J. of Supercritical Fluids 40 (2007) 344–348
Supercritical carbon dioxide extraction of acorn oil M. Gabriela Bernardo-Gil a,∗ , Isabel M.G. Lopes a , Miguel Casquilho b , M. Albertina Ribeiro a , M. Mercedes Esqu´ıvel a , Jos´e Empis a a
Centre for Biological and Chemical Engineering, DEQB, IST, Instituto Superior T´ecnico, Av. Rovisco Pais, 1049-001 Lisboa, Portugal b Centre of Chemical Processes, DEQB, IST, Instituto Superior T´ ecnico, Av. Rovisco Pais, 1049-001 Lisboa, Portugal Received 2 January 2006; received in revised form 9 June 2006; accepted 6 July 2006
Abstract Acorn fruit oil of Quercus rotundifolia L. (holm-oak) was extracted with compressed carbon dioxide in the temperature range of 35–60 ◦ C and in the pressure range of 12–21 MPa. The influences of particle size, CO2 density, solvent flow rate, and extractor geometry were studied. Two different tubular extractors were used: extractor 1 of 0.2 L of capacity, internal diameter D = 45.7 mm, height/diameter ratio H/D = 1.5, and extractor 2 of 0.09 L of capacity, internal diameter D = 21.3 mm, and H /D = 12. It was found that the yield and the initial extraction rate depend on the carbon dioxide density and superficial velocity, and on the ratio of bed height to D, at the same conditions of temperature, pressure, and particle size. In the beginning of extraction, CO2 density is the preponderant factor, but after some time of extraction, the fraction of oil directly exposed to the solvent, which is dependent on particle size, becomes the most important factor. The Sovov´a model was successfully applied to the description of the supercritical extraction curves of acorn oil. © 2006 Elsevier B.V. All rights reserved. Keywords: Supercritical carbon dioxide; Quercus rotundifolia L.; Acorn; Extraction; Sovov´a model
1. Introduction Since ancient times, in Portugal, namely in the Alentejo region, fruits of Quercus rotundifolia L., in particular, and of Quercus suber L., have been used to feed swine raised under loose-housing. In the sixties, the acorn oil was valorised as raw material for the alimentary oil industry. At present, Portuguese legislation [1] places the acorn oil in the category of directly alimentary oils, although no industrial oil is being produced. The oil composition of Q. rotundifolia L. acorn has been reported as containing about 12% of crude fat. The major fatty acids found in acorn oil are oleic acid (18:1), about 65%, linoleic acid (18:2), about 17%, and palmitic acid (16:0), about 13% [2]. As linoleic acid is considered the quite significant and valuable as regards human health [3], the acorn oil can be considered as an important oil. Several vegetable and other oils have been obtained by using supercritical fluid extraction (SFE) [4–9]. Advantages of the use of supercritical carbon dioxide (SCCO2 ) include shorter extrac∗
Corresponding author. Tel.: +351 21 841 7582; fax: +351 21 841 9176. E-mail address:
[email protected] (M.G. Bernardo-Gil).
