Optimization of Ginger (Zingiber officinale) Oil Yield from Malaysia in Different Hydrodistillation Physical Parameters via Central Composite Design of Response Surface Methodology (RSM)

Response Surface Methodology (RSM) with Central Composite Design face cantered (CCD) is the statistical analysis used for optimization process of ginger oil yield from Malaysia. In this study the three factors predictors (Water volume, extraction time and particle size) and one response criterion (Ginger oil yield) were optimized. The final optimization value for high yield of ginger oil is at water ratio of 2660 mL water to 100 g sample at 23.15 h of hydro distillation using medium size particles of dried ginger. Under these conditions the oil constitute majority of sesquiterpenes hydrocarbon which is 33.49% with alpha-zingiberene as a major compound.


INTRODUCTION
The trend for Malaysia's export volumes of ginger had increased between years 2000 to 2007 suggesting gingers are among important herbs usage in Malaysia (Globinmed, 2012). The essential oil compound properties as antifungal, antibacterial and antioxidant were the major interesting medicinal benefits discovered. It is also approved by FDA as food additive. There are many studies in wide world on different source of ginger and different essential oil compound yield. The variation in compound constituents depends on geographical factor and extraction methods used.
Sample preparation is crucial to obtain high yield of ginger oil. Factors involved include sample physical characteristic, drying methods and particle size. Sample physical characteristic can be either it is fresh or dried, peeled or unpeeled skin and young or old ginger. Upon drying, factors involved are time and methods (oven dried, microwave dried, sun dried, freeze dried and air dried). Particle sizes which involved slicing methods are as well important as others.
Unpeeled gingers were found to have highest yield of oil/oleoresin (Yiljep et al., 2005). Peeling or removal of skin may remove thick fibrous tissue of ginger suggesting loss of oil cells that may also reside at the cell parenchymal wall (Jayashree et al., 2012;Noor et al., 2004). Upon using fresh ginger in opposition to dried, it was found the chemical composition in fresh ginger are more than dried ginger due to drying process. Most monoterpene and sequiterpene alcohol compounds are decrease in ginger oil from dried ginger compared to fresh ginger (Sasidharan and Menon, 2010).
The particle size contributes on the yield of oil where it is suggested concentration of essential oil in ginger will decrease when there is increase in size (Okafor et al., 2009). However exact particle size requires for high yield of oil is undetermined.
To control several variables in an experiment is challenging upon acquiring a maximum and quality optimum response. Conducting One Factor at Time (OFAT) experiment per se failed to detect interaction between variables as the variables are not varied at one time. Design of Experiment is a statistical tool approaching controllable variables (input) and responses variables (output) allowing detection of possible interaction in a well designated structured and effective experiment plan. Response Surface Methodology (RSM) is suitable as it is an economical, robust and widely uses methods in optimization process of essential oil successfully (Tan et al., 2012;Rezzoug et al., 2005).
The objective of this study was to apply Central Composite Design (CCD) based RSM to analyze the effects of the process parameters on essential oil production and to search for the optimal values for attaining a higher essential oil yield.

Plant material:
The fresh rhizomes of ginger plant (Zingiber officinale) were purchased from herbal plantation (Kiza Herbs Sdn. Bhd) located in Pahang. The voucher specimens were identified and deposited at UKM Herbarium, Bangi (UKMB 30013). Ginger plants collected were washed and unpeeled. Dry gingers were obtained by slicing thin rhizome and were air dried for 3 days. The water content was determined using Dean's Stark apparatus. The dried rhizomes were then ground and sieved at 3 different sizes; coarse (>2000 µm), medium (2000-800 µm) and fine (<800 µm).

Isolation of essential oil:
The ground dried gingers (100 g) were hydro distilled using Clevenger apparatus at three different water-sample ratio content of 1000 mL: 100 g, 2000 mL:100 g and 3000 mL: 100 g at three different times of 8, 16 and 24 h, respectively. Temperature was remained constant at 40°C. The extracted oil was dehydrated with anhydrous sodium sulfate to isolate ginger oil.
Experimental design: RSM were employed in this study to improve the isolation process by optimizing the yield via CCD. A set of random experimental trial work with combination of factors involved in process for high and optimum production of desired results were effectively designed by Design of Experiment (DOE).
A central composite face centered design with was used. The design includes three variables factors (Water volume, extraction time and particle size) and 1 response (Ginger oil yield). There is 33 actual experiments with 3 factors (k = 3), 3 level with 3 centre points to form a central composite design. The experimental ranges of each variable factor were based on the results from preliminary trials and usual standard lab protocol practice in the laboratory. Variables, code, range and level of the experimental work are as in Table 1.
The suggested optimized values later were used to do a confirmation run to validate the model is fits and accurate.
Essential oil analysis: The ginger oil extracted was dissolved in hexane and was analyzed using Shimadzu GC 14A using a FID detector and DB-5 stationary phase column (30 m×0.25 mm, 0.25 | | m film  thickness) programmed from 75°C for 10 min, then 3°C /min to 230°C and held for 10 min). It was operated under nitrogen as carrier gas at 50 cm/sec, injector and detector temperatures were maintained at 250°C. The areas and retention times from chromatogram were used to measure linear retention indices of the compounds which were relative to n-alkanes (Adam, 2001). For GC-MS analysis, it was operated using Hewlett Packard GC-MSD 5890 with combination of 70 Ev Electron Impact Ionisation (EI) zone and BPX5 column (30 m×0.25 mm, 0.25 µm film thickness) programmed similar protocol as in GC. Identification of the compound were compared with reference spectra in NIST library and confirmed via retention indices of compound from literature from GC spectrum (Adam, 2001).

