INTRODUCTION
Jarak pagar or the scientific name is Jatropha curcas is live widly
and belongs to Euphorbiaceae family. It is cultivated in central and south America,
Southeast Asia, India and Africa (Gübitz et al.,
1996). The Jatropha curcas seed oil can produce about 4060% oil
which is rich with unsaturated fatty acid. The fatty acids are palmitic (13.00%),
stearic (2.53%), oleic (48.80%) and linoleic acid (34.60%). However, the chemical
compositions of the oil vary according to the climate and locality (MartínezHerrera
et al., 2006).
Linoleic acid is an essential fatty acid; the high content of linoleic acid
makes this oil very importance to the industries. The linoleic acid can be used
in protective coatings, plastics, urethane derivatives, surfactant, dispersants,
cosmetics, lubricants and varieties of synthetic intermediate, stabilizers in
plastic formulations and in the preparations of other long chain compounds.
The high content of linoleic acid in seed oil such as Jatropha curcas
seed oil is very important to the production of oleochemicals (Hosamani
and Katagi, 2008).
There are several method can be used to obtain unsaturated fatty acid including
freezing crystallization, urea complexation, molecular distillation, supercritical
fluid extraction, silver ion complexation, lipase concentration (Lui
et al., 2006) and HPLC (highperformance liquid chromatography) (Yamamura
and Shimomura, 1997). The cheapest and efficient technique to obtain linoleic
acid in the form of free fatty acid is urea complex fractionation. This is well
established technique to elimination of saturated and monounsaturated fatty
acids (Wanasundara and Shahidi, 1999; Chen
and Ju, 2001). Urea complexation has the advantage that complexes crystal
is extremely stable and filtration does not necessarily have to be carried out
at the very low temperatures which solvent crystallization of fatty acids would
be required. This methods also favored by many researchers because complexation
depends upon configuration of the fatty acid moieties due to the presence of
multiple double bonds, rather than pure physical properties such as melting
point or solubility (Wanasundara and Shahidi, 1999).
In the urea complex fractionation, the saturated and monounsaturated fatty acids
easily complex with urea and crystallize out on cooling and may subsequently
be removed by filteration. The liquid and NonUrea Complexed Fraction (NUCF)
is enriched with unsaturated fatty acids and the crystals formed or Urea Complexed
Fraction (UCF) is consist saturated fatty acids.
In this study, urea complex fractionation of a mixture of free fatty acids
of local Jatropha curcas oil was carried out to obtain concentrated with
high unsaturated fatty acids. The effects of urea/FFA ratio (w/w), crystallization
temperature (°C) and crystallization time (h) on the percentage of linoleic
acid and yield of linoleic acid were systematically studied.
MATERIALS AND METHODS
Materials: Jatropha curcas seeds were obtained from the Universiti Kebangsaan Malaysia field at