12EE Fish Oil Using AgNO3(aq

Article pubs.acs.org/IECR

Extraction of EPA/DHA from 18/12EE Fish Oil Using AgNO3(aq): Composition, Yield, and Effects of Solvent Addition on Interfacial Tension and Flow Pattern in Mini-Fluidic Systems Kirubanandan Shanmugam and Adam A. Donaldson* Department of Process Engineering and Applied Science, Dalhousie University, PO Box 15000 Halifax, Nova Scotia, Canada B3H 4R2

Downloaded by UNIV OF NEBRASKA-LINCOLN on September 6, 2015 | http://pubs.acs.org Publication Date (Web): August 12, 2015 | doi: 10.1021/acs.iecr.5b01780

S Supporting Information *

ABSTRACT: Solvent extraction of Omega-3 polyunsaturated fatty acids (PUFA) ethyl esters (Et) from 18/12EE fish oil was performed with an aqueous silver nitrate solution in a mini-fluidic reactor framework. The resulting extraction was compared within conventional stirred-tank reactors, with practical extraction yields of EPA-Et and DHA-Et approaching 60 to 70 wt %. Equilibrium was reached in less than 36 s at 10 °C, despite stratified flow being observed rather than previously reported slugflow profiles in idealized fluid pairs at these scales. The deviation in flow pattern was attributed to a measured order of magnitude reduction in the interfacial tension between fish oils and AgNO3 solution relative to idealized solvent mixtures using hexane/ heptane as a carrier for purified EPA/DHA. The impact of solvent addition to fish oils on interfacial tension with silver nitrate solutions was subsequently explored through spinning drop tensiometry, suggesting that the heterogeneous nature of raw fish oils will yield significantly different flow patterns than previously considered in extraction studies utilizing AgNO3 solutions. These results are applicable both to mini-fluidic systems and for the approximation of interfacial tension in fish oil/organic/aqueous extraction systems.

1. INTRODUCTION Omega-3 polyunsaturated fatty acids (PUFA) are increasingly recognized as nutraceutical and food additives, with increasing production requirements of high-purity products with EPA/ DHA contents exceeding 55%. Solvent extraction has been widely used for the separation and concentration of various biomolecules from their feed stocks, 1−3 including the purification of fatty acid ethyl esters based on the complexation between double bonds of PUFA-Et and silver ions.4,5 This separation method is achieved through a reversible chemical reaction between the silver nitrate ions and the 5 to 6 unsaturation sites of Omega-3/6 PUFA and is viewed as a promising method for the purification and concentration of PUFA from fish oil ethyl esters. Despite preliminary success with this system for idealized EPA-laden solvents (primarily pharmaceutical grade EPA/DPA dissolved in hexane or heptane) as shown by the work of Belarbi et al.,6 there is limited data available on the practical extraction performance of Omega-3 from 18/12EE fish oils using Ag-based solvents. Given that this is a liquid−liquid extraction and that fish oils are complex mixtures of fatty acids with various chain lengths and degrees of unsaturation, the heterogeneous nature of the fish oils is anticipated to significantly affect overall yield and fluid dynamics within these systems. The complexation reaction between silver and unsaturated fatty acids is still not well characterized, specifically with respect to the effects of molar ratios of Ag ions to EPA/DHA molecules on extraction, yield, and selectivity. Traynham et al.7 reported that silver ions could form reversible π bond complexations with double bonds, with subsequent literature exploring its use to remove olefins from paraffinic systems. In these reactions, the binding of silver ions to the double bond of © 2015 American Chemical Society

