Supercritical carbon dioxide as solvent in the lipase-catalyzed ethanolysis of fish oil: kinetic study

Supercritical carbon dioxide as solvent in the lipasecatalyzed ethanolysis of fish oil: kinetic study Rodrigo Melgosaa, M. Teresa Sanz*a, Ángela G. Solaesaa, Esther de Paza,Sagrario Beltrána, Daniela L. Lamasb aDepartment of Biotechnology and Food Science (Chemical Engineering Section), University of Burgos, Burgos, Spain bInstituto Nacional de Investigación y Desarrollo Pesquero (INIDEP), Buenos Aires, Argentina *[email protected] Fax: 0034947258831 Abstract Supercritical carbon dioxide (SC-CO2) has been used as green solvent in the lipase-catalyzed ethanolysis of fish oil by Lipozyme RM IM at mild, non-oxidative conditions and with no solvent residues. The effect of experimental conditions, initial substrate ethanol/oil molar ratio (2-38), pressure (7.5-30 MPa), and temperature (323.15-353.15 K) on equilibrium conversion, reaction rate and oxidative status of the products has been studied. No ethanol inhibition has been observed at high concentrations of ethanol, when putting in contact first the fish oil with the enzyme avoiding direct contact between the biocatalyst and ethanol. Operating pressure affected positively the reaction performance in the range investigated. Visual observation of the phase behaviour of the initial reaction mixture showed an “expanded liquid phase” that helped enhancing reaction rate, and a gas phase. Raising temperature accelerated the reaction up to a limit (343.15 K), observing higher enzyme thermal stability than in other reaction media (313.15 K). However, lipid oxidation increases with temperature. Up to 86 ± 1 % FAEE yield has been found at MR = 6:1, 30 MPa and 323.15 K. Kinetic data have been correlated by using a mathematical model based on the elementary reactions of the 3-step transesterification. Kinetic rate constants, apparent activation volumes and energies are reported for the first time for a lipase-catalyzed ethanolysis reaction in SC-CO2. Keywords Omega 3, lipase, ethanolysis, supercritical carbon dioxide. 1. Introduction Fish oil is a natural source of omega-3 polyunsaturated fatty acids (n-3 PUFAs) such as eicosapentaenoic acid (EPA, 20:5 n-3) and docosahexaenoic acid (DHA, 22:6 n-3). Health benefits of these compounds have been well established in the literature [1]. As a consequence, functional foods enriched with n-3 PUFAs have been the type of functional food products whose production in Europe and USA has increased the most in the last years [2]. Nevertheless, in a recently published review it has been found that the excess of oxidation in commercial n-3 PUFA supplements affects between 11-62% of the analysed supplements [3]. Traditional methods for production of n-3 PUFA concentrates from their natural sources have been recently reviewed, and a number of novel techniques have been proposed [2]. Among the latest, enzymatic modification of oils rich in n-3 PUFAs in supercritical fluids (SCFs) rises as an alternative for obtaining less oxidized fish oil derivatives, compared to conventional methods. Several studies have been carried out on enzymatic reactions in different SCF media. A comprehensive review on this subject was carried out by Knez [4], with references on different enzymatic reactions in dense gases, such as oxidation, hydrolysis, esterification and transesterification. Supercritical carbon dioxide (SC-CO2) is probably the most used SCF due to its benefits (non-toxic, non-flammable, readily available at high purities and low costs, and relatively mild critical conditions, easily separated from the reaction products by simple depressurization) that are appealing when choosing environmental replacement for organic solvents. Besides, by varying the temperature and pressure it allows the fractionation of the products. Some previous studies of enzymatic ethanolysis of natural lipid sources in SC-CO2 have been reported in the literature. Different immobilized lipases have been used as biocatalyst, such as the non-specific lipase Novozyme 435 from Candida Antarctica [5-8] and the sn-1,3regiospecific lipase Lipozyme TL-IM from Thermomuces lanuginosa [9, 10]. In this work, Lipozyme RM IM from Rhizomucor miehei, a sn-1,3 specific lipase, has been used as biocatalyst. Ethanolysis of palm kernel oil in SC-CO2 by the homologous Lipozyme IM was studied by Oliveira and Oliveira [7]. Although the biocatalyst was reported to be sn-1,3specific, it did not behave as a regiospecific lipase and considerable amounts of glycerol