0896-8446/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.supflu.2006.07.026
tion times and the potential for selective extractions by varying the pressure and temperature. Besides, the use of SCCO2 is an innovative replacement of traditional organic solvents, and the technique has been rapidly growing in parallel with the increased more stringent legislation rules against the use of volatile organic compounds (VOC’s). The aim of this work was to study the supercritical fluid extraction (SFE) of the Q. rotundifolia acorn oil. Carbon dioxide was used in the temperature range of 35–60 ◦ C, at pressures between 12 and 21 MPa. The influence of several parameters on yield was studied. The parameters analysed were: CO2 density (depending on pressure and temperature of extraction), CO2 flow rate, residence time, feed particle size, and extractor geometry. The experimental data were used to obtain the parameters of Sovov´a model [10,11]. 2. Materials and methods 2.1. Raw material Q. rotundifolia L. acorns were manually gathered near Portalegre, Alto Alentejo region (Eastern Portugal). Samples were
M.G. Bernardo-Gil et al. / J. of Supercritical Fluids 40 (2007) 344–348
dried at 60 ◦ C during 3 days, manually barked and then crushed using an Armfield FT2 hammer mill (Armfield Technical Education Company Limited, Ringwood Hampshire, England) to achieve the desired average particle diameter (Dp = 0.68 or 0.27 mm). The milled acorn was stored in glass flasks at about 0 ◦ C, under N2 atmosphere, to avoid oxidation. Moisture content was measured by determining the loss of mass after drying overnight in an oven at 110 ◦ C. A value of 11.8 ± 0.2 wt.% was found. Total oil content was determined by using the ISO 659 standard method (Soxhlet extraction using hexane as a solvent). A value of 12.1 ± 0.4 wt.% was obtained. The density of acorn was determined as 930 ± 18 kg m−3 , and the bulk densities before extraction, were 654 ± 11 and 542 ± 10 kg m−3 for ground acorn with Dp = 0.68 and 0.27 mm, respectively. 2.2. Solvents Carbon dioxide used was CO2 N45 (purity ≥99.995%), supplied by AR LIQUIDO-Portugal. All other solvents and reagents used in analytical determinations were from Merck, pro analysis grade. 2.3. Supercritical fluid extraction apparatus and procedure Extraction measurements were carried out in a semi-batch flow extraction apparatus built at Instituto Superior T´ecnico (Lisbon). The equipment was identical to that described by Esqu´ıvel and Bernardo-Gil [4] and Oliveira et al. [8]. The extraction experiments were performed within two different tubular extractors: extractor 1 of 0.2 L of capacity, internal diameter D = 45.7 mm, height/diameter ratio H/D = 1.5, and extractor 2 of 0.09 L of capacity, internal diameter D = 21.3 mm, H /D = 12. The mass of oily extract (and hence the yield of extract) was determined as a function of extraction time and the mass of CO2 passed under each of the conditions of temperature, pressure, particle size, and superficial velocity. 2.4. Modelling The extraction curves were modelled by the Sovov´a model [10,11]. The overall mass transfer coefficients in the supercritical and the solid phase (Kf , Ks ), and the fraction of oil directly exposed to CO2 (fk ), were used as adjustable parameters. The solubility (yr ) of the acorn oil in the solvent, at work conditions of pressure and temperature, was calculated assuming that the solubility of vegetable oils in supercritical CO2 is proportionally dependent on the content in fatty acids. At phase equilibrium Chastil [12] assumed a linear relationship between the logarithm of solubility, y (kg/kg CO2 ), and the logarithm of supercritical CO2 density, ρ (kg m−3 ), which can be expressed as a ln y = (k − 1) ln ρ + +b (1) T
345
Table 1 Chrastil’s correlation parameters Fatty acid Palmitic Oleic Linoleic
16:0 18:1 18:2
k
a
b
7.53 9.02 11.6
−6632 −8232 −9472
−27.5 −32.8 −45.5