Omega-3 PUFA is considered a mass transfer limited interfacial reaction where mixing and interfacial area play major roles in the reaction. Temperature has also been found to significantly affect solvent extraction with silver nitrate solutions. In the work of Teramoto et al., 8 the distribution ratios of polyunsaturated fatty acid ethyl esters between aqueous AgNO3 extraction phase and an organic phase of esters dissolved in heptane were found to increase drastically as temperature was lowered. There was, however, an accompanying increase in viscosity at lower temperatures, creating additional mass transfer concerns and pumping limitations. The obvious drawback of these methods is also the need for large amounts of expensive silver salts, for both high recovery and freezing point depression when an aqueous carrier phase is used at low temperatures. Over the past decade, micro/mini-fluidic technology has emerged as a process intensification technology suitable for solvent extraction systems. The flow patterns generated in micro/mini-fluidic technology have been proven to offer high mass transfer rates and consistent extraction performance, while minimizing solvent inventory requirements compared to other conventional extraction systems.9−11 The high cost of AgNO3, the reversible exothermic silver−PUFA complexation reaction, and the desire to operate at lower temperatures all suggest that the compact framework of the mini-fluidic system will offer efficiency gains within a processing facility using this technology. This was, in fact, observed recently by Kamio et Received: Revised: Accepted: Published: 8295

May 13, 2015 August 6, 2015 August 6, 2015 August 6, 2015 DOI: 10.1021/acs.iecr.5b01780 Ind. Eng. Chem. Res. 2015, 54, 8295−8301

Article

Industrial & Engineering Chemistry Research Table 1. Comparison of Various Processes on Concentration of Omega-3 PUFA, Adapted from Lembke1 molecular distillation

supercritical fluid chromatography

liquid chromatography

mechanism

boiling point

molecular weight and structure

conditions

140−220 °C, 0.001 mbar 65−75% very high

35−50 °C, >140 bar 99% very high

molecular weight and structure 20−50 °C, 1 bar

low continuous moderate

very low semi-batch high

efficiency decontamination efficacy oxidation operation mode capital investment

supercritical extraction

urea precipitation

low temperature crystallization

solvent extraction

molecular weight

degree of saturation

melting point

complexation with CC

−10−90 °C, 1 bar 45−65% low

0−70 °C, 1 bar

20−30 °C, 1 bar

99% high

35−50 °C, >140 bar 75−85% medium

>90% low

88−91%a low

possible semi-batch high

very low continuous high

possible batch low

possible batch moderate

very low continuous low

a


The extraction efficiency of the solvent extraction based on aqueous phase silver nitrate solution is 88−91% from concentrate. However, the separation performance is >94% in silver salts in ionic liquids.

Figure 1. Overview of the process employed within this study for contacting and subsequent sample extraction/preparation.

exploration of surface tension and flow-pattern analysis to clarify the hydrodynamic impact of using raw fish oils in the place of idealized EPA/solvent mixtures.

al.,4,5 who described liquid−liquid extractions of DHA from an organic carrier (heptane) using silver salt solutions in a micro reactor framework (0.5 mm length scale). In their work, contact times on the order of 10 to 20 s were sufficient to reach equilibrium at the conditions tested, leading to the current interest in this approach. In order to consider this system within a production framework, practical data using real fish oils and information on scalability was needed. Using a 1.59 mm mini-fluidic reactor and a batch reactor system to collect preliminary data on flow patterns, extraction yields, and compositions, this work attempts to compare extraction from practical fish oils against results from idealized solvent systems. Where previous literature in this field has consistently reported slug-flow profiles, this work observed a stratified flow regime within the channels, prompting an

2. METHODS AND MATERIALS FOR CONCENTRATION OF PUFA FROM FISH OILS A recent review of concentration processes commonly encountered in the literature was published by Lembke,1 from which a summarized overview and comparison of relative parameters is provided in an effort to illustrate their relative advantages and disadvantages (Table 1). The main processes reviewed include urea precipitation, low temperature crystallization, molecular distillation, supercritical extraction, chromatographic methods, enzymatic enrichment, and solvent extraction. 8296

