ent and lecithin in microencapsulation of fi sh oil by spray-drying

Departamento de Ciencia de los Alimen Ciencias Qúımicas y Farmacéuticas, Univ Independencia, Santiago, Chile Instituto de Ciencia y Tecnoloǵıa de Aliment Novais 10, 28040 Madrid, Spain Fundación Fraunhofer Chile Research, Av. Las Condes, Santiago, Chile Departamento de Ciencia y Tecnoloǵıa d Universidad de Santiago de Chile, Av. Ecu Chile. E-mail: bego.gimenez@usach.cl Cite this: RSC Adv., 2018, 8, 4172


Introduction
Fish oil (FO) is one of the main sources of long chain polyunsaturated omega-3 fatty acids (LCu3-PUFA). These fatty acids, especially eicosapentaenoic acid (EPA, C20 : 5 u3) and docosahexaenoic acid (DHA, C22 : 6 u3) are necessary in human nutrition because of their health benets. EPA is associated with cardiovascular protection, reducing blood cholesterol and triglyceride levels, 1 whereas DHA is related to the development and function of the nervous and visual system of humans and other mammals. 2 The intake recommendations from the World Health and North Atlantic Treaty Organizations are 0.3-0.5 g day of EPA + DHA. However, the worldwide average consumption is currently below these levels. 2 EPA and DHA can be obtained from plant sources rich in alpha-linolenic acid (ALA, C18 : 3 u3), but the conversion of ALA into EPA and DHA is low. 3 For this reason, healthy or functional foods supplemented with LCu3-PUFA are increasing in the food market. The main drawbacks of the incorporation of FO into foods are its low solubility in most food systems and its high susceptibility to oxidation. [4][5][6][7] Microencapsulation of FO with a coating material has been proposed as a strategy to retard lipid auto-oxidation, thus improving oil stability, prolonging its shelf life and limiting off-avours development. Furthermore, microencapsulation of FO allows its application in different food matrices, as well as controlled release of FO. 2,3,8 Conventional spray-drying is currently the most common method used to encapsulate FO. 8 However, the encapsulation of FO requires a previous elaboration of oil-in-water emulsions as infeed that must be stable over a certain period prior to and during spray-drying. 7,[9][10][11] An alternative to conventional spray-drying is the use of solvents that would allow evaluating the effect of incorporating emulsiers in the infeed system on the powder properties. To the best of our knowledge, this is the rst study in the literature focused on this topic. Both the encapsulating agent and the FO must be at least dispersible in the solvent used to prepare the infeed system. Ethanol may be used as solvent in the microencapsulation of FO by spray-drying, since its use is permitted in food industry. In this sense, ethanol-water mixtures have been used as solvent in the microencapsulation of cromolyn sodium with lactose as encapsulating agent, in order to obtain powers with different crystalline degrees and to study its effect on the solid-state powder properties. 12 Hydroxypropylcellulose, a partially-substituted poly(hydroxypropyl)ether of cellulose, was selected as encapsulating agent in this study due to its solubility in water and organic solvents.
The objective of this work was to study the effect of the emulsier (soy lecithin) and solvent (water and ethanol) in infeed systems for encapsulation of FO oil by spray-drying. FO microparticles were evaluated for encapsulation efficiency and oxidative stability of sh oil.  13 FAMEs were analyzed using a gas chromatograph (7890B, Agilent Technologies, USA), with a HP-88 fused-silica capillary column (100 m, 0.25 mm i.d., 0.20 mm lm thickness) and a ame ionization detector. The injector and detector temperatures were 250 C. The oven temperature was set at 180 C for 32 min, and increased to 230 C at 20 C min À1 , hold 32 min. The injection volume was 1 mL. EPA and DHA were quantied using tricosaenoic acid (C23 : 0) as internal standard, according to AOCS. 14 The results were expressed as mg EPA or DHA/g FO (1). The determination was performed in triplicate.