RESULTS AND DISCUSSION
Ginger essential oil yield: The spectrum result of CG for analyzed ginger oil is as in Fig. 1. The ginger oil yield was average of 0.95% at 24 h. Upon comparing with other ginger from different location, the ginger oil yield from Malaysia ginger is among the lowest of all (Onyenekwe and Hashimoto, 1999;Sultan et al., 2005).
The compound profile listed above shows good quality ginger oil. Ginger oil must be characterized by high sesquiterpene hydrocarbons (alpha zingiberene, arcurcumene, beta-sesquiphellandrene) follows by other compounds (Sivasothy et al., 2011;Onyenekwe and Hashimoto, 1999;Wohlmuth et al., 2006;Sasidharan and Menon, 2010). The quality of oil was also contributed by the age of oil which is by determination ratio of zingiberene+beta-sesquiphellandrene to arcurcumene (Zachariah and Gopalan, 1987). A ratio of 2:3 is the recommended ratio for a good quality of ginger oil (Zachariah, 2008).
Different physical characteristic and distillation parameter will provides different quality of ginger oil odor. In industry where aroma is essential, a mixture of alpha terpineol, neral, geraniol, beta sesquiphellandrene, ar-curcumene and nerolidol is characteristic of ginger oil aroma (Prabhakaran, 2013). Table 3 shows the actual factorial experiment and response value that were designed by Design of experiment and were performed which comprised three factors predictors (Water volume, extraction time and particle size) and one response criterion (Ginger oil yield).

Optimization of ginger essential oil: The
Result for Analysis of Variance (ANOVA) is shown in Fig. 2. The significant level of each coefficient is determined by p value <0.0001. From ANOVA, the model is statistically significant with Fvalue of 38.15, thus the model is accepted and there is only 0.0001% chances that this due to noise.
In this model, A (Water volume) B (Extraction Time), C (Particle size), AC (Water volume×Particle size) and A 2 (Water volume) 2 are significant model terms with significant p-value <0.0001. These factors have significant effect on response individually or via interaction. Non-significant lack of fit F-value of 0.99 indicate that the replication variation on the region and design point of the model are fit and there is that 0.52% chance the value occur due to noise.
Factor A, B and C when act independently preferred gained higher ginger oil yield at water ratio <1000 mL to 100 g, time >24 h and using medium particle sizes. However upon synergistic relationship, interaction between AC was significantly exist (F-value 17.68) where at water ratio 2000 mL to 100 g and using medium particle size, ginger oil are highly obtained. Here is where the desirability parameter was selected to optimize the process.
Individually, water ratios have a trend of at certain point, the ginger oil production will drop and further increase in ratio will not be beneficial (Teoh et al., 2012;Li et al., 2009). This may due to oil cells that are not rebuilt due to fully rupture of parenchyma cell wall (Noor et al., 2004).
Particle size wouldn't matter if the plant do not have secretory duct (Xu et al., 2011). When the size is big, there is longer diffusion or mass transfer between water and the plant particles. Thus the interaction between water ratio and particle size is important as stated in this studies to optimized ginger oil yield. The regression analysis was performed besides for prediction; it is also to understand or explained the relationship of three inputs to one output response. The coefficient of determination R 2 value is 0.95 suggesting there is a good prediction of the model. The prediction R 2 of 0.89 is also in reasonable agreement with adjusted R 2 of 0.92.
Model measurement for accuracy can be achieved by adequate signal of "Adeq Precision" more than 4. The ration of 23.396 in this model is adequate. Stability of this model to predict well in other sample can be evaluated from PRESS (Predictive Residual-Error Sum of Squares). A smaller value is desired thus 0.30 in this model is accepted (Palmer and O'Connell, 2009).
The residual analysis emphasizing on different from actual observed and predicted value would address the appropriateness of the model. From normal probability plot in Fig. 3, the residuals falls on a straight line suggesting the random error from experiments are normally distributed.
The residual versus predicted plot as in Fig. 4 shows that there were no unusual or skewed data suggesting no possible outliers. Upon optimization, a solution combining three factors above suggest at desirability value of 1, to obtained maximum ginger oil yield of 1.34%, the optimal water ratio to sample value is 2660 mL to 100 g, at 23.15 h using medium size particles. The 3D surface graph is as showed in Fig. 5. A confirmation run to validate the model were done at chosen suggested optimal parameter and revealed results of average ginger oil yield of 1.34% (Fig. 6).

CONCLUSION
From the analysis, we can conclude interaction of water ratio and particle size interaction contributes significantly for production of high ginger oil. From the optimize value, to obtained maximum ginger oil yield of 1.34%, the optimal water ratio to sample value is 2660 mL to 100 g, at 23.15 h using medium size particles via hydro-distillation. A confirmation run have validate the model is accurate.