DOI: 10.1021/acs.iecr.5b01780 Ind. Eng. Chem. Res. 2015, 54, 8295−8301

Article


Industrial & Engineering Chemistry Research

1.47 mL/min, thus maintaining an approximate salt to oil solution flow ratio of ∼3.4:1.4,5 The fish oil ethyl ester and silver nitrate solution are precooled in a 1.5 m length of tubing prior to being contacted together in a “Y” junction, after which the immiscible fluids were allowed to be in contact for residence times ranging from 0.5 to 6 min before being collected in a syringe and manually processed to recover the PUFA fraction by decomplexation and dewatering. The extraction was also carried out in a batch reactor for comparison purposes. A 65 mL three neck flask equipped with an exterior cooling jacket, stirrer, and thermometer was initially charged with 50 mL of silver nitrate solution, blanketed with nitrogen, and allowed to cool to 10 °C. Once cooled, 14.7 mL of 18/12EE fish oil was added and the fluids allowed contact under stirred conditions for 15, 30, 60, 90, and 120 min. Agitation was removed at the end of reaction period, allowing for density-based separation of the oil layer. The final aqueous layer was then subjected to the same decomplexation and dewatering procedure as used for the mini-fluidic system. After extraction, the aqueous phase enriched with PUFA was processed to determine yield and purity. The processing of the extraction mixture consisted of 4 steps (Figure 1): gravity separation of oil phase, demulsification with hexane, decomplexation using hexane and water, and sample drying and filtering prior to analysis at DSM Nutritional Products. In the stirred tank vessel and the sample syringes, a small residual oil phase remained following the initial reaction and gravity partitioning. In the case of the batch system, the entire contents of the batch were transferred to a separatory funnel, where the aqueous/emulsion phase was removed and the residual oil phase collected through the addition of hexane to assist in minimizing residue losses on the glass surface. For the syringe samples from the mini-fluidic process, the syringe was held vertically and the oil phase discharged into a second syringe using a short interconnecting tube. The residual oil phases were then stored in a nitrogen blanketed sample vial for subsequent drying and filtration. An aqueous emulsion phase between fish oil ethyl ester and silver nitrate solution was then demulsified through the addition of 10% by volume of hexane. In the batch system, hexane was added to a separatory funnel containing the previously collected aqueous phase, resulting in the removal of the emulsified organic fraction. For the syringe samples, hexane was transferred directly to the sample syringe containing the emulsion/aqueous phase, with contact achieved by repeated transfer between two syringes through a short connecting tube length. In both cases, the hexane breaks the emulsion and separates into an oil phase dissolved in hexane fraction and an aqueous phase. The resultant organic component is subsequently referred to as Fraction 1, consisting of the added hexane and organic content initially entrained with the aqueous phase during gravity-based separation of the reactor product. To recover the complexed ethyl esters from the aqueous phase, the fractionated aqueous phase typically having a volume of ∼50 mL was placed in a separatory funnel where it was brought into contact with 20 mL of hexene and 600 mL of water and allowed to be in contact for 2 h under a nitrogen blanket. The aqueous phase was then separated off and the organic fraction stored for analysis. The nonpolar solvent (hexane) was added into the aqueous phase to weaken the bond between EPA/DHA and silver ion and to increase the volume of organic phase recovered. This final organic fraction is