where T is the absolute temperature (K), and k, a, and b are the parameters that depend on the nature of the fatty acid. The parameter k represents the number of solvent molecules associated with one molecule of solute in the solvate complex, a is related to the heat of solvation and to the heat of vaporisation of the solute, whereas b depends on the molecular mass and melting points of solvent and solute. Chrastil’s correlation was used to estimate the solubility of the main fatty acids (oleic, linoleic and palmitic) of acorn oil [2]. The values of constants k, a, and b were obtained from data of Vasconcellos and Cabral [13], and are presented in Table 1. 3. Results and discussion Experimental results of extraction curves were fitted by using the Sovov´a model. In Table 2 are presented the parameters Kf , Ks and fk obtained. This model successfully describes the acorn oil extraction by supercritical carbon dioxide, as can be seen in Figs. 2–4. The value of fk = 0.66 for ground acorn with Dp = 0.27 mm is the mean of the fk values obtained for all the extraction curves. This value was then kept constant, and Kf and Ks were calculated again. The values of these two parameters are close to those obtained before, with differences being of about 2–9%. The values of Kf (=kf a) vary from 1.19 × 10−3 to 6.02 × 10−3 s−1 when extractor 1 is considered, varying for the extractor 2 from 0.465 × 10−3 to 3.63 × 10−3 s−1 , which Table 2 Parameters of the Sovov´a model based on experimental data Pressure (MPa)
Dp (mm)
Vs (cm s−1 )
fk
Kf × 103 (s−1 )
Extractor 1 18
0.68
0.022
0.39
1.340
2.923
0.27
0.022 0.021 0.022 0.022 0.034 0.044 0.057 0.075 0.020 0.022
0.66
1.852 1.188 2.317 1.606 4.531 4.158 5.949 6.021 1.413 2.368
4.939 1.864 17.21 10.76 23.04 23.97 52.20 43.29 11.38 11.19
0.27
0.121 0.027 0.025
0.66
3.629 0.4654 0.7713
46.43 12.46 18.28
15 15 18 18 18 18 18 18 21 21 Extractor 2 18 18 18
Ks × 105 (s−1 )
fk , Fraction of oil directly exposed to CO2 ; Kf , overall mass transfer coefficient in the solvent phase; Ks , overall mass transfer coefficient in the solid phase.
346
M.G. Bernardo-Gil et al. / J. of Supercritical Fluids 40 (2007) 344–348
are within the ranges cited in several works [14–17]. The kf values can also be estimated by using the well known dimensionless relation Sh = 0.206 Re0.8 Sc0.33 [18], where Sh = kf Dp /D12 , Re = Vs ρ Dp /μ, and Sc = μ/(ρ D12 ) are the dimensionless Sherwood, Reynolds, and Schmidt numbers, respectively, with Dp the particle diameter, ρ the solvent density, and μ its viscosity. The diffusion coefficient, D12 , was estimated by using the Stokes-Einstein equation D12 = k T/(6πrμ) [19], where k is the Boltzman constant, and r is the solute molecular radius. The Kf values calculated by this method vary from 0.888 × 10−3 to 7.18 × 10−3 s−1 for the extractor 1, and from 2.28 × 10−3 to 8.85 × 10−3 s−1 for the extractor 2, agreeing well with those calculated by the Sovov´a model. The values of Ks calculated by the Sovov´a model with results obtained by using extractor 1 vary from 1.86 × 10−5 to 5.22 × 10−4 s−1 , and for the extractor 2, the values of this parameter vary from 1.25 × 10−4 to 4.64 × 10−4 s−1 . These values are within the ranges cited in several works [14–17].
Fig. 2. Acorn oil recovered by supercritical CO2 extraction as a function of mass of CO2 passed through bed per mass of oil-free ground acorn at 40 ◦ C and Vs = 0.022 cm s−1 , at different values of pressure, at Dp = 0.27 mm and Dp = 0.68 mm, by using extractor 1. Curves represent the simulation by using the Sovov´a model.
3.1. Effect of CO2 density and particle size Fig. 1 shows the percentage of acorn oil recovered by SFE, in relation to the quantity extracted by means of a Soxhlet, versus the CO2 density for medium particle size (Dp ) equal to 0.68 mm, superficial velocity (Vs ) of 1.3 cm min−1 , for 4 h of extraction time, and different values of pressure and temperature, by using the extractor 1. As expected, yields increased with the increase in pressure at a given temperature, a fact due to the increase of oil solubility. This increase results from the increase of the CO2 density, leading to the increase of its solvent power. Similarly, yields increase with the decrease in temperature at a given pressure, an effect attributed to the decrease of the CO2 density, which dominates over the increase of the solute vapour pressure at the pressure range in study, some possible retrograde effect not being reached. The percentage of oil recovered increases linearly with density in the studied ranges of pressures and temperatures. In Fig. 2 are presented the extraction curves of acorn oil, representing the acorn oil recovered by supercritical CO2 extraction as a function of mass of CO2 passed through the bed per mass of oil-free acorn, at 40 ◦ C, Dp = 0.27 mm, Vs = 1.3 cm min−1 , for
Fig. 1. Acorn oil recovered by supercritical CO2 extraction as a function of CO2 density at different values of pressure and temperature, being Dp = 0.68 mm, t = 4 h, and Vs = 0.023 cm s−1 , by using extractor 1.
