Supercritical fluid carbon dioxide extraction of Nigella sativa (black cumin) seeds using taguchi method and full factorial design

Optimum condition for Nigella sativa seeds oil and its bioactive compound, thymoquinone (TQ) using supercritical fluid carbon dioxide extraction (SCFE-CO2), were investigated. The optimization process was performed with Taguchi method and full factorial design (FFD) under the following condition: pressure (150, 200 and 250 bar), temperatures (40, 50 and 60oC) and carbon dioxide (CO2) flowrate (10, 15 and 20 g/min), in which solvent to feed (SF) ratio was set constant at 24. The highest yield of Nigella sativa seeds oil from SCFE-CO2 process with FFD method was 12% at 250 bar, 60 oC and 20 g/min. Whereas Taguchi method was performed at 250 bar, 50oC, 10 g/min with oil yield 11.9%. The highest thymoquinone content in Nigella sativa seeds oil from both experimental design was achieved through SCFE-CO2 extraction condition at 150 bar, 60 oC and 20 g/min with thymoquinone content 20.8 mg/ml. In addition, conventional methods such as high pressure soxhlet with liquid CO2, n-hexane soxhlet and percolation with ethanol were performed with oil yield 5.8%, 19.1% and 12.4%; and thymoquinone content 8.8 mg/g oil, 6.3 mg/g oil and 5.0 mg/g oil, respectively. The analysis of variance (ANOVA) with 95% confidence interval, indicates effect of pressure on Nigella sativa seeds oil yield and thymoquinone content from SCFE-CO2 process. The oil was then evaluated for its activity. The antibacterial activity of the oil from Taguchi Method, shows that all samples were unable to inhibit Escherichia coli O157 and Salmonella typhimurium. For other pathogenic bacterias, all samples show similar inhibition at concentration 10% oil for MRSA, 3% oil for Stapylococcus aureus and 3% oil for Bacillus subtilis.

Supercritical fluid extraction (SCFE) has many advantages compared to conventional extraction methods such as its environmentally compatible fluids, low temperature, shorter extraction time and ease separation of solute from supercritical fluid solvent by simple expansion. Moreover, this extraction does not leave any trace of organic solvent [13][14][15][16][17]. The most frequently employed supercritical solvent is carbon dioxide (CO 2 ). It has low toxicity, good safety, cheap, abundant in pure form (food grade) and low critical temperature and pressure [13,14,16,17].
Optimization process is critical as various parameters may potentially affect the SCFE process. The optimum condition can be achieved by investigating multiple factors in all possible conditions [18]. This condition was considered achieved when high yield of oil and high content of thymoquinone was obtained. The objective of this present research is comparing Nigella sativa seeds oil extracted by SCFE process using two different experimental designs, Taguchi's method and full factorial design. Other conventional extraction methods were also performed as comparison.

Materials and methods
Nigella sativa seeds were imported from Arab by "Intisari Herbal" supplier (Solo, Indonesia). Prior to the extraction process, all dried plant materials were grounded in milling machine (China) with stainless filter mess #10. Particle distribution of grounded raw material is shown at ( Table 1). Moisture content was determined using H83 Halogen Moisture Analyzer (loss on drying 6.87% ) and Volumetric KF Titrator V30 (water content 5.99%).

Design of experiment Taguchi method
The Taguchi method utilizes orthogonal array design (OAD) to study a large number of parameters with a small number of experiments. It significantly reduces number of experimental configurations to be studied without affecting the quality of results [19] and maintain the experimental cost at minimum level [20]. Taguchi Method was employed a special set of OAD in each number of experimental conditions to systematically look for the favorable operating conditions [18]. In this study, the following parameters are considered in the SCFE-CO 2 process: pressure (A), temperature (B) and CO 2 flowrate (C), and each parameter consists with three levels (values). Thus, the L 9 orthogonal arrays improved by Taguchi method as shown in ( Table 2), was chosen to determine the experimental design.
The quality of results were measured by the signal-to-noise (S/N) ratio with three different characteristic of target values: "larger is better", "smaller is better" or "nominal is better" [18,20]. Since purpose of this study was the highest oil yield and TQ content, target values of "larger is better" was chosen. In this study, interactions among variables were not considered and focus was placed on the main effects of three parameters.