Silver nitrate salt solutions have previously been used as a solvent for the purification of the ethyl ester of PUFA (PUFAEts) as previously reported by Yazawa,12 whereby silver ions interact with the double bonds in unsaturated fatty acids to produce complexes which are soluble in ionic/polar carrying phases. Since the separation is performed exclusively via a reversible chemical reaction, this technique is very simple and would be a promising method for the purification of PUFAs. Limited reports are available of extraction performance using silver salts to isolate polyunsaturated fatty acids from a mixture of FFA, with previously known methods not providing for a sufficiently selective and/or efficient process for concentrating Omega-3 fatty acids. WO Patent WO 2012038833 A113 reported that silver salt solutions could be used in the extraction of PUFA from marine oils. Within a process-oriented framework, an industrial extraction system might involve multiple contacting, separation, solvent recovery, and solvent regeneration steps. This process was subsequently adapted to lab-scale analysis, with the reaction/contacting taking place in either a mini-fluidic or batch system and the subsequent washing and separations occurring in separatory funnels and nitrogen bubblers (Figure 1). Prior to each experiment, a fresh batch of the silver nitrate solution with 50 wt % AgNO3 (ACS grade, 99.9% assay) and 5 wt % NaNO3 (Fisher scientific, 99% assay) was prepared to avoid long-term oxidation of the nonstabilized silver nitrate solution. The solution was prepared by dissolving 400 g of silver nitrate into 400 mL of distilled deionized water, after which 40 g of NaNO3 was added as an ionic strength adjuster to provide a comparable composition as reported by Kamio et al.4,5 During this process, the solution volume increased to 500 mL due to the addition of silver nitrate and sodium nitrate. The silver nitrate solution was then purged with nitrogen and refrigerated in a blanketed opaque container to minimize oxidation and sunlight exposure prior to use. The 18/12EE fish oils were obtained from DSM Nutritional Products in Dartmouth, NS, Canada, derived from anchovy (Engraulis ringens) and sardine (Sardinops sagax sagax). During transfer operations and prior to storage, the fish oils were purged with nitrogen and refrigerated to limit oxidation prior to testing. As a raw material, the 18/12EE fish oil ethyl ester was not winterized prior to acquisition, creating the potential for crystal formation at lower extraction temperatures. In order to avoid potential channel blockages, the fish oils were winterized prior to use. Winterization is the process of crystallization of oils and fats followed by the separation of solids, which are often used to make high quality food oils. On heating the fish oil ethyl ester to 60 °C and cooling down to 4 °C, a gel forms within the solution and settles in the container. Subsequent analysis of nonwinterized and winterized feedstock oils used in this process by gas chromatography confirmed that the overall DHA, EPA, and total Omega-3 PUFA was not significantly modified by the winterization process. The mini-fluidic reactor consisted of a 1.59 mm ID Tygon tube submerged in an insulated reservoir maintained at 10 °C using an external refrigerated circulating bath. Prior to each experiment, the tubing was purged with water and ethanol and then dried with compressed air. The fish oil and silver nitrate solution from the reservoir was then injected using a double syringe pump (2 NE 4000), whereby a 60CC syringe was used for the silver nitrate solution and a 10CC syringe was used for the 18/12EE fish oil. The syringe pump was set to produce a silver salt solution flow rate of 5 mL/min and an oil flow rate of 8297


Article



subsequently referred to as Fraction 2, representing the extracted PUFA dissolved in hexane. The residual oil, Fraction 1, and Fraction 2 recovered from this process were subsequently filtered and subjected to a nitrogen flow to vaporize the volatile solvent. The solvent free fractions were then weighed for determination of yield, sealed, and transported to DSM Nutritional Products where they were analyzed using in-house GC methods similar to that of Stenerson et al.14 to determine EPA, DHA, and total Omega content.

3. EXTRACTION RESULTS: EPA/DHA CONTENT AND YIELD The initial composition of the 18/12-EE fish oil was determined for both nonwinterized and winterized oil to be approximately 15 wt % EPA and 10 wt % DHA, with a total Omega-3 content of ∼30.8 wt %. The EPA-Et and DHA-Et content in the fish oil did not change over the winterization process performed, confirming that crystallization of waxes within the feedstock did not artificially concentrate the 18/ 12EE fish oils relative to the process stream which was initially sampled to provide the tested oils. The extraction temperature of 10 °C was also slightly above the winterization temperature of 4 °C, avoiding any additional crystallization during the actual extraction process. The mass of each sample collected was recorded during the postextraction process, both for the purpose of performing a rough yield calculation and to determine the approximate volumes of each phase which might be expected. The sum of the oil fractions recovered was typically less than that of the initial oil fed, primarily due to oil residue losses on the glassware and syringes used in the postextraction processing. Subsequent compositional analysis of each of the sample fractions collected provides insight into the partitioning of EPA, DHA, and Omega-3 during the extraction process (Figure 3). The first organic fraction obtained was the residual oil layer which did not form a complex or emulsion with the silver nitrate solution. Given the relatively larger sample volume in this phase (Figure 2), the mass and composition was the primary basis for determining overall extraction yield. Between 85% and 95% of the EPA and DHA was extracted into the emulsion and complexed media. After collecting the residual oil layer, hexane was added to the aqueous phase to de-emulsify any suspended oils, with the resulting organic layer denoted as Fraction 1. For the mini-fluidic reactor, this suspended phase represented a minor fraction of the total sample masses collected (5 wt %) and exhibited a concentration of DHA and total Omega-3 relative to EPA when compared to that of the raw feedstock. Within the batch reactor, Fraction 1 was typically greater in mass (attributed to increased agitation by the magnetic stirrer relative to the mini-fluidic system), with an increase in Omega-3 compounds not classified as EPA or DHA and a EPA/DHA ratio similar to that of the raw feedstock. The most likely reason for this difference is that in the batch system there is significant agitation of the immiscible phases, providing significant opportunity emulsified EPA and DHA to complex with the surrounding silver ions (resulting in low EPA/DHA levels within the suspended phase), while in the mini-fluidic system the stratified flow profile observed resulted in minimal agitation and emulsified oil, so the relatively larger DHA molecule with 6 unsaturated sites (possibly requiring 6 silver ions to go into solution) had a reduced complexation rate compared to the smaller EPA molecule with only 5 unsaturated