were found in the reaction products. Furthermore, the reaction conversion was followed in terms of glycerol production, not taking into account the reaction intermediates (diand monoacylglycerides). Kondo et al. [11] carried out the synthesis of fatty acid ethyl esters (FAEEs) from SC-CO2-extracted canola oil in a continuous supercritical extraction-reaction (SFE-SFR) system by using Lipozyme RM IM as biocatalyst, observing a decrease in the FAEE production at high ethanol concentration. Enzymatic reactions in SC-CO2 can be affected by operating pressure in different ways. According to transition state theory and standard thermodynamics, operating pressure can affect the reaction rate constants. Besides, density-related changes in the physical parameters of SC-CO2 may indirectly affect the enzyme catalytic activity, and thus the reaction performance [12]. Loss et al. [13] have recently reviewed different applications of supercritical fluids as alternative solvent for biocatalysis processes, concluding that there seems to be no “rule of thumb” for predicting the effect of pressure on enzyme activity in SCCO2. Direct effects of pressure on enzyme residual activity and stability of Lipozyme RM IM have been previously investigated, finding that almost no changes occurred in the range between 10 and 25 MPa at 323.15 K [14]. Different results have been found in the literature regarding the effect of pressure on Lipozyme RM IM-catalyzed reactions in SC-CO2. For instance, in the study of esterification of stearic acid with ethanol catalyzed by Lipozyme IM in SC-CO2 in the range from 6 to 20 MPa at 323.15 K, Nakaya et al. [15] found an increase in esterification rate with an increase in pressure, but a maximum was found in the hydrolysis rate of the corresponding ethyl stearate. Laudani et al. [16] performed a detailed kinetic and thermodynamic study of the esterification of oleic acid with 1-octanol catalyzed by Lipozyme RM IM in dense carbon dioxide. These authors reported a positive effect of increasing pressure from 8 to 10 MPa at 323.15 K. Further increase in pressure up to 30 MPa led to a decrease in reaction conversion from 84 % to 77 %. Therefore, the effect of pressure on enzyme activity in CO2 is very dependent not only on the specific enzyme, but also on the reaction studied and the phase behaviour of the system at different pressure and temperature conditions. In this work, the effect of the initial molar ratio of substrates, pressure, and temperature on equilibrium yield and reaction rate has been studied for the ethanolysis of fish oil in SC-CO2, covering a wider range than previous works reported in the literature. Additionally, oxidation parameters of the refined fish oil and the reaction products obtained from the reactions in SCCO2 have been determined and compared with those obtained from reactions performed in conventional organic solvents and in solvent-free media at atmospheric pressure. This way, optimal reaction conditions considering kinetic aspects and quality of the products can be determined. Experimental data have been satisfactorily correlated by a simple semi-empirical kinetic model based on the elementary reactions that may occur in this system and taking into account reaction intermediates. 2. Material and Methods 2.1. Materials Lipozyme RM IM, a lipase from Rhizomucor miehei immobilized on a macroporous resin, was purchased from Novozymes A/S (Denmark). Refined fish oil was kindly provided by AFAMSA S.A. (Spain) being a mixture of tuna (Thunnus sp.) and sardine (Sardina pilchardus) oil. Fatty acid profile of the fish oil has been previously reported with a 24 % mol of docosahexaenoic acid (DHA) and 7 % mol of eicosapentanoic acid (EPA) [17]. Absolute ethanol (99.9 %) was purchased from Merck KGaA. Carbon dioxide (99.9%) was supplied by Air Liquide S.A. (Spain). All other chemicals used in different analyses were of analytical or HPLC grade. 