LIQUID-LIQUID EXTRACTION (LLE) IN MINI-FLUIDIC FLOW REACTORS
An overview of the initial contacting and subsequent sample preparation and extraction process is illustrated in Figure 6.6. Experiments were carried out in a 1/16" ID food-grade Tygon tube connected via Luer fittings to a dualsyringe continuous pumping device. The aqueous and organic phases were pumped at 5 ml/inin and 1.47 ml/min by using the 60CC and 10CC syringes on the pump for each stream, respectively. A 2 m long tubing being dipped in a reservoir maintained at 10°C was employed to pre-cool each phase and then both phases were contacted in Y-junction where it was anticipated that a slug-flow profile would be produced. The combined flow was then allowed to proceed through set lengths of tubing (0.5, 1, 2, 4, and 6 meters) before being sampled for subsequent analysis. Due to the limitations of the syringe pumps employed, flow rates above those stated were not possible for these lengths (skipping of the screwdriver on the syringe pumps). The resulting residence times for each of these sample ports was approximately 0.6, 1.2, 2.4, 4.8, and 7.3 minutes.
FIGURE 6.6 Overview of the process employed within this study for contacting and subsequent sample extraction/preparation.
A few notable observations and cautions regarding the experimental setup are as follows. The initial system constructed used Leur check valves in the syringe pump arrangement for controlling flow direction, which subsequently failed after one use due to the high solubility of the silicone diaphragms used in the fittings. Three-way valves with Teflon stop-cocks were subsequently installed and manually adjusted at 10-minutes intervals to enable access to the feedstock reservoirs when the syringes were withdrawing during loading.
The same three-way valves were initially installed at the downstream sample lengths to simplify sampling but were subsequently removed due to the Teflon causing oil retention in the line, resulting in collected samples containing no appreciable oil fractions.
The silver nitrate solution used for these extractions contained 47.5 wt% of AgN03 and 4.76% wt% NaN03 in aqueous solution. Batches of solution were prepared prior to each run to avoid long-term oxidation of the non-stabilized silver nitrate solution. Each solution was prepared by slowly adding 400 grams of 99.9% ACS grade silver nitrate to distilled deionized water until dissolved, after which 40 grams of NaN03 was added as an ionic strength adjuster to provide a comparable composition to what was previously reported by Kamio et al. (2010, 2011). Following preparations, solutions were purged with nitrogen and stored in a blanketed opaque container in the fridge to minimize oxidation prior to use.
The samples collected for each contact time were subsequently fractionated and conditioned for subsequent analysis, as per the qualitative description provided in Figure 6.7. The initial immiscible oil fraction was separated by retaining it within the sample syringe when transferring the aqueous fraction to a separate vessel. De-emulsification was then performed on the aqueous phase through the addition of hexane (typically 10 ml), which was agitated following mixing to promote contact prior to fractionation in a separatoiy funnel. The remaining aqueous phase theoretically contained EPA and DHA which had formed a complex with the silver ions in solution. To recover these complexed ethyl esters, the fractionated aqueous phase (typically having a volume of ~50 ml) was placed in a separate vessel and diluted with 600 ml of water. 20 ml of hexane was then added to the diluted solution to facilitate recovery of the de-complexed oils, with fractionation subsequently
FIGURE 6.7 Step up a process for recovery of co-3 PUFA from aqueous phase and emulsion phase.
occurring in a separatory funnel. The hexane and hexene fractions recovered from this process were subsequently filtered and subjected to a nitrogen flow to vaporize the hexane and hexene solvents. The masses of fractions during each processing step were recorded in an attempt to determine approximate yields, although oil residuals on glassware, sample losses during transfer, and residual water content are expected to limit the quantitative accuracy of mass balances based on these results.
LIQUID-LIQUID EXTRACTION (LLE) IN STIRRED REACTOR
The extraction was earned out in a batch-wise stirred vessel using a traditional system and the results were compared to the mini-fluidic system to determine if there was a significant difference in extraction performance. A 65 ml three-neck flasks equipped with an outer 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 FO was added and the fluids allowed contacting under stirred conditions for 15, 30, 60, 90, and 120 minutes. At the end of reaction time, the stir bar was shut off and the contents allowed separating by gravity. After the removal of the oil layer formed, the aqueous layer was processed for separation of co-3 PUFA.
POST EXTRACTION PROCESSING OF AQUEOUS AND ORGANIC EXTRACT
After extraction of co-3 PUFA in the mini-fluidic or batch reactors, the aqueous phase enriched with co-3 PUFA was further processed to determine yield and purity. The processing of the extraction mixture consisted of 4 steps (Figure 6.7) Gravity separation of the oil phase, de-emulsification with hexane, de-complexation using hexane and water, and sample drying and filtering prior to analysis.
- 1. Oil Residue Recovery: 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 separatoiy funnel, where the aqueous/emulsion phase was removed and the residual oil phase collected. 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.
- 2. De-Emulsification of Oil Phase: An emulsion phase between the ethyl esters of FO and AgN03 solution was formed during the extraction process, allowing for fast mass transfer between co-3 PUFA and silver ions. This emulsion was not the complexed PUFA and needed to be separated prior to the decomplexation of the PUFA from the silver nitrate solution. De-emulsification was carried out by the addition of hexane in the ratio 1 part hexane to 10 parts sample volume. In the batch system, hexane was added to the separatory funnel and mixed thoroughly prior to recovering the aqueous and organic fractions in separate containers. For the syringe samples, Hexane was transferred to the sample syringe containing the emulsion/aqueous phase, with contacting achieved by repeated transfer between two syringes through a short coimecting 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 de-emulsified hexane fraction is henceforth named as fraction 1.
- 3. Decomplexation of Aqueous Phase: 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 separatoiy funnel where it was brought into contact with 20 ml of hexane and 600 ml of water and allowed to contact for 120 minutes. The aqueous phase was then separated off and the organic fraction was stored for analysis. The non-polar solvent (hexane) was added into the aqueous phase to weaken the bond between EPA/DHA and silver ion and to increase the volume of the organic phase recovered. This final organic fraction was henceforth referred to as Fraction 2, representing the extracted PUFA.
SAMPLE PREPARATION AND ANALYSIS
The hexane (de-emulsified fraction) and hexene (de-complexed fraction) fractions recovered from this process were subsequently filtered and subjected to a nitrogen flow to vaporize the hexane and hexene solvents. The solvent-free fractions were then sealed and were subjected to gas chromatographic analysis. These values were used in conjunction with the mass of each of the collected fractions from these experiments to perform an approximate material balance to determine the wt% yields. While the exact analytical procedure used for detennining EPA and DHA content cannot be included in this thesis, a suitable method was identified from open literature which offers a brief description of what would be required. In order to find the content of EPA/DHA-Et and co-3 PUFA infractions, A BPX70 capillary column (30 m length, 0.25 mm i.d., phase ratio, b = 250) was specifically used to analyze the fatty acid methyl esters (FAMEs), utilizing the phasing technology with the aromatic group (Silphenylene) on a siloxane backbone. The total time for analysis is nearly 20 min which is reduced to 8 min using the shorter BPX70 column (b=250) that effectively separates the fraction into saturated (C16:0 and 08:0) and monounsaturated (C18:ln9 and C18:ln7) fatty acids. The content of fractions of GC chromatogram is expressed in tenns of peak area percentage of each component in the fractions, which is mentioned in Appendix Al.