Figure 2. Mass fraction distribution during silver-based extraction within the mini-fluidic (top) and batch (bottom) reactors.

sites. This would also explain the lower net yield of Fraction 1 for the mini-fluidic setup. Following demulsification, the silver nitrate solution was decomplexed through watering out, with hexane added to increase the volume of organics which could be recovered using a separatory funnel. The final EPA, DHA and total Omega-3 content in Fraction 2 approached 41, 26.8, and 80.8 wt %, respectively, for each of the contact times tested. While slight variations are present between the mini-fluidic and batch reactor with time, these results suggest that equilibrium extraction levels were achieved by the minimum contact times used in each system (36 s for the mini-fluidic system and 15 min in the batch reactor). While not useful for kinetics analysis, these results demonstrate that equilibrium has been reached in the mini-fluidic 1.59 mm channel on a similar time scale as that observed in a 0.5 mm channel by Kamio et al.,4,5 despite the observed difference in flow pattern. In their work, pure DHA or EPA dissolved in an organic solvent reached equilibrium with a silver salt solution in 10 to 20 s with a silver ion concentration comparable to that used here and an operating temperature of −5 to 15 °C. For the purpose of process feasibility, looking at shorter residence times was not required as most of the batch-wise processes using retrofitted stirred reactors would operate for longer than 15 min due to filling and emptying considerations, and a 36 s residence time within a continuous process provides an upper bound of what might be needed within a double-pipe or plate heat exchanger design. The approximate total yield of EPA/DHA/Total Omega-3 recovered within the de-emulsification oil layer (Fraction 2) ranged from 75% to 78% of the initial content present within 8298


Article



Figure 3. Compositional analysis of feedstock and oil layers obtained during the extraction process outlined in Figure 1.

the 18/12EE feedstock, representing ∼28.9% of the total initial 18/12EE mass fed to the contactor. Compared to total yields of 85% (25.1% of initial tuna oil mass) obtained in lab-scale urea complexation experiments carried out over a 20 h period,15 the method presented here potentially offers significant advantages in terms of kinetics for concentration of Omega-3. Of particular note is the significantly higher recoveries observed in this work relative to that of Kamio et al.,4,5 where a synthetic mixture of DHA in n-heptane exhibited DHA recoveries ranging from 35% to 65% as temperatures were decreased from 25 to −5 °C. While some of this discrepancy may be accounted for through variations in [Ag+] ion concentrations and aqueous/oil flow ratios, the heterogeneous nature of commercial 18/12EE fish oils is thought to promote extraction relative to idealized mixtures due to combined effects of natural surfactants and synergistic complexation of EPA, DHA, and other Omega-3’s. At a minimum, this work demonstrates the successful application of silver-based solvent extraction to nondiluted commercial fish-oil feedstock. Recovery estimates given here are considered conservative, representing only the actual mass of Omega-3 PUFA’s collected after sample losses during processing. For the CSTR, the amount added was well-defined due to the batch-wise operation, limiting potential errors in yield calculations to losses during handling. For the mini-fluidic system, an unanticipated stratified flow pattern within the channels may have potentially affected the relative flow rates and contact times of the two species in question, creating additional uncertainty in the yield. It is worthwhile to note that the mass of samples collected from the mini-fluidic reactor ranged from 5.5 to 7.5 g for the residual oil layer, 0.5 to 1.5 g for Fraction 1, and 2 to 4 g for Fraction 2. Solely on the basis of analysis of the residual oil layer (highest sample mass, limited opportunity for residual losses), the maximum recovery of Omega-3 should approach 90% to 95%.