Material and methods
where A x is the peak area of EPA or DHA; A IS is the peak area of the internal standard; C f is the correction factor (0.99 for EPA, 0.97 for DHA); W IS is the weight of the internal standard added to the sample (mg); W S is the weight of the sample (mg) and 1.04 is the factor to express the results as mg EPA or DHA/g oil. and the inlet air temperature (75-135 C) were evaluated as independent variables, whereas the FO encapsulation efficiency (EE) was the dependent variable. The data were tted to a second-order regression model, according to eqn (2).
where Y was the response; subscripts i and j ranged from 1 to the number of variables (n ¼ 2); b o was the intercept term; b i , b ii and b ij values were linear, quadratic and cross-product coefficients, respectively; and X i and X j were the levels of independent variables. The analysis of variance (ANOVA), test of lack of t, and determination of regression coefficients were performed with the soware Statgraphics (5.0 program, Manugistics Inc., Rockville, MA). Response surface methodology (RSM) was applied to maximize the response variable.
The FO-EtOH-Lec infeed solution (100 g) was prepared as follows: lecithin (0.6 g) was dispersed in ethanol (12.5 g) at 40 C, using a magnetic stirrer (500 rpm for 20 min). Subsequently, this dispersion was added to FO (2.5 g) and homogenized (PT 2100, Polytron, Kinematica AG, Switzerland) at 13 000 rpm for 3 min. HPC (1.7-10.8 g) was dispersed in ethanol (82.7-73.6 g) and stirred in an orbital shaker by 12 h (200 rpm) (JSSI-100C, JSR, Korea). Fish oil-in-ethanol emulsion was added to the HPC solution and homogenized at 19 000 rpm for 3 min (Polytron PT 2100, Kinematica AG, Switzerland). The droplet size of the FO-EtOH-Lec infeed solution was determined using a Malvern Mastersizer 2000 particle size analyzer (Malvern Instrument Ltd., Worcestershire, UK), with a He-Ne laser (l ¼ 633 nm) and a measurement range 0.01-2000 mm. Six measurements were performed and the results were expressed as the average of the volume average diameter (D (4,3) ).
FO microparticles by ethanol spray-drying with FO incorporated as oil-in-ethanol dispersion (FO-EtOH) were prepared according to optimal conditions (inlet air temperature of 135 C and FO-HPC ratio of 1 : 3 (submitted)). Briey, ethanol (12.5 g) at 40 C were added to FO (2.5 g) and homogenized at 13 000 rpm for 3 min (Polytron PT-2100, Kinematica AG, Switzerland). HPC (7.5 g) was dispersed in ethanol (77.5 g) and stored at room temperature for 12 h with orbital shaking (200 rpm) (JSSI-100C, JSR, Korea). Fish oil-in-ethanol dispersion was added with constant stirring to the HPC solution and homogenized at 19 000 rpm for 3 min (Polytron PT 2100, Kinematica AG, Switzerland).
2.2.2.2. Preparation of FO microparticles by conventional spray-drying. In this case, FO was incorporated as oil-in-water emulsion (with lecithin as emulsier) and FO microparticles (FO-W-Lec) were prepared according to a central composite 2 2 + star design. The FO : HPC ratio (1 : 0.7-1 : 4.3) and the inlet air temperature (150-200 C) were evaluated as independent variables, and the EE of FO was the dependent variable. The data were tted to a second-order regression model, according to eqn (2). The ANOVA, test of lack of t, and determination of regression coefficients were performed with the soware Statgraphics (5.0 program, Manugistics Inc., Rockville, MA). RSM was used to maximize the response variable.
The FO-W-Lec infeed solution (100 g) was prepared as follows: lecithin (0.6 g) was dispersed in water (30 g) at 40 C using a magnetic stirrer at 500 rpm for 20 min. These dispersions were added to FO (2.5 g), and homogenized with a high speed blender (26 300 rpm for 3 min). HPC (1.7-10.8 g) was dispersed in water (65.7-56.6 g) and stirred for 12 h in an orbital shaker at 200 rpm (JSSI-100C, JSR, Korea). Fish oil-in-water emulsion was added to the HPC solution and homogenized at 19 000 rpm for 3 min using a rotor-stator blender. The droplet size of the FO-W-Lec infeed solution was determined as described for FO-EtOH-Lec in 2.2.2.1.