Fig. 3. Yields of acorn oil extracted by supercritical CO2 as a function of extraction time at 18 MPa, 40 ◦ C, and Dp = 0.27 mm, by using extractor 1, for several values of “HB /D; Vs ”. Curves represent simulations obtained with the Sovov´a model.
15, 18 and 21 MPa, by using extractor 1. In this figure is also presented the extraction curve obtained at 40 ◦ C, Dp = 0.68 mm, Vs = 1.3 cm min−1 , for 18 MPa, by using the same extractor 1. For a given temperature, and using the same particle size, the initial rate of extraction increased as the pressure increased.
Fig. 4. Acorn oil recovered by supercritical CO2 extraction as a function of extraction time at 18 MPa, 40 ◦ C, and Dp = 0.27 mm, at several values of “extractor; HB /D; Vs ”. Curves represent simulations obtained with the Sovov´a model.
M.G. Bernardo-Gil et al. / J. of Supercritical Fluids 40 (2007) 344–348
Besides, the yield at a given mass of CO2 per mass of oil-free acorn is highly dependent on pressure. However, when extraction curves of acorn oil obtained at 18 MPa and at 21 MPa are compared, the differences on recoveries are not so significant. At the same conditions of CO2 density and superficial velocity, the initial extraction rate depends on the particle size. For smaller particles, greater fluid–solid contact areas exist, hence greater extraction rates. Besides, a higher amount of oil is released, as the seed cells are destroyed by milling, the amount directly exposed to the CO2 being more easily extracted, as related to fk (0.39 for Dp = 0.68 mm, and 0.66 for Dp = 0.27 mm). When extraction curves at 15 MPa and Dp = 0.27 mm and at 18 MPa and Dp = 0.68 mm are compared, it is evident that at the beginning of extraction, CO2 density is the preponderant factor, but after some time of extraction, the fraction of oil directly exposed to CO2 , fk , becomes more important.
347
Fig. 5. Influence of residence time in the supercritical extraction of acorn oil at 40 ◦ C, and Dp = 0.27 mm at different values of pressure, in extractor 1.
3.2. Effect of solvent flow rate and extractor geometry The effect of SCCO2 superficial velocity on the extraction rate and on the yield of acorn oil, when operating at 18 MPa, 40 ◦ C, and Dp = 0.27 mm, by using extractor 1, is shown in Fig. 3, which addresses two values of the ratio of the bed height, HB , to the extractor diameter (HB /D), 1.0 and 1.5. For the first value of this ratio, the superficial velocity was varied, and an increase of the initial extraction rate with this velocity was observed. If the yield is plotted versus the ratio of SCCO2 amount to the initial amount of ground acorn (kg CO2 /kg ground acorn), no significant differences in the extraction curves are found within the range of superficial velocities studied. This weak dependence upon the SCCO2 superficial velocity suggests that in the first part of the curve, the process is solubility controlled. The same can be said for the last part of the curve, where mass transfer is controlled by internal diffusion in the particle, an effect already observed by Kiriamiti et al. [20] and Salgin et al. [21]. In the same Fig. 3, it is possible to compare the curves at the same value of Vs = 1.3 cm min−1 (flow rate = 18.3 g min−1 ) for extractor 1 (D = 4.57 cm), but different values of HB /D. As was expected, for the same extraction time, as HB /D increases, the yield decreases, conducting to a lower initial extraction rate, the same fact being observed by Moura et al. [22]. The same effect as above can be observed in Fig. 4, with extractor 2 (D = 2.13 cm), when operating at the same conditions of 18 MPa, 40 ◦ C, and Dp = 0.27 mm. For HB /D = 3.8, as superficial velocity increases from 1.5 cm min−1 (flow rate = 4.41 g min−1 ) to 7.3 cm min−1 (flow rate = 18.8 g min−1 ), an increase of the initial extraction rate was observed (0.021 and 0.076 kg oil/kg ground acorn/s, respectively). If the yield is plotted versus the ratio of SCCO2 amount to the initial amount of ground acorn (kg CO2 /kg ground acorn), no significant differences in the extraction curves are, once again, found for the superficial velocities studied. This weak dependence upon the SCCO2 superficial velocity points to the same conclusions as above, both for the first and the last parts of the curve. A comparison can also be made between the curves at the same value of Vs = 1.5 cm min−1 (flow rate = 4.41 g min−1 ) for extractor 2, but different values of HB /D, in Fig. 4. For the same
Fig. 6. Influence of residence time in the supercritical extraction of acorn oil at 40 ◦ C, 18 MPa, and Dp = 0.27 mm, in extractor 2.