2.2. Ethanolysis of fish oil in SC-CO2 The ethanolysis reaction has been performed in a high pressure batch stirred tank reactor (HPBSTR) made of stainless steel (SS-316) and having an internal volume of 100 mL. A schematic diagram of the experimental apparatus is shown in Figure 1. In a typical experiment, a weighed amount of enzyme (5.0 % wt. of substrates) was added into the reactor together with a known amount of fish oil. Subsequently, ethanol was added according to the established initial substrate molar ratio. This procedure was adopted in order to avoid direct contact of ethanol with the enzyme, which may cause inactivation of the catalyst. The reactor was then closed, connected to the pressure circuit and placed in a thermostatic water bath at the desired operating temperature. Subsequently, SC-CO2 was fed into the reactor by means of a high pressure pump (ISCO 260 D) up to the desired pressure, which was maintained by a digital pressure controller. A Bourdon pressure gauge also provided a secondary lecture. Once the established conditions have been reached, magnetic stirring was connected and the reaction was initiated. Figure 1. Schematic diagram of the high pressure apparatus used for the ethanolysis reactions in SC-CO2. 1: CO2 reservoir; 2: syringe pump; 3: cryostat; 4: rupture disk; 5: high pressure batch stirred tank reactor; 6: thermostatic water bath; 7: magnetic stirrer; 8: sampling device. Operating temperature and pressure have been varied in the range between 323.15-353.15 K and 7.5-30 MPa, respectively. The effect of the initial substrate molar ratio has been studied in the range from 2:1 to 38:1 (ethanol:fish oil). Samples were taken periodically during 24 h through a siphoned capillary equipped with a microfilter made of sintered steel, which prevented the withdrawal of the enzyme from the reaction mixture. Samples were collected in glass screw-top vials immersed in a cold trap and stored at -18 ºC up to analysis. Pressure drops up to 0.5 MPa were observed during the withdrawal of the samples, which were compensated by feeding fresh SC-CO2 at the desired pressure into the reactor. According to the low mass of the samples (ca. 0.1 g) compared with the initial loading of the HP-BSTR, disturbances of the batch process were considered negligible. 2.3. Determination of the composition Neutral lipid profile of the samples (fatty acid ethyl esters, FAEEs; Monoacylgycerides, MAGs; diacylglycerides, DAGs; and unreacted triacylglycerides, TAGs) has been determined by normal phase HPLC. Chromatographic equipment, method and calibration procedure have been previously described in detail [18]. Chromatographic analysis of glycerol (GLY) content in the reaction samples was performed using High-Temperature Gas Chromatography (HTGC). Method and calibration procedure have been previously described [17]. GLY content in the reaction samples was also theoretically calculated by a balance of the glycerol backbone, as proposed by Sovová et al. for the enzymatic hydrolysis of blackcurrant oil in SC-CO2 [19]. A modified expression for an ethanolysis reaction gives: nGLYt = (nFAEE – nDAG – 2·nMAG)/3 (1) where nFAEE , nDAG , and nMAG are the FAEE, DAG and MAG mole content in the reaction samples and nGLYt is the theoretical GLY mole content. Theoretical calculation deviated less than 10% from experimental data, thus HT-GC determination of GLY and neutral lipid profile analysis were satisfactorily related. Unreacted EtOH was theoretically calculated considering the reaction stoichiometry, in which the production of 1 mol of FAEE consumes 1 mol of EtOH, giving: nEtOH = nEtOH,o – nFAEE (2) where nEtOH,o is the initial mole content of ethanol. 2.4. Measurement of lipid oxidation Determination of the peroxide value (PV), p-anisidine value (p-AnV), and acid value (AV) of the samples before and after the kinetic experiments have been performed in order to evaluate potential lipid oxidation processes during the ethanolysis reactions. The peroxide value, PV, measures the concentration of peroxides and hydroperoxides formed in the initial stages of lipid oxidation (primary oxidation). The p-anisidine value (p-AnV) is an estimation of the concentration of secondary oxidation products. Determination of the acid value (AV) has been also performed as an estimation of the hydrolytic rancidity of the fish oil and the reaction samples. All determinations were performed according to standard methods [20-22]. In the case of reaction samples, lipid fractions were obtained by means of evaporation of unreacted ethanol in a vacuum rotary evaporator (Heidolph). 