2.2.3. Spray-drying. Conventional and ethanol infeed systems were fed into a laboratory scale spray-dryer (B-290 mini spray-dryer Büchi, co-current form, Switzerland). The operational conditions were: spray gas ow, 600 L h À1 ; gas ow rate (aspirator), 35 m 3 h À1 ; feed ow (peristaltic pump); 1 mL min À1 for both conventional and ethanol methods. Air was used as drying gas in FO-W-Lec. However, in the case of FO-EtOH-Lec and FO-EtOH systems, nitrogen was used as drying gas and the B-290 mini spray-dryer was linked to mobile unit inert loop B-295 (Büchi, Switzerland). The powders were stored at À18 C.

Total FO determination.
The coating material structure of the microparticles was completely destructed by the following procedure: microparticle powders (0.6-1 g) were dissolved in ethanol (20 mL). For oil extraction, a washing was performed with distilled water (5 mL) and hexane (5 mL), stirred for 1 min at room temperature, sonicated for 5 min and centrifuged at 4000 rpm for 4 min. The supernatant (organic phase) was separated and the washing was repeated three times. The supernatants were joined and the solvent was removed using rotatory evaporator at 40 C until constant weight.
Encapsulation efficiency (EE) and yield (Y) of sh oil were calculated using eqn (3) and (4). Water activity (a w ) was determined by the dewpoint method using a Hygrolab 2 (Rotronic, USA) at 20 AE 0.3 C. Hygroscopicity was determined according to Cai and Corke. 15 The moisture content was determined with an infrared moisture analyzer (PMC50, Radwag, USA).
The particle size and size distribution were determined by light scattering using a laser diffraction particle size analyzer (LV 950-V2 Horiba, Kyoto, Japan). The microparticles were dispersed in recirculating water, and the results were expressed as D (4,3) . The outer structures of all the FO microparticle systems were examined by scanning electron microscopy (SEM). The microparticles were coated with gold/palladium, using a Varian Vacuum Evaporator PS 10E and analyzed using a LEO 1420VP SEM (LEO Electron Microscopy Ltd., UK) operated at 20 kV. The images were collected digitally (EDS 7424 soware, Oxford Instruments, UK).
2.2.6. Release of sh oil from FO microparticles. Release of FO from microparticles obtained under optimal conditions (FO-W-Lec, FO-EtOH, FO-EtOH-Lec) was evaluated in aqueous models (phosphate buffer pH 6.5 to simulate milk and acetate buffer pH 4.6 to simulate yogurt). FO-microparticles (5 g) in cellulose lter bags were placed into glass bottles (250 mL) with buffer pH 4.6 or pH 6.5 (100 mL), stoppered and stored at 30 AE 2 C in an orbital shaker at 55 rpm (JSSI-100C, JSR, Korea) in the dark. Aliquots (15 mL) were removed (triplicate) at specic time intervals, and the initial volume was maintained by the addition of the corresponding buffer. FO was extracted three times with hexane by liquid-liquid extraction. The supernatants were joined and the solvent was removed using a rotatory evaporator at 40 C until constant weight.
The data were tted to Higuchi Kinetic model, 16 according to eqn (5): M t is dened as the quantity of FO released at any time t, and M N is the initial FO loading. The release rate constants (k) were obtained from the slope of a plot M t /M N versus (time) 1/2 . 2.2.7. Storage stability assays of FO microparticles. Open Petri dishes (10 cm) with 2.5 g of FO microparticles obtained under optimal conditions (FO-W-Lec, FO-EtOH-Lec, FO-EtOH) were stored at 40 AE 1 C in a forced-air oven (BE 500, Memmert, Germany) in the dark. Petri dishes (in triplicate) were removed at specic time intervals for analyses. Additionally, samples were kept refrigerated in the dark at 5 AE 1 C and analyzed aer 5 months.