4. FLOW PATTERN VARIATION AND INTERFACIAL TENSION A notable difference between this study and previous literature4,5 was the formation of a stratified flow profile in the mini-fluidic channels. Stratified flow is characterized by the formation of parallel flow paths between phases, where the relative residence time and velocity of each phase is dependent on the cross-section occupied. The observed stratified flow (Figure 4) was unexpected under these conditions, as one of

Figure 4. Stratified flow pattern formed in mini-fluidic reactor technology for 18/12EE fish oil (top) and 50 wt % AgNO3 solution (bottom) flowing at 1.47 and 5 mL/min, respectively. Interface indicated by arrows.

the characteristic properties of channels of this scale is the predominance of interfacial forces over gravity forces which should limit the formation of stable stratified flow. The dimensionless parameter often used to characterize this interaction between gravity and surface forces is the Bond number, Bo, which is typically less than 1 for microgravity systems with the transition commonly occurring between 1 and 10.

Δρgdh 2 (1) σ where Δρ is the density difference between fluids, g is the gravity constant (9.81 m/s2), dh is the hydraulic diameter of the channel, and σ is the interfacial tension (IFT). The majority of previous literature exploring this extraction method has employed 0.5 mm channels with a heptane-based carrier Bo =

8299


Article



of ∼2.6 mN/m was observed. In contrast, fish oil in silver nitrate solution was at the lower end of the instrument’s measurement capabilities, where the interface would breakup rapidly at rotation rates above 4000 rpm before stable surface tension measurements could be obtained. The presence of the silver ions and the complexation process resulted in at least a single order of magnitude reduction in IFT to ∼0.25 mN/m. The addition of nonpolar organic solvents into the fish oil ethyl ester generally increased the IFT, with hexene additions having a significantly greater impact than hexane. While 10% hexane addition had a negligible impact (possibly increasing the IFT from ∼0.1 to ∼0.3 mN/m at low interfacial renewal rates), 50% hexane addition increased the IFT to as much as 0.7 mN/ m. The addition of 10 wt % hexene had the equivalent impact on surface tension as 50 wt % addition of hexane. Increasing the hexene fraction to 50% had the equivalent effect of counteracting the complexation effects observed between the fish oil sand silver ions. In all cases where organic solvent was added, capillary instability occurred in the spinning drop tensiometry unit, limiting the maximum rotation rate which could be applied before droplet breakup or coalescence occurred. These trends are important for future processing considerations, both in recognizing that the low surface tension between fish oil and silver nitrate will facilitate contacting and mass transfer, while possibly making separation difficult. The addition of hexane or, preferably, hexene, could be used to help in emulsion breaking and separating the two phases. As the variations in IFT observed here correspond to the ranges determined for Bo at which flow could transition from gravity dominated to surface force dominated (i.e., 0.26 to 2.1 mN/m), preliminary flow pattern analysis was performed for oil flow rates above and below those tested within the extraction studies for three of the fluid pairs (Table 2). The images were

