SUPERCRITICAL FLUID EXTRACTION OF HEMP AND PAPAYA SEED OILS: CHARACTERIZATION, ANALYSIS, MODELING AND OPTIMIZATION

Abstract

In the developing world, demands of natural and organic free products are increasing in safety concern of food, health, environment, etc. Therefore, in the area of separation/extraction techniques, supercritical fluid extraction (SFE) is being the most applied technique due to its emission free approach, efficient recovery and organic free products. SFE with solvent CO2 acquires significant attention due to its commendable properties such as inert, cost and environment efficient, inflammable, easily separable, non-toxic, etc. Various types of species have been extracted through SFE. Extract obtained through SFE is pure and it contains various components having food, health and medicinal values. However, there are various species, which are rarely explored despite containing valuable components. Considering these facts, this study applies SFE (SFE 1000F) for the extraction of hemp seed oil (HSO) and papaya seed oil (PSO). Hemp and papaya seeds are selected as these are less explored despite of being rich in nutrition. Proximate analysis of these materials is performed to investigate the moisture, ash and oil content. Materials are characterized using field emission scanning electron microscope (FESEM), Fourier-transform infrared spectroscopy (FTIR) and thermo-gravimetric analysis (TGA)/differential thermal analysis (DTA) to analyze the surface morphology, functional groups and thermal degradation, respectively. SFE parameters such as temperature, pressure, solvent flow rate, particle size and co-solvent (ethanol) flow rate are optimized through central composite design (CCD) to maximize the extraction yields. Moreover, artificial neural network (ANN) modeling is also carried out to optimize SFE for both seeds. Obtained oils are analyzed through gas chromatography (GC) and ultra violet (UV) spectroscopy. The primary fatty acids of HSO such as ω-6 linoleic and ω-3 α-linolenic acids are optimized through CCD while maintaining the rare and optimized ratio of ω-6/ω-3 as 3:1. Similarly, primary component of PSO i.e. oleic acid is optimized through CCD due to its rich and nutritious values. Physico-chemical properties such as refractive index, specific gravity, flash point, fire point, peroxide value, acid value, saponification value, unsaponifiable matter and iodine value for both oils are determined. Further, SFE is compared with other extraction processes for HSO such as Soxhlet (SOX), percolation (PER), ultrasonication (ULT), ultrasonication treated Soxhlet (UTS) and Soxhlet treated ultrasonication (STU) to investigate the best process in terms of yields, functional groups, surface morphologies, chemical compounds and industrial scale economic analysis. Gas chromatography–mass spectrometry (GC-MS) is used for analyses of obtained oils. Further, two mathematical models (PM-1 and PM-2) are proposed and validated for HSO and ii PSO yields. These models are also compared with the existing models such as Reverchon model (RM) and Sovova model (SM) (Reverchon, 1996; Sovova, 1994). Proximate analysis of hemp seed estimates moisture, ash and oil contents as 5.39 %, 8 % and 36.30 %, respectively. Characterizations indicate high oil concentration due to shiny surface, presence of various functional groups and thermal stability at high extraction temperature (≈200 °C). CCD suggested 32 experimental runs, which are performed through SFE and estimates the extraction yield of HSO in the range from 3.70 % to 36.26 %. ANOVA analysis exhibits that all five parameters are significant (p<0.05) and contribute in the model individually and interactively. A significant reduced quadratic model is developed to predict the yield with R2 value of >0.92. ANN modeling demonstrates the most optimized configuration as FFBP-ANN [5 8 1], which denotes the feed forward back propagation ANN with 5, 8 and 1 neurons in input, hidden and output layers, respectively. It develops an equation to predict the HSO yield. Physico-chemical properties of HSO are observed within the range of edible oil requirement. GC estimates the concentrations of fatty acids such as ω-6 linoleic, ω-3 α-linolenic, oleic, palmitic, stearic, γ-linolenic, eicosanoic (arachidic), eicosenoic and behenic in the range of 48.85-63.66 %, 16.71-26.20 %, 8.04-13.62 %, 3.00-10.08 %, 0.40-3.99 %, 0.11-0.95 %, 0.22-0.91 %, 0.10-0.82 % and 0.08-0.78 %, respectively. UV spectroscopy estimates chlorophyll a, b and carotene contents in the range of 84.0-293.9 μg/g, 46.4-216.4 μg/g and 4.30-66.6 μg/g, respectively. CCD optimization of ω-6 linoleic and ω-3 α-linolenic fatty acids exhibits the significant effect of SFE parameters on concentrations except temperature. Mathematical models; PM-2, PM-1, RM and SM exhibit the satisfactory fitting with experimental data of HSO with average AARD as 11.37 %, 11.45 %, 11.68 % and 13.44 %, respectively. The extraction of HSO is compared with other extraction processes such as SOX, PER, ULT, UTS, STU and pyrolysis (PYR). Maximum yield is obtained through UTS followed by STU, SOX, SFE, ULT, PER and PYR. GC-MS of obtained HSO estimate the fatty acids in the range of 14.91-52.39 % (ω-6 linoleic acid), 2.6-21.41 % (ω-3 α-linolenic acid), 0.55-3.01 % (palmitic acid), 0.11-0.44 % (stearic acid) and 6.17-8.09 % (oleic acid). GC-MS of hemp bio-oil through PYR exhibits its potential to use it as bio-fuel. FESEM analysis of each extracted sample verifies the maximum yield through UTS. FTIR peaks of fresh sample are sharper and longer than the extracted samples through different processes. For industrial scale economic assessment, SFE is observed as most feasible process followed by SOX, STU, UTS and ULT. iii Moisture, ash and oil contents in papaya seed are determined as 6.84 %, 7.8 % and 36.85 %, respectively. The surface of papaya seed is observed as net type structure possessing high oil content. Further, characterizations of seed demonstrate the presence of various functional groups and thermal stability at high temperature. Similar to HSO, 32 experiments are performed using SFE and obtained PSO yield as 10-36.67 %. CCD optimization of PSO yield indicates that all parameters except temperature are significant. However, temperature exhibits significant effect in the interaction with other parameters. Reduced quadratic model is observed as significant for PSO yield with high R2 value as >0.91. Fatty acids in PSO such as oleic, ω-6 linoleic, ω-3 α-linolenic, palmitic and stearic are observed in the range of 53.05-86.99 %, 0.44-12.47 %, 1.11-27.55 %, 0.56-20.41 % and 2.22-21.74 %, respectively. CCD optimization of oleic acid develops a significant reduced quadratic model with R2 >0.78 to predict the concentration. Similar to HSO, FFBP-ANN [5 8 1] configuration is optimized for PSO and an equation is developed to predict the yield. Properties of PSO are observed in the limit of oil edibility. The best model fitting of PSO yield is obtained through PM-2 with average AARD as 12.83 % followed by PM-1, RM and SM with AARD values of 14.06 %, 17.77 % and 19.91 %, respectively. Industrial scale economic analysis of PSO shows SFE as profitable process. Hence, SFE of HSO and PSO demonstrates the high quality of oil with 99 % recovery. Further, it also proves its economic feasibility at industrial scale.