extraction time, as HB /D increases, the yield decreases, conducting to lower initial extraction rate (0.012 kg oil/kg ground acorn/s), as above. The extraction results obtained in the two extractors were further assessed for similar values of superficial velocities. Some conclusions can be also drawn from Fig. 4. By comparing extractors 1 and 2, with H/D of 1.5 and 12, respectively, the determinant factor with similar superficial velocities is HB /D, conducting to a decrease on the initial extraction rate as HB /D increases. 3.3. Effect of residence time on loading Figs. 5 and 6 show the influence of residence time on the supercritical extraction of acorn oil, in extractors 1 and 2, respectively. As can be seen, the curves of loading (gram of extracted oil/kg of CO2 passed) are independent of residence time, and start decreasing at about 80% of extracted oil, under the conditions studied. In extractor 2, loadings are lower than in extractor 1 until almost 80% of extracted oil, and decrease after that, as in extractor 1. 4. Conclusions For the conditions used in this work, which was developed for the extraction of Q. rotundifolia L. acorn oil with supercritical
348
M.G. Bernardo-Gil et al. / J. of Supercritical Fluids 40 (2007) 344–348
carbon dioxide, it can be concluded that the oil recoveries are linearly dependent on the CO2 density. Also, at the same conditions of CO2 density and superficial velocity, the initial extraction rate depends on the ground acorn particle size. Besides, at the beginning of extraction, CO2 density is the preponderant factor, but, after some time of extraction, the fraction of oil directly exposed to CO2 , fk , becomes more important. For the same superficial velocity, yields and initial extraction rates increase with the decrease of HB /D. For a given HB /D, yields and initial extraction rates increase with the superficial velocity. The Sovov´a model was successfully applied to model the acorn oil extraction achieved in this work. Acknowledgement The authors thank the financial support offered by the project PRAXIS XXI – 2/2.1/BIO/1066/95 (FCT – Fundac¸a˜ o para a Ciˆencia e Tecnologia – Portugal). References [1] Gorduras e o´ leos comest´ıveis – o´ leo de bolota. Definic¸a˜ o, caracter´ısticas e acondicionamento (fats and edible oils – acorn oil, definition, characteristics and packaging) NP 3373, 1989. [2] I.M.G. Lopes, M.G. Bernardo-Gil, Characterisation of acorn oils extracted by hexane and by supercritical carbon dioxide, Eur. J. Lipid Sci. Technol. 107 (2005) 12–19. [3] D. Horrobin, M. Manku, How do polyunsaturated fatty acids lower plasma cholesterol levels? Lipids 18 (1983) 558–562. [4] M.M. Esqu´ıvel, M.G. Bernardo-Gil, Extraction of olive husk oil with compressed carbon dioxide, J. Supercrit. Fluids 6 (1993) 91–94. [5] M.A. G´omez, C.P. L´opez, E. Mart´ınez de la Ossa, Recovery of grape seed oil by liquid and supercritical carbon dioxide: a comparison with conventional solvent extraction, Chem. Eng. J. 61 (1996) 227–321. [6] C. Devittori, D. Gumy, A. Kusy, L. Colarow, C. Bertoli, P. Lambelet, Supercritical fluid extraction of oil from millet bran, J. Am. Oil Chem. Soc. 77 (2000) 573–579. [7] M.G. Bernardo-Gil, C. Oneto, A. Antunes, M.F. Rodrigues, J.M. Empis, Extraction of lipids from cherry seed oil by supercritical carbon dioxide, Eur. Food Res. Technol. 212 (2001) 170–174.