solvent having a density of approximately 684 kg/m3, compared to the measured 18/12EE fish oil density of 898 kg/m3. Both this study and that of Kamio et al.4,5 used a similar silver nitrate solution, for which the measured density was 1751 kg/m3. Accounting for the 1.59 mm diameter of the channels used here, the IFT at which Bo = 10 for a 1.59 and 0.5 mm channel would be 2.1 and 0.26 mN/m, respectively. Interfacial tension near or above these values would be needed in each system for surface forces to start affecting the flow profiles to the point of creating slug flow. Given the observed transition between slug and stratified flow, IFT measurements were carried out between solvent-diluted 18/12EE fish oil and both water and AgNO3 solutions to determine the potential variability in IFT between raw fish oils and synthetic alkane mixtures used previously. The raw 18/12EE fish oil and silver nitrate fluid pair represents a heterogeneous mixture of organics and aqueous salt solutions, with significant presence of surface-active compounds and active complexation occurring on the interface which should result in similar dynamics to a surfactant loaded system. Adsorption of extraction molecules at the interface lowers the IFT and makes it easier to disperse one phase into the other. In previous studies where EPA-doped alkane carriers were used in the place of fish oils, the use of nonpolar organic solvents could elevate the IFT, potentially leading to a change in the predominant flow pattern. It is important to understand this transition as a variation in flow pattern may affect optimal system design during scale-up (i.e., maximum stable bubble size, mass transfer, etc.) and could lead to early onset of emulsion formation which could make separation difficult if not anticipated. Figure 5 illustrates the IFT determined via spinning drop tensiometry16 for various mixtures of 18/12EE, organic

Table 2. Flow Patterns Observed in a 1.59 mm Tygon Minichannela

a

The lighter oil phase enters from the bottom while the heavier silver nitrate enters from the top.

collected using a pco.dimax HD high-speed camera with LED backlight with all inlet fluids cooled to the same temperature as used in the extraction experiments (i.e., 10 °C). These experiments were performed with the inlets oriented to maximize initial instability, with the more dense aqueous phase entering from the top and the less dense organic phase entering from the bottom. This configuration required a phase inversion to occur when first in contact, increasing the probability of slug flow pattern formation. The results suggest that for IFT’s approaching that of 18/12EE fish oils and water, or alkanes (i.e., hexane) and AgNO3 solutions, slug flow was able to form within the 1.59 mm Tygon channels. When 18/ 12EE fish oils were injected with the silver nitrate solution, the

Figure 5. IFT vs RPM obtained using spinning drop tensiometry for different wt % mixtures of fish oils and hexane/hexene. Droplet breakup limited the determination of IFT above 4000 rpm for all but the fish oil/water system, which plateaued at ∼2.6 mN/m as RPM increased to 9000.

solvents, water, and silver nitrate solutions. A significant variation in IFT was observed depending on the presence of complexation (fish oil/silver nitrate vs fish oil/water) and the degree to which an organic solvent was employed. The fish oil/ DI water mixture was the only fluid pair for which IFT could be determined at RPMs approaching 9000, at which a stable value 8300


Article


(3) Shahidi, F.; Wanasundara, U. N. Omega-3 fatty acid concentrates: nutritional aspects and production technologies. Trends Food Sci. Technol. 1998, 9, 230−240. (4) Kamio, E.; Seike, Y.; Yoshizawa, H.; Ono, T. Modeling of extraction behavior of docosahexaenoic acid ethyl ester by utilizing slug flow prepared by micro reactor. AIChE J. 2010, 56 (8), 2163− 2172. (5) Kamio, E.; Seike, Y.; Yoshizawa, H.; Matsuyama, H.; Ono, T. Microfluidic extraction of docosahexaenoic acid ethyl ester: comparison between slug flow and emulsion. Ind. Eng. Chem. Res. 2011, 50 (11), 6915−6924. (6) Belarbi, E. H.; Molina, E.; Chisti, Y. A process for high yield and scaleable recovery of high purity eicosapentaenoic acid esters from microalgae and fish oil. Enzyme Microb. Technol. 2000, 26, 516−529. (7) Traynham, J. G.; Sehnert, M. F. Ring size and reactivity of cyclic olefins: Complexation with aqueous silver ion. J. Am. Chem. Soc. 1956, 78, 4024−4027. (8) Teramoto, M.; Matsuyama, H.; Ohnishi, N.; Uwagawa, S.; Nakai, K. Extraction of ethyl and methyl esters of polyunsaturated fatty acids with aqueous silver nitrate solutions. Ind. Eng. Chem. Res. 1994, 33, 341−345. (9) Burns, J. R.; Ramshaw, C. The intensification of rapid reactions in multiphase systems using slug flow in capillaries. Lab Chip 2001, 1 (1), 10−15. (10) Charpentier, J. C. Process intensification by miniaturization. Chem. Eng. Technol. 2005, 28 (3), 255−258. (11) Kashid, M. N.; Harshe, Y. M.; Agar, D. W. Liquid−liquid slug flow in a capillary: An alternative to suspended drop or film contactors. Ind. Eng. Chem. Res. 2007, 46 (25), 8420−8430. (12) Yazawa, K. Purification of highly unsaturated fatty acids, AA, EPA, DHA. In Highly Unsaturated Fatty Acids; Kayama, M., Ed.; Kouseisyakouseikaku Co. Ltd.: Tokyo, Japan, 1995; pp 1−10. (13) Breivik, H.; Libnau, F. O.; Thorstad, O. Process for concentrating omega-3 fatty acids. WO Patent WO 2012038833 A1, March 29, 2012. (14) Stenerson, K.; Halpenny, M. R.; Sidisky, L. M.; Buchanan, M. D. GC Analysis of Omega 3 Fatty Acids in Fish Oil Capsules and Farm Raised Salmon. Reporter US Applications Newsletter 2013, 31.2, 14−15. (15) Liu, S.; Zhang, C.; Hong, P.; Ji, H. Concentration of docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) of tuna oil by urea complexation: Optimization of process parameters. J. Food Eng. 2006, 73, 203−209. (16) Currie, P. K.; van Nieuwkoop, J. Buoyancy effects in the spinning-drop interfacial tensiometer. J. Colloid Interface Sci. 1982, 87, 301−316.