Full factorial design (FFD)
FFD is the most basic structures experimental design. In FFD, responses are measured at all combinations of experimental factor levels [21]. Each experimental condition is called a "run" and each run represents variation of one variable. Each response will be measured for an observation. The entire set of runs is the "design". Since, we consider three parameters with three levels, the total run with full factorial design is 3 3 (27) number of trials. The variation levels for parameters is shown in (Table 3).

Supercritical carbon dioxide extraction
The supercritical carbon dioxide (SCFE-CO2) extraction was carried out using supercritical fluid extractor with CO 2 cycle system (KIST-Korea) as shown in (Figure 1). The extractor vessel with 2 x 1000 ml capacity was loaded with 100 g powdered material of Nigella sativa. Food grade liquid CO 2 was delivered to extraction vessel using high pressure pump (Thar, USA). Extraction pressure was varied from 150-250 bar and temperature 40-60 o C. The pressure in the extraction vessel was controlled by back pressure regulator (Tescom, USA). Heat exchangers (Lab. Companion, USA) were provided in system to maintain temperature in the extractor and separator vessel. CO 2 flow rate was varied from 10 to 20 g/min for static extraction time of 60 minutes (fixed for all the extractions) and followed by dynamic extraction time (120, 160 and 240 min) that had been determined based on each CO 2 flowrate. Therefore, the solvent to feed (SF) ratio was constant at 24. Extract was collected every 20 min (cummulated) until dynamic extraction time finished. Yield of extract was determined as summation of extract from the experiments.

High pressure soxhlet (HPS) extraction
About 50 g of powdered Nigella sativa seeds were submitted to high pressure soxhlet extraction (KIST-Korea) with CO 2 liquid as solvent. The operating condition for HPS extraction is at boiler temperature 35 o C, chiller temperature -5 o C (Lab. Companion, USA), pressure 40-50 bar, CO 2 cycle-rate 5 min and extraction time 7 hours. The remaining extract was considered as experiment yield.

Soxhlet n-hexane extraction
About 50 g of powdered Nigella sativa seeds were extracted with 600 ml n-hexane (ratio 12:1). Operating condition is at boiler temperature 100 o C, chiller temperature -5 o C (Lab. Companion, USA), pressure 1 bar and extraction time 6 hours. The extract was filtered and then evaporated to separate   the organic solvent (n-hexane). The remaining extract was considered as experiment yield.

Solvent extraction (ethanol 96%) with percolator
About 200 g of powdered Nigella sativa seeds were extracted with 2700 ml Ethanol (96%, Technical Grade). Operating condition for percolation process is at pressure solvent pump (HKS, Korea) 10 bar, room temperature with solvent flowrate at 180 ml/min and the extraction time 7 hours. The extract was filtered and then evaporated to separate the organic solvent (ethanol). The remaining extract was considered as experiment yield.   280ºC and held isothermal for 5 min. The injector and detector temperature were maintained at 220ºC and 240ºC, respectively. Sample solution was prepared with concentration of 4000 ppm in 2-propanol (chromAR). The injection volume was 1.0 µl with a split ratio at 10:1. Flowrate carrier nitrogen (N 2 ) gas at 1.0 ml/min, Hidrogen (H 2 ) at 45.0 ml/min and compress air UHP (Ultra High Purity) at 450.0 ml/min.

Anti-microbial analysis
Bacterial species used in this study were Escherichia coli O157, Bacillus subtilis ATCC 6633, Salmonella typhimurium ATCC 13311, Stapylococcus aureus ATCC 6538 and Methicillin Resistant Staphylococcus Aureus (MRSA). Antibacterial activity was determined by disc diffusion assay. Nigella sativa seeds oil was dissolved in dimethylsulphoxide (DMSO; Merck, Germany) to reach final concentration 3%, 5%, 10% and 100% (without dilution) and sterilized using filtration method with 0.22 μm membrane filters (Iwaki, Japan). The medium used were MHA (Mueller-Hinton Agar) supplied from Merck (Germany). The 6 mm diameter disc (Whatman, USA) were impregnated with 20 µl of the oil and placed on inoculated agar (10 6 CFU/ml of bacterial). Ciprofloxacin 0.2 mg/ml was used as positive reference standards to determine the sensitivity of tested microbial species. Control test was dissolved in 100% DMSO and showed no inhibition of microbial growth. The inoculated plates were incubated at 37 o C for 24 h for bacterial species growth. Antibacterial activity was evaluated by measuring zone of inhibition against those bacterial species. All inhibition assays and controls were carried out triplicate.