2.2.7.1. Extraction of oil from microparticles during storage. Surface oil was extracted following the procedure detailed in 2.2.4. For extraction of the encapsulated oil from the microparticles devoid of surface oil, 2 g-samples were rubbed in a mortar with 0.5 mL water to obtain a clotted mass. Then, 2.5 g Na 2 SO 4 were added and the mixture was rubbed again until a granular mass was obtained. The oil was extracted with diethyl ether (3 Â 20 mL). Aer ltration through anhydrous Na 2 SO 4, the solvent was removed using a rotatory evaporator at 40 C and the sample dried until constant weight under stream of nitrogen. 17 2.2.7.2. Determination of polar compounds. Formation and distribution of polar compounds in both surface FO and encapsulated FO was carried out by solid phase extraction followed by high performance size exclusion chromatography analysis, described in detail, including precision, accuracy and recovery data. 18 2.2.7.2.1. Solid phase extraction (SPE). The polar fraction of both surface and encapsulated FO in each system was isolated by SPE. For this, extracted oil (50 mg) was dispersed in hexane (2 mL) and separated into two fractions by Sep-Pack Vac Silica columns (6 cm 3 /1 g), previously conditioned with petroleum ether : diethyl ether (15 mL; 90 : 10, v/v). The rst fraction (unoxidized triacylglycerols) was eluted with petroleum ether-: diethyl ether (15 mL; 90 : 10, v/v). The second fraction (containing triacylglycerol polymers, triacylglycerol dimers, oxidized triacylglycerol monomers, diacylglycerols, free fatty acids and polar unsaponiable matter) was eluted with diethyl ether (25 mL). For quantitative purpose, monostearin (1 mg) was used as internal standard. The separation efficiency was checked by thin layer chromatography, using a silica plate and hexane : diethyl ether : acetic acid (60 : 40 : 2) as elution solvent, and revealed with iodine vapour.
2.2.8. Statistics. All the results were reported as mean AE standard deviation (SD). One-way ANOVA was applied to determine the statistical differences among systems, using Statgraphics, Version 7.0 (Manugistic Inc., Statistical Graphics Corporation, Rockville, MD).

Fish oil microparticles
A central composite design was applied to evaluate the effect of the FO/HPC ratio (X 1 , formulation variable) and the inlet gas temperature (X 2 , process variable) on the encapsulation of FO by ethanol (FO-EtOH-Lec) and conventional spray-drying (FO-W-Lec), respectively. Table 1 shows the experimental design for both spray-drying systems. The EE of FO was higher in FO-EtOH-Lec, with values ranging from 41% to 85%, whereas the FO-W-Lec systems showed values of EE between 17.1% and 72.3%. The FO recovery was high (close 100%) for both spraydrying systems, and thereby this variable was not considered as response variable.
The ANOVA for the EE of FO-microparticle systems by ethanol and conventional spray-drying is shown in Table 2. For FO-EtOH-Lec, the EE of FO was signicantly (p # 0.05) affected by the linear and quadratic forms of FO/HPC ratio and the linear form of the inlet gas temperature; whereas in FO-W-Lec, the EE of FO was signicantly affected by the linear and quadratic forms of FO/HPC ratio and the linear and quadratic forms of the inlet gas temperature. The models explained over 92% of the variability (R 2 adj. for d.f.) in EE for both FO-EtOH-Lec and FO-W-Lec systems (Table 2), with residual values below 6.0. The equations describing the effect of the independent variables on the EE of FO are the following: EE(FO-EtOH-Lec) ¼ 11.0527 + 28.9611 Â X 1 + 0.113565 Â X 2 À 3.70529 Â X 1 2  The surface response plots for both FO microparticles systems (Fig. 1) showed a maximum EE included in the experimental domain. Furthermore, the variation of the FO/HPC ratio had higher inuence on the EE than that of the inlet gas temperature, and the highest values were found at FO/HPC ratios with higher HPC content. Thus, the optimal FO/HPC ratio was 1 : 4 for both FO-EtOH-Lec and FO-W-Lec systems, showing the same content of total solids in both cases. As shown in Fig. 1, the higher the encapsulating agent content, the higher EE of FO because the faster the formation of the dry crust over the oil droplets surface, leading to higher emulsion retention. 8 FO/encapsulating agent ratios from 1 : 1 to 1 : 6 have been reported in microencapsulation of FO by conventional spray-drying. 19 However, to the best of our knowledge, this is the rst study dealing with the microencapsulation of sh oil-inethanol emulsions by solvent spray-drying.