[8] R. Oliveira, M.F. Rodrigues, M.G. Bernardo-Gil, Characterization and supercritical carbon dioxide extraction of walnut oil, J. Am. Oil Chem. Soc. 79 (2002) 225–230. [9] M.G. Bernardo-Gil, L.M.C. Lopes, Supercritical fluid extraction of Cucurbita ficifolia seeds oil, Eur. Food Res. Technol. 219 (2004) 593– 597. [10] H. Sovov´a, Rate of the vegetable oil extraction with supercritical CO2 —I. Modelling of extraction curves, Chem. Eng. Sci. 49 (1994) 409– 414. [11] M.M. Esqu´ıvel, M.G. Bernardo-Gil, M.B. King, Mathematical models for supercritical extraction of Olive Husk Oil, J. Supercrit. Fluids 16 (1999) 43–58. [12] J. Chrastil, Solubility of solids and liquids in supercritical gases, J. Phys. Chem. 86 (1982) 3016–3021. [13] R.V. Vasconcellos, F.A. Cabral, A new method for estimating solubility of fatty acids, esters, and triglycerides in supercritical carbon dioxide, J. Am. Oil Chem. Soc. 78 (2001) 827–829. [14] H. Sovov´a, J. Kucera, J. Jez, Rate of the vegetable oil extraction with supercritical CO2 —II. Extraction of grape oil, Chem. Eng. Sci. 49 (1994) 415–420. [15] N.P. Povh, M.O.M. Marques, M.A.A. Meireles, Supercritical CO2 extraction of essential oil and oleoresin from chamomile (Chamomilla recutita [L.] Rauschert), J. Supercrit. Fluids 21 (2001) 245–256. [16] S.R.S. Ferreira, M.A.A. Meireles, Modeling the supercritical fluid extraction of black pepper (Piper nigrum L.) essential oil, J. Food Eng. 54 (2002) 263–269. [17] V. Louli, G. Folas, E. Voutsas, K. Magoulas, Extraction of parsley seed oil by supercritical CO2 , J. Supercrit. Fluids 30 (2004) 163–174. [18] J.M. del Valle, J.C. Germain, E. Uquiche, C. Zetzl, G. Brunner, Microstructural effects on internal mass transfer of lipids in prepressed and flaked vegetable substrates, J. Supercrit. Fluids 37 (2006) 178– 190. [19] M.B. King, O. Catchpole, Physico-chemical data required for the design of near-critical fluid extraction process, in: M.B. King, T.R. Bott (eds), Extraction of Natural Products using near-critical solvents, Blackie Academic & Professional, Glasgow, UK, 1993. [20] H.K. Kiriamiti, E. Rascol, A. Marty, J.S. Condoret, Extraction rates of oil from high oleic sunflower seeds with supercritical carbon dioxide, Chem. Eng. Proc. 41 (2001) 711–718. [21] U. Salgin, A. C ¸ alimli, B.Z. Uysal, Supercritical fluid extraction of jojoba oil, JAOCS 81 (2004) 293–296. [22] L.S. Moura, R.N. Carvalho Jr., M.B. Stefanini, L.C. Ming, M.A.A. Meireles, Supercritical fluid extraction from fennel (Foeniculum vulgare): global yield, composition and kinetic data, J. Supercrit. Fluids 35 (2005) 212– 219.