2.5. Kinetic model Lipase-catalyzed ethanolysis of triacylglycerydes (TAG) of fish oil can be considered as a 3step transesterification. At each step, one molecule of FAEE and a glyceride containing one fewer ester bond are obtained. Glycerides involved in the reaction are diand monoacylglycerides (DAG and MAG), and glycerol (GLY) as the last product. Following the proposed model, the reaction takes place through the following steps: 1. Conversion of trito diacylglycerides: TAG +EtOH k1 ⇄ k−1 DAG +FAEE (3) 2. Conversion of dito monoacylglycerides: DAG +EtOH k2 ⇄ k−2 MAG + FAEE (4) 3. Conversion of monoacylglycerides to glycerol: MAG + EtOH k3 ⇄ k−3 GLY +FAEE (5) To correlate the experimental kinetic data, a semi-empirical model based on the mass balance equations of all the species in the reaction system has been employed. Although the sn-1,3specific catalyst cannot deacylate the sn-2 position of the acylglyceride, step 3 (Eq. 5) should be considered because isomerization of 2-MAG to 1(3)-MAG (acyl-migration) may occur. As the regioisomers 1,2and 1,3-DAG; and 1(3)- and 2-MAG could not be distinguished with the applied analytical procedure, no difference was made between them in the model. Hydrolysis reaction has not been taken into account since no free fatty acids were detected (<0.1%). The kinetic equations involved in the ethanolysis system are the following: d(nTAG/ntotal)/dt = – k’1·xTAG·xEtOH + k’–1·xDAG·xFAEE (6.1) d(nDAG/ntotal)/dt = k’1·xTAG·xEtOH – k’–1·xDAG·xFAEE – k’2·xDAG·xEtOH + k’–2·xMAG·xFAEE (6.2) d(nMAG/ntotal)/dt = k’2·xDAG·xEtOH – k’–2·xMAG·xFAEE – k’3·xMAG·xEtOH + k’–3·xGLY·xFAEE (6.3) d(nGLY/ntotal)/dt = k’3·xMAG·xEtOH – k’–3·xGLY·xFAEE (6.4) d(nFAEE/ntotal)/dt = k’1·xTAG·xEtOH – k’–1·xDAG·xFAEE + k’2·xDAG·xEtOH – k’–2·xMAG·xFAEE + k’3·xMAG·xEtOH – k’–3·xGLY·xFAEE (6.5) d(nEtOH/ntotal)/dt = – k’1·xTAG·xEtOH + k’–1·xDAG·xFAEE – k’2·xDAG·xEtOH + k’–2·xMAG·xFAEE – k’3·xMAG·xEtOH + k’–3·xGLY·xFAEE (6.6) Equilibrium constants for each reaction step, evaluated as Ki = k’i/k’–i, are also reported. To our knowledge, this is the first time that kinetic rate constants are reported for the three steps of a lipase-catalyzed ethanolysis in SC-CO2. Figure 4. a) Equilibrium FAEE yield (%) vs. MR (ethanol :oil) in the ethanolysis of fish oil catalyzed by Lipozyme RM IM in SC-CO2 medium. Solid line is from the non-linear regression of the experimental data (Eq. 9). b) Logarithmic relationship between initial reaction rate and MR. Dashed line is from the linear regression of the experimental data. Experimental conditions: p = 10 MPa, T = 323.15 K and enzyme loading 5 % wt. of substrates As it can be observed from Table 2, the forward and reverse rate constants follow the order k’3 > k’2 > k’1 and k’–2 > k’–3 > k’–1. An increase in MR leads to an increase in both the forward and the reverse rate constants. In the MR range studied, the forward rate constant of the third step (MAG to produce FAEE, k’3) is larger than the other two forward rate constants, being the initial breakdown of TAG the slowest step and therefore the rate-limiting step of the ethanolysis reaction. This behaviour has been also described in the literature for either acid or base-catalyzed transesterification [30]. It can be also noticed that equilibrium constants monotonously decreased with increasing MR due to the excess of ethanol employed. Besides, the equilibrium constant of the third reaction step, K3, is one order of magnitude higher than those of the other two steps. In the literature, different mechanisms have been proposed for lipase-catalyzed transesterification systems, most of them based on Ping-Pong Bi-Bi models [31]. Comparison with these studies is difficult since in most cases thermodynamic parameters are not provided. In this work, a simple model was adopted and surprisingly, leads to kinetic and equilibrium parameters of the same order as those reported by chemicalcatalysis [30]. Table 2. Effective forward (k’i) and reverse (k’–i) reaction rate constants, equilibrium constants (Ki) and objective function (O.F.) values of the proposed kinetic model for the ethanolysis of fish oil by Lipozyme RM IM in SC-CO2 at different initial molar ratio of substrates (MR). Reactions were performed at p = 10 MPa, T = 323.15 K, enzyme loading 5 % wt. of substrates. MR (ethanol:oil) k’ 1 (min -1 ) k’ –1 (min -1 ) K1 k’ 2 (min -1 ) k’ –2 (min -1 ) K2 k’ 3 (min -1 ) k’ –3 (min -1 ) K3 O.F. 2:1 0.0971 0.5998 0.1543 1.3972 2.4197 0.5511 3.7926 0.9194 4.1315 0.1704 4:1 0.1703 0.8898 0.1786 1.5682 4.0452 0.3684 5.3823 1.3874 4.0460 0.0592 6:1 0.4764 1.3465 0.4867 3.2605 5.9020 0.5290 7.4221 2.5333 2.6911 0.0309 10:1 0.5177 6.7735 0.0946 22.2968 164.8420 0.1426 11.5229 7.5545 2.1538 0.0137 38:1 3.5370 34.7671 0.0152 16.3786 158.3258 0.1178 27.8656 56.5390 0.2191 0.0047 3.2.Pressure effect Pressure has been varied in the range from 7.5 to 30 MPa at fixed MR = 6:1 (ethanol:fish oil), 323.15 K and an enzyme loading of 5 % wt. of substrates. Table 1 shows that initial reaction rate steadily increases with pressure from 7.5 to 30 MPa, and FAEE yield