The inlet gas temperature showed a low effect on the EE of FO in the range of studied values. However, the extent of lipid oxidation can be inuenced by the inlet gas temperature, and higher lipid oxidation would be expected with higher inlet gas temperatures. In the case of FO-W-Lec system, the optimal inlet gas temperature was 200 C, in the range reported in the literature for conventional spray-drying (140-215 C). 19 However, the optimal inlet gas temperature was lower for FO-EtOH-Lec and FO-EtOH systems (135 C), attributed to the lower boiling point of ethanol. Table 3 shows the characteristics of FO microparticles obtained under optimal conditions (FO-EtOH-Lec, FO-EtOH and FO-W-Lec). The presence of an oil-in-ethanol emulsion stabilized by lecithin in the infeed system (FO-EtOH-Lec) did not have any signicant effect on EE, in relation to FO-EtOH system, showing similar (p > 0.05) EE of FO (81.1% and 80.4%, respectively), in spite of the encapsulation mechanism could be different. When a sh oil emulsion is encapsulated by spray-drying the encapsulation efficiency represents the sh oil emulsion retention within the microparticles, whereas in the case of absence of lecithin in the infeed system, the encapsulation mechanism involves the direct interaction between HPC and triglycerides of FO mainly by hydrophobic forces and van der Waals interactions. Although the FO-W-Lec (control) and FO-EtOH-Lec systems were prepared with different solvents (water and ethanol, respectively), the EE was higher in FO-EtOH-Lec (Table  3), in spite of the encapsulation mechanism is based on the emulsion retention in both cases. This behavior may be explained by the lower droplet size of the infeed (FO-emulsion with HPC) in the case of FO-EtOH-Lec (D (4,3) ¼ 0.48 mm), when compared to FO-W-Lec (D (4,3) ¼ 0.97 mm), attributed to lower dielectric constant of ethanol (25.8 vs. 81.1 for water) and/ or the lower interfacial tension at oil-EtOH interface (2.4 mN m À1 reported in a 95% ethanol-oil system) 20 compared to that of oil-water systems (19.0-24.5 mN m À1 ). 21 The total FO content in the microparticles containing lecithin (FO-EtOH-Lec and FO-W-Lec) and without lecithin (FO-EtOH) was similar (p > 0.05), around 200 mg g À1 (Table 3). The total FO content for conventional spray-dried microparticles is variable, and both higher and lower values than those obtained in this study have been reported (108-119 mg g À1 ; 22 244-492 mg g À1 (ref. 6)). Furthermore, similar behavior to total FO content was obtained for the content of EPA and DHA, with similar values in all the systems (p > 0.05; Table 3). In this study, one gram of FO-microparticles provides on average 39 mg of EPA + DHA, which is approximately 7.8-13% of the recommended daily dose (0.3-0.5 g of EPA + DHA per day). All the  systems (FO-W-Lec, FO-EtOH, FO-EtOH-Lec) showed similar moisture content (p > 0.05) with values ranging from 4.5 to 5.0%, whereas a w values ranged from 0.25 to 0.33, ensuring the microbiological stability of the microparticles. The hygroscopicity was also similar (p > 0.05) in all the systems studied (Table  3), because HPC was used as encapsulating agent in all of them. Fig. 2(A.1-C.1), shows the SEM photographs and particle size distributions of FO-microparticle powders (FO-EtOH-Lec, FO-EtOH and FO-W-Lec) obtained under optimal conditions. All the microparticles showed spherical shapes with smooth surface, although a higher proportion of dented microparticles was found in FO-W-Lec system, which may be attributed to particle shrinkage due to the high inlet gas temperature (200 C) during the spray-drying process. A unimodal particle size distribution was found in the FOmicroparticle systems with lecithin, with sizes ranging from 0.5 mm to 6 mm and from 0.5 mm to approximately 10 mm in FO-EtOH-Lec and FO-W-Lec, respectively ( Fig. 2A.2 and 2C.2). Particles with higher sizes in FO-W-Lec system also led to higher D (4,3) values (2.43 mm vs. 1.80 mm obtained for FO-EtOH-Lec; Table 3). In the case of the FO-EtOH system, a bimodal particle size distribution was found, with a dominant population ranging from 0.05 to 0.8 mm and a small population between 0.8 and 6 mm (Fig. 2B.2). Furthermore, D (4,3) value was lower than in the FO-microparticle systems with lecithin (0.54 mm; Table 3). This particle size distribution can be also observed in the SEM photographs ( Fig. 2B.1). Therefore, smaller FO microparticles were obtained when the FO was incorporated as oil-in-ethanol dispersion (without emulsier).