drop in IFT observed in Figure 5 was sufficient to cause the transition to stratified flow.

5. CONCLUSION Liquid−liquid extraction of EPA/DHA from raw 18/12 EE fish oils was carried out using 50 wt % AgNO3 solution at 10 °C both in a 1.59 mm mini-fluidic system and in a 65 mL batch stirred reactor. EPA/DHA recoveries after ∼36 s of contact time in the mini-fluidic system were similar to those obtained from a CSTR after 15 min, with concentrations and yields comparable to other novel extraction technology such as supercritical CO2, urea precipitation, or molecular distillation. Typical concentrations within the final extracted product were 38 to 42 wt % EPA and 27 to 30 wt % DHA and a total Omega3 content between 78 and 82 wt %. The observed flow patterns in the mini-fluidic contacting system were notably different than those reported in previous work on this extraction method. Through the completion of interfacial tension measurements using spinning drop tensiometry, a significant increase in IFT was observed when real 18/12EE fish oils were diluted or replaced by alkane-based solvents (0.2 to 2.6 mN/ m). Given the sensitivity of the flow patterns observed to this variation in IFT, the change in flow regime and potential modifications needed for postextraction processing at an industrial scale warrant attention. The presence of a stratified flow system could potentially lead to different interstitial velocities for each phase, creating an unanticipated disparity in residence times of the organic and aqueous phases.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.5b01780. Tables including feedstock composition (winterized and nonwinterized), sample masses collected during testing by residence time, individual EPA/DHA/Omega3 levels obtained for each reactor and residence time tested for the Residual Oil, Fraction 1, and Fraction 2, and overall yield estimates for the trials carried out (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors would like to acknowledge financial support for this work from the Natural Sciences and Engineering Research Council of Canada, Grant 449352, and additional contributions from A. Neima at Dalhousie University.

■

REFERENCES

(1) Lembke,P. Production techniques for Omega-3 concentrates. In Omega-6/3 Fatty Acids: Functions, Sustainability Strategies and Perspectives; De Meester, F. et al., Eds.; Springer Science & Business Media: New York, 2013; pp 353−364. (2) Rubio-Rodríguez, N.; Beltrán, S.; Jaime, I.; de Diego, S. M.; Sanz, M. T.; Carballido, J. R. Production of Omega-3 polyunsaturated fatty acid concentrates: A review. Innovative Food Sci. Emerging Technol. 2010, 11, 1−12. 8301


12EE Fish Oil Using AgNO3(aq

Recommend Documents