mushroom odor and taste 1.0 1-Penten-3-ol Pungent, grassy, alliaceous-like, green vegetable, fruity taste 400 3-Methylbutanal Powerful, penetrating, cheesy-sweaty-fruity in dilution 0.2-2 Hexanal Strong, penetrating, fatty-green, grassy unripe fruit odor 4.5 Benzaldehyde Odor of bitter almond oil, characteristic sweet cherry taste 350 2,3-Pentadione Oily-buttery, fatty odor, butter, cream, milk taste 30 Ethyl Acetate Ethereal, sharp, wine-brandy like odor 5.0 410 411 22 412 Table 2. Physicochemical properties of the volatile compounds 413 Compound 1-Octen-3-ol 1-Penten-3-ol 3-Methylbutanal Hexanal Benzaldehyde 2,3-Pentadione Ethyl acetate MW 128.2 86.1 86.1 100.2 106.1 100.1 88.1 BP (ºC) 174-5 114.4 92.5 128.3 178.8 108 77.1 p s (26ºC), Pa 65 1348 7035 1540 181 2918 13045 γ ∞ (26ºC) 4955.4 17.0 164.6 1047.2 273.7 282.4 75.3 VM BP (cc/mol) 186.6 117 118.9 140.0 118.6 121 106.3 414 23 List of Figure Caption 415 Figure 1. Vapor pressure of water and volatile compounds as a function of temperature. 416 Figure 2. Effect of temperature on water permeation flux (pp = 300 Pa). 417 Figure 3.Effect of permeate pressure on water permeation flux (T = 26ºC). 418 Figure 4. Water (▲) and 1-octen-3-ol (i, Δ) permeation flux at different Reynolds (pp = 419 400 Pa, T = 26ºC, Ci,feed ≈ 5 ppm). 420 Figure 5. Effect of volatile feed concentration on volatile compound permeation flux (T = 421 26 ºC, pp = 300 Pa). 422 Figure 6. Volatile compound permeance as a function of volatile feed concentration (T = 423 26 ºC, pp = 300 Pa). 424 Figure 7. Effect of feed temperature on volatile compound permeation flux (Ci, feed ≈ 425 10 ppm, pp = 300 Pa). 426 Figure 8. Enrichment factor of volatile compound at different operating temperatures 427 (Ci,feed ≈ 10 ppm, pp = 300 Pa). 428 Figure 9. Ratio of volatile compound permeation flux at 35.7 ºC and 26ºC. 429 Figure 10. Enrichment factor of volatile compound at different operating permeate 430 pressure (Ci,feed ≈ 10 ppm, T = 26 ºC). 431 432 433 24 434 435 436 437 438 439 440 0 5 10 15 20 25 30 35 40 0 10 20 30 40 50 60 T(ºC) p s ·10 3 (Pa) 441 442 Figure 1. Vapor pressure of water and volatile compounds as a function of temperature. 443 444 445 446 (1) (2) water (3) (4) (5) (6) (7) (1) ethyl acetate (2) 3-methylbutanal (3) 2,3-pentadione (4) 1-penten-3-ol (5) hexanal (6) benzaldehyde (7) 1-octen-3-ol 25 447 448 449 450 451 452 -5.0 -4.6 -4.2 -3.8 -3.4 -3.0 3.05 3.15 3.25 3.35 3.45 1000/T(K) ln Jwater (g/sm2) Pure water Multicomponent mixture 453 Figure 2. Effect of temperature on water permeation flux (pp = 300 Pa). 454 455 456 26 457 458 459 460 461 462 0.000 0.005 0.010 0.015 0.020 0 500 1000 1500 2000 p, Pa Jwater (g/sm2) Multicomponent mixture Pure water 463 Figure 3. Effect of permeate pressure on water permeation flux (T = 26ºC). 464 465 466 33 532 533 534 535 536 537 0 25 50 75 100 125 150 1-octen-3-ol 1-penten-3-ol 3-methylbutanal hexanal benzaldehyde 2,3-pentanodione ethyl acetate Enrichment factor 100 Pa 300 Pa 400 Pa 600 Pa 900 Pa 1800 Pa 538 Figure 10. Enrichment factor of volatile compound at different operating permeate 539 pressure (Ci,feed ≈ 10 ppm, T = 26 ºC). 540 541 542