Release of sh oil from FO microparticles
The FO release prole in aqueous model (pH 4.6 and 6.5) showed one phase (Fig. 3), where the release percentage was lower than 16 g FO/100 g power for all FO-systems during the studied period at 30 C. This FO released corresponded only to surface FO and release of encapsulated FO was not observed, in spite of the encapsulation mechanism could be different due to presence of lecithin and in agreement with the low affinity of the hydrophobic triglycerides for aqueous medium. The data  were tted to Higuchi mathematical model, with correlation coefficient over 0.97 in all the systems, showing Fickian diffusion because only surface FO was released. The FO-W-Lec (0.030 AE 0.003 days 0.5 and 0.029 AE 0.001 days 0.5 at pH 4.6 and 6.5, respectively) and FO-EtOH (0.020 AE 0.002 days 0.5 and 0.027 AE 0.012 days 0.5 at pH 4.6 and 6.5, respectively) systems showed the highest release rate constants in both aqueous media, followed by FO-EtOH-Lec (0.013 AE 0.003 days 0.5 and 0.012 AE 0.001 days 0.5 at pH 4.6 and 6.5, respectively).
3.4. Stability of microencapsulated sh-oil during storage Table 4 shows the results obtained for polar compounds, including total level and distribution in different groups of oxidation compounds and hydrolysis products, in surface and encapsulated fractions of the oils extracted from initial microparticles. This analytical approach has been successfully applied to quantitate oxidation compounds in microencapsulated oils and proved to be more complete and sensitive than determinations traditionally used, such as TBARS or peroxide value. 17,23,24 Results showed that, even though values obtained among surface and encapsulated oil fractions and among microparticles systems were very close, it was clear that preparation of microparticles with ethanol using lecithin led to lower levels of oxidation compounds, especially regarding dimers and polymers. Fig. 4 shows time-course of formation of total oxidation compounds (oxidized triacylglycerol monomers plus triacylglycerol dimers and polymers) in surface and encapsulated oil fractions during oxidation at 40 C in the dark. Oxidized triacylglycerol monomers include monomeric triacylglycerols with one or more oxygenated functions, mostly hydroperoxides during the early oxidation stage or epoxides, alcohols and ketones during advanced oxidation stages. Dimers and polymers start forming once oxidation is accelerated. Therefore, global quantitation of oxidized triacylglycerol monomers, dimers and polymers provides a complete measurement of the oxidation level throughout the entire oxidation process. 24 Results showed remarkably lower oxidation for FO-EtOH-Lec microparticles both in surface and encapsulated oil fractions throughout oxidation. Aer 72 hours, values higher than 25% of oxidation compounds had been reached in FO-W-Lec and FO-EtOH microparticles, in marked contrast with only 12-13% in the case of FO-EtOH-Lec samples. Evaluation of the levels of EPA and DHA remaining aer storage for 72 hours at 40 C in the dark, revealed that FO-EtOH-Lec microparticles contained 32 mg g À1 microparticle versus 27.7 mg g À1 and only 19 mg g À1 in FO-EtOH and FO-W-Lec microparticles, respectively. At lower temperature, 5 C (Fig. 5), results showed consistently the highest stability for FO-EtOH-Lec microparticles. In this case, polymers plus dimers in total microencapsulated oil were determined aer 5 months-storage. As it is well-known, polymerization compounds increase rapidly in FO even at low  temperature due to their considerable content in polyunsaturated fatty acids, which are highly susceptible to oxidation. 17,25 The results obtained in the assays carried out on oxidative stability seem not to be associated to differences in particle size distribution or EE between microparticles.

Conclusions
For FO-microparticles with lecithin, the replacement of water by ethanol contributed to the signicantly highest EE and stability of FO. The lower FO-oxidation for FO-EtOH-Lec may be related to the substantially lower inlet gas temperature, the use of nitrogen as drying gas and the physical barrier that lecithin forms around the oil droplets to the oxygen ow.

Conflicts of interest
There are not conicts to declare.