at equilibrium increases from 51.4 % near the critical pressure (7.5 MPa) to 80 % at 9 MPa, showing a plateau around 81-85 % at higher pressures up to 30 MPa (Figure 5). In the range investigated (7.5-30 MPa), no decay in the reaction performance was observed when increasing pressure. Figure 5. Effect of operating pressure (p) on the ethanolysis of fish oil catalyzed by Lipozyme RM IM in SCCO2 medium. Experimental conditions: MR = 6:1, T = 323.15 K and enzyme loading 5 % wt. of substrates. Lines represent the fitting of the proposed kinetic model to the experimental data Figure 6 shows that a semi-logarithmic relationship can be established between initial reaction rate and operating pressure (dashed line, ln ro = 0.4220·ln pr + 3.6157; R2 = 0.9774). Besides, a similar expression to the one proposed in Eq. 9 can be established to describe the effect of pressure on the equilibrium FAEE yield. Operating pressure was expressed in terms of reduced pressure (pr = p/pC). Non‐linear regression was performed by using the Marquardt algorithm (Statgraphics) giving a limiting equilibrium FAEE yield of 83.47 %, b = 0.079 and pr0 = 0.982 (p0 = 7.22 MPa) with R2 = 0.973. The continuous line in Figure 6 corresponds to this adjustment. Figure 6. Open triangles: Equilibrium FAEE yield (%) vs. reduced operating pressure (pr = p/pC) in the ethanolysis of fish oil catalyzed by Lipozyme RM IM in SC-CO2 medium. Solid line is from the non-linear regression of the experimental data (Eq. 9). Filled circles: logarithmic relationship between initial reaction rate and pr. Dashed line is from the linear regression of the experimental data. Experimental conditions: MR = 6:1, T = 323.15 K and enzyme loading 5 % wt. of substrates The effect of operating pressure on the ethanolysis of fish oil by Lipozyme RM IM in SC-CO2 could be explained considering the phase behaviour of the fish oil/ethanol reaction mixture in SC-CO2. Visual observations of the initial reaction mixture at the different pressures assayed in this work showed that the reaction system consisted of an “expanded” liquid and a gas phase. At the conditions studied in this work, solubility of CO2 in ethanol is high [32], whereas Borch-Jensen and Mollerup [33] reported moderate solubility of CO2 in fish oil at pressures from 6 to 65 MPa and temperatures from 293.15 to 393.2 K. At none of the temperatures, carbon dioxide and fish oil were completely miscible, but at each temperature, CO2 solubility in fish oil was found to increase with increasing pressure [33]. This increase in CO2 solubility with pressure may correspond to the increase in the reaction rate with pressure as shown in Figure 6, since CO2 solvation results in a better mass transfer reaction medium due to an increase in diffusivity and a reduction of medium viscosity. The different effects of pressure depending on the phase behaviour of the system can be noted when comparing the results obtained in this work with those from the esterification of oleic acid with n-octanol in SC-CO2 [16]. Phase behaviour of the system oleic acid + n-octanol + CO2 shows that CO2 is highly soluble in the reaction mixture (oleic acid and ethanol) and the liquid phase can contain up to 70 % mole of CO2 near the critical point of CO2 [16]. Further increase in operating pressure promoted a decrease in conversion due to more CO2 solvating in the reaction bulk and leading to dilution of the substrates. On the contrary, the mixture fish oil + ethanol can dissolve a much lower amount of CO2 (30 % mole at MR = 6:1, 10 MPa and 323.15 K, according to analytical determination of the phase behaviour in a high pressure view cell). Therefore, dilution of the substrates at high pressure is not supposed to strongly affect the reaction performance in the fish oil + ethanol + CO2 reaction system. Oliveira and Oliveira [7] adopted a Taguchi experimental design to assess the influence of the process variables on the ethanolysis of palm kernel oil in SC-CO2 catalyzed by Lipozyme IM. According to their results, Lipozyme IM was positively affected by pressure, although MR was the variable that more strongly affected the conversion. For this enzyme they found an optimum at 14.6 MPa. On the contrary, in the present work, no maximum in the pressure was found in the range from 7.5 to 30 MPa, although similar phase behaviour could be expected for both reaction mixtures (fish oil + ethanol + CO2 or palm kernel oil + ethanol + CO2). The effect of pressure on the kinetic rate constants has been taken into account by using the