Statistical analysis
The data obtained from supercritical carbon dioxide extraction were subjected to analysis of variance (ANOVA) to determine significant difference among all extract yield. P-value less than 0.05 were considered significant. All statistical analysis were performed using MINITAB v.15 (Minitab Inc., USA) statistical software package.

Results and Discussion
( Taguchi method (Figure 2a and 2b the optimum conditions for Nigella sativa seeds oil yield were at P = 250 bar, T = 50 o C and CO 2 flowrate = 10 g/min (Figure 2a). Whereas the optimum conditions for thymoquinone content were obtained at P = 150 bar and T = 60 o C with CO 2 flowrate 20 g/min. The antibacterial activity of Nigella sativa seeds oil were tested with disc diffusion assay method on several strains of bacteria such as Escherichia coli O157, Bacillus subtilis, Salmonella typhimurium, Stapylococcus aureus and Methicillin-Resistant Staphylococcus Aureus (MRSA) ( Table 5). Based on the results of antibacterial activity test, all samples did not inhibit E. coli O157 and Salmonella typhimurium even without oil dilution treatment. For other pathogenic bacterial, all samples show similar inhibition (insignificant difference) at 10% oil for MRSA, 3% oil for Stapylococcus aureus and 3% oil for Bacillus subtilis. All samples show a better inhibition against Gram-positive bacteria compared to Gram-negative bacteria.
The higher resistance of Gram-negative bacteria to external agents is due to the presence of lipopolysaccharides in their outer membranes, which make them inherently resistant to antibiotics, detergents and hydrophilic dyes.

Full factorial design (FFD)
FFD requires more experimental points, however it can provide more information on the effect of various combinations. ( Table 6) show various combination factors including pressure (P, 150-250 bar), temperature (T, 40-60 bar) and CO 2 flowrate (10-20 g/min). The optimum conditions to obtain high oil yield was at 250 bar and 60 o C. All three level of CO 2 flowrate at this condition yield similar results 12% of Nigella sativa seeds oil. Therefore, 20 g/min was considered as optimum level for CO 2 flowrate This result is different compared to Taguchi method at 250 bar, 50 o C and 10 g/min with oil yield 11.9%. However, the difference between these two optimum condition is   insignificant (0.1%). For TQ content, both experimental designs gave similar results at 20.8 mg/g oil. Therefore, taguchi method was able to simplify the experimental procedure without affecting the quality of results. Generally, higher pressure at isothermal conditions result an increase in solvent density, subsequently the solvent power and solubility of compounds. As the density increased, the distance between the molecules decreased; the interaction between the compounds and CO 2 increased; which leads to increase compounds solubility in CO 2 . Therefore, increase in pressure will also accelerate mass transfer analytes and solvent in supercritical extractor vessel system and increase the extraction yield of Nigella sativa seeds oil. Solubility of solutes in SCFE-CO 2 is affected by solvent density and volatility of solutes which depend on temperature in opposing ways. Higher temperatures increase the volatility of solutes and improve their solubility and extraction. On the other hand, at a given pressure, density of supercritical CO 2 decreases with increase in temperature, reduction in solvation power of CO 2 and thus reduction in solubility and extraction efficiency [23][24][25].
The significance of each independent variable determined by analysis of variance (ANOVA) are presented in (Table 7a and 7b). With 95% confidence interval, only pressure affect the oil yield, while two others parameters (temperature and CO 2 flowrate) have no significant effects (P>0.05). The ANOVA results are in accordance with S/N ratio that identify pressure as the main parameter.
Other extraction methods such as high pressure soxhlet with liquid CO 2 , n-hexane soxhlet extraction and percolation with ethanol were performed with yield 5.8%, 19.1% and 12.4%, respectively ( Table 6). Although, yield from n-hexane soxhlet extraction and percolation process higher than optimum results of SCFE-CO 2 but the extract from both methods still contain organic solvent. Moreover, the final extract from these three different methods has lower content of thymoquinone compared to the SCFE-CO 2 process, with concentration of 8.8 mg/g oil, 6.3 mg/g oil and 5.0 mg/g oil, respectively. Therefore, SCFE-CO 2 seems to be a better alternative for Nigella sativa seeds since it offers the usage of non-toxic, non-explosive, environmental friendly, cost effective, time saving and selectivity-adjustable solvent in the extraction process.