transition-state theory and classical thermodynamics. Following this approach, the variation of the reaction rate constant k’i with pressure for a bimolecular reaction “A + B = M* = products” can be expressed as follows [12]: (∂ln (k)/∂p)T = – ΔV*/RT (10.1) where k is the rate constant of the reaction expressed in pressure-independent concentration units, ΔV* is the apparent activation volume, T is the operating temperature and R is the gas constant. The direct integration of Eq 10.1 is not straightforward since activation volume changes with pressure [12]. However, within a small range of pressure it can be assumed that ΔV* does not change with pressure and Eq. 10.1 can be easily integrated, giving: k’i = k0,i·exp(– p·ΔVi*/RT) (10.2) where the subscript i refers to the different steps in the ethanolysis reaction. Following this expression, pre-exponential kinetic constants, k0i, and apparent activation volumes, ΔVi*, have been simultaneously estimated for the experiments performed in the range 9-30 MPa. Results obtained are listed in Table 3. Kinetic parameters for the experiment performed at p = 7.5 MPa were estimated separately because of the marked changes in the physical properties of the solvent near the critical region (pr = 1.02). In the pressure range 9-30 MPa, apparent activation volumes were negative (ΔVi* < 0) for all the reaction steps, which indicates that reaction rate constants will increase with increasing pressure. It can be observed that ΔV* for the forward first and third reaction steps are lower (or higher in absolute value) than the corresponding ΔV* for the reverse reaction, being this steps more sensitive to an increase in operating pressure. On the contrary, the DAG to MAG conversion shows similar ΔV* values for the forward and reverse reactions, which suggest little influence of pressure in this step. Overall consideration of ΔV* values indicates that FAEE production may be favoured by increasing pressure, as it can be observed from the experimental results. To our knowledge, no ΔV* values have been previously reported for lipase-catalyzed ethanolysis in SC-CO2. However, a similar ΔV* value of ca. -206 cm3·mol-1 was reported by He et al. [34] for the transesterification of soybean oil without catalyst in suband supercritical methanol between 8.7 MPa and 36 MPa (553 K and MR methanol:soy bean oil = 42:1). In any case, Kamat et al. [12] stated that the use of ΔV* in enzyme-catalyzed reactions must be treated with caution, since changes in pressure will result in multiple variables being changed that also influence the ability of the enzyme to catalyse a given reaction. Therefore ΔV* should not be used to compare the effects of pressure for catalyzed and uncatalyzed reactions. For an uncatalyzed reaction, data are only dependent of direct pressure-effects and no indirect effects of pressure are transmitted via the enzyme. Table 3. Values for the forward (k0,i) and reverse (k0,–i) pre-exponential constants, apparent activation volume of each forward (ΔV*i) and reverse (ΔV*–i) reaction step and values of the objective function (O.F.) of the proposed kinetic model for the ethanolysis of fish oil by Lipozyme RM IM in SC-CO2 at 7.5 MPa and in the range 930 MPa. Reactions were performed at MR = 6:1, T = 323.15 K, enzyme loading 5 % wt. of substrates. p (MPa) k 0,1 (min -1 ) k 0,–1 (min -1 ) ΔV* 1 (cm 3 mol -1 ) ΔV* –1 (cm 3 mol -1 ) k 0,2 (min -1 ) k 0,–2 (min -1 ) ΔV* 2 (cm 3 mol -1 ) ΔV* –2 (cm 3 mol -1 ) k 0,3 (min -1 ) k 0,–3 (min -1 ) ΔV* 3 (cm 3 mol -1 ) ΔV* –3 (cm 3 mol -1 ) O.F. 9 – 30 0.2366 1.1770 -158.0394 -47.6628 1.5824 2.6906 -126.3835 -127.7673 3.0135 1.7217 -145.7125 -88.7143 0.0889 7.5 k’1= 0.3099 k’-1= 43.1938 k’2= 2.3257 k’-2= 12.1366 k’3= 43.8870 k’-3= 45.9396 0.0087 3.3. Temperature effect To assess the effect of temperature on the kinetics of the ethanolysis of fish oil by Lipozyme RM IM in SC-CO2, operating temperature has been varied between 323.15 and 353.15 K. Initial substrate molar ratio (6:1 ethanol:fish oil), pressure (10 MPa) and enzyme loading (5% wt. of substrates) remained unchanged. Figure 7 shows that FAEE yield at equilibrium was similar in the temperature range from 323.15 to 343.15 K. The heat of reaction is generally small for many transesterification systems; therefore the equilibrium conversion observed for the ethyl esters is essentially temperature independent [35]. Raising temperature from 323.15 to 343.15 K resulted in an increase of the initial reaction rate (Table 1 and Figure 8), probably because of a higher kinetic energy of the molecules. Besides, lower viscosity and higher diffusivity of the solvent at higher temperatures may lead to lower mass transfer limitations [4]. The highest temperature assayed in this work (353.15 K) led to lower equilibrium FAEE yield (Table 1 and Figure 7), which may be due to thermal deactivation of the catalyst. The effect of temperature on initial reaction rate is shown in the Arrhenius plot (Figure 8). From this figure, it can be seen that the Arrhenius dependence is no longer valid at temperatures higher than 343.15 K. As it has been previously mentioned, this behaviour may correspond to thermal deactivation of the catalyst. Figure 7. Effect of operating temperature (T) on the ethanolysis of fish oil catalyzed by Lipozyme RM IM in SC-CO2 medium. Experimental conditions: MR = 6:1, p = 10 MPa and enzyme loading 5 % wt. of substrates. Lines represent the fitting of the proposed kinetic model to the experimental data. Similar results for thermal behaviour of Lipozyme RM IM have been reported in the literature for the synthesis of n-octyl oleate in SC-CO2 [16]. Conversion around 80 % was observed in the range 308.15-333.15 K at 10 MPa, whereas higher temperatures (343.15 and 353.15 K) led to lower conversion (around 65 %) yet slightly higher initial reaction rates, which may indicate that thermal deactivation is not immediate. Oliveira and Oliveira [7] reported T = 324 K as the optimum temperature for the ethanolysis of palm kernel oil by Lipozyme IM in SC-CO2, whereas 313.15 K was found to be the optimum for the same reaction in n-hexane [8]. Recently, Calero et al. [28] have found the same operating temperature (T = 313.15 K) as the optimum for the ethanolysis of sunflower oil by Lipozyme RM IM in solvent-free media Figure 11. Effect of the ethanolysis of fish oil catalyzed by Lipozyme RM IM on the peroxide value (PV) in different reaction media: SC-CO2 at different operating pressure, SF (solvent-free media) and 20 wt% of TP. Dashed line represents the recommended limit set by the FDA [38]. To our knowledge, no other studies in the literature assessed lipid oxidation processes during enzymatic ethanolysis of fish oil in SC-CO2. Park et al. [39] determined PV and conjugated diene (CD) content of commercial salmon oil before and after enzymatically (Lipozyme IM) and chemically catalyzed ethanolysis at atmospheric pressure and n-hexane as the reaction media. They found that both methods increased PV and CD, yet little oxidation and isomerization of PUFAs were found when Lipozyme IM was used as the catalyst [39]. Results obtained in this work show that enzymatic ethanolysis of fish oil in SC-CO2 can be considered a suitable method to obtain less oxidized reaction products compared to those obtained by enzymatic ethanolysis in conventional organic solvents or in solvent-free media at atmospheric pressure. 4. Conclusions SC-CO2 has been used as a green solvent in the transesterification of fish oil by Lipozyme RM IM, providing an environmentally benign reaction medium. Advantages of using SC-CO2 include replacing organic solvents, enhancing reaction kinetics by reducing mass transfer limitations and preventing oxidation due to displacement of oxygen. The latter is especially important when working with easily oxidizable compounds such as n-3 PUFAs. Enzyme and phase behaviour are key parameters to understand bioconversion in SC-CO2. The lipase showed higher thermal stability in SC-CO2 reaction medium than in other conventional reaction media. Besides, no ethanol inhibition has been observed when avoiding high concentrations of ethanol in the enzyme environment. Operating pressure affected positively the reaction performance due to solvation of CO2 in the reaction mixture, which reduces viscosity and improves diffusion coefficients. Lipase-catalyzed ethanolysis in SC-CO2 has been shown as suitable method to obtain less oxidized n-3 PUFA FAEE compared to other reaction media. Correlation of the kinetic data to a semi-empirical model showed that the rate-limiting step is the breakdown of triacylglycerides. Similar trends of kinetic and equilibrium parameters have been observed as those reported by chemical-catalysis. Acknowledgements To the Spanish Government (MINECO) and the European Regional Development Fund (ERDF) for financial support of the project CTQ2012-39131-C02-01. Financial support from the Junta de Castilla y León and European Regional Development Fund (ERDF) through project BU055U16 is also gratefully acknowledged. RM acknowledges MINECO for a grant (BES-2013-063937). AGS acknowledges University of Burgos for a fellowship. EP acknowledges MINECO for a fellowship (FJCI-2014-19850). References [1] C. H. S. Ruxton, S. C. Reed, M. J. A. Simpson and K. J. Millington, Journal of Human Nutrition and Dietetics, 2004, 17, 449–459. [2] N. Rubio-Rodríguez, S. Beltrán, I. Jaime, S. M. de Diego, M. T. Sanz and J. R. Carballido, Innovative Food Science and Emerging Technologies, 2010, 11, 1–12. [3] B. B. Albert, D. Cameron-Smith, P. L. Hofman and W. S. Cutfield, BioMed Research International, 2013, 2013, 464921. [4] Ž. Knez, Journal of Supercritical Fluids, 2009, 47, 357–372. [5] H. Gunnlaugsdottir and B. Sivik, Journal of American Oil Chemists’ Society, 1995, 72, 399–405. [6] H. Gunnlaugsdottir, J. Mattias and B. Sivik, Journal of Supercritical Fluids, 1998, 12, 85–93. [7] J. V. Oliveira and D. Oliveira, Industrial and Engineering Chemistry Research, 2000, 39, 4450–4454. [8] D. Oliveira and J. V. Oliveira, Journal of Supercritical Fluids, 2001, 19, 141–148. [9] S.-K. Shin, J.-E. Sim, H. Kishimura, B.-S. Chun, Journal of Industrial and Engineering Chemistry, 2012, 18, 546-550. [10] H.-Y. Lee, Tanbirul Haque A.S.M., S.-B. Kim, Y.-B. lee, B.-S. Chun, Journal of Industrial and Engineering Chemistry, 2014, 20, 1097-1102. [11] M. Kondo, K. Rezaei, F. Temelli and M. Goto, Industrial and Engineering Chemical Research, 2002, 41, 5770–5774. [12] S. V Kamat, E. J. Beckman and A. J. Russell, Critical Reviews in Biotechnology, 1995, 15, 41–71. [13] R. Loss, L. Lerin, J. V. Oliveira and D. Oliveira, eds. M. A. Z. Coelho and B. D. Ribeiro, White Biotechnology for Sustainable Chemistry, The Royal Society of Chemistry, 2015, 104–135. [14] R. Melgosa, M. T. Sanz, Á. G. Solaesa, S. L. Bucio and S. Beltrán, Journal of Supercritical Fluids, 2015, 97, 51–62. [15] H. Nakaya, K. Nakamura and O. Miyawaki, Journal of the American Oil Chemists’ Society, 2002, 79, 23–27. [16] C. G. Laudani, M. Habulin, Ž. Knez, G. Della Porta and E. Reverchon, Journal of Supercritical Fluids, 2007, 41, 92–101. [17] S. L. Bucio, A. G. Solaesa, M. T. Sanz, S. Beltrán and R. Melgosa, Journal of Chemical and Engineering Data, 2013, 58, 3118–3124. [18] A.G. Solaesa, S.L. Bucio, M. T. Sanz, S. Beltrán, and S. Rebolleda, Journal of Oleo Science, 2014, 63, 449–460. [19] H. Sovová, M. Zarevúcka, P. Bernášek and M. Stamenić, Chemical Engineering Research and Design, 2008, 86, 673–681. [20] American Oil Chemist’ Society, AOCS Official Method Cd 3d-63, 2009. [21] AOAC International, AOAC Official Method 965.33, 2002. [22] American Oil Chemist’ Society, AOCS Official Method Cd 18-90, 2011. [23] A. Millqvist Fureby, C. Virto, P. Adlercreutz and B. Mattiasson, Biocatalysis and Biotransformation, 1996, 14, 89–111. [24] X. Xu, A. R. H. Skands, C. E. Høy, H. Mu, S. Balchen and J. Adler-Nissen, Journal of the American Oil Chemists’ Society, 1998, 75, 1179–1186. [25] O. N. Ciftci and F. Temelli, Journal of Supercritical Fluids, 2011, 58, 79–87. [26] W. Piyatheerawong, Y. Iwasaki, X. Xu and T. Yamane, Journal of Molecular Catalysis B: Enzymatic, 2004, 28, 19–24. [27] V. Kumari, S. Shah and M. N. Gupta, Energy & Fuels, 2007, 21, 368–372. [28] J. Calero, C. Verdugo, D. Luna, E. D. Sancho, C. Luna, A. Posadillo, F. M. Bautista and A. A Romero, New Biotechnology, 2014, 31, 596–601. [29] D. M. Chesterfield, P. L. Rogers, E. O. Al-Zaini and A. A. Adesina, Chemical Engineering Journal, 2012, 207-208, 701–710. [30] A. Sivasamy, K. Y. Cheah, P. Fornasiero, F. Kemausuor, S. Zinoviev and S. Miertus, ChemSusChem, 2009, 2, 278–300. [31] B. Cheirsilp, A. H-Kittikun and S. Limkatanyu, Biochemical Engineering Journal, 2008, 42, 261–269. [32] J. S. Lim, Y. Y. Lee and H. S. Chun, Journal of Supercritical Fluids, 1994, 7, 219–230. [33] C. Borch-Jensen and J. Mollerup, Fluid Phase Equilibria, 1997, 138, 179–211. [34] H. He, S. Sun, T. Wang and S. Zhu, Journal of the American Oil Chemists’ Society, 2007, 84, 399–404. [35] S. L. Bucio, Á. G. Solaesa, M. T. Sanz, R. Melgosa, S. Beltrán and H. Sovová, Journal of Oleo Science, 2015, 64, 431–441. [36] D. Oliveira, A. C. Feihrmann, A. F. Rubira, M. H. Kunita, C. Dariva and J. V. Oliveira, Journal of Supercritical Fluids, 2006, 38, 373–382. [37] R. C. Rodrigues and R. Fernandez-Lafuente, Journal of Molecular Catalysis B: Enzymatic, 2010, 66, 15–32. [38] U.S. Department of Health and Human Services, Food and Drug Administration. GRAS Notice 000200, 2006. [39] S.-B. Park, Y. Endo, K. Maruyama and K. Fujimoto, Food Science and Technology Research, 2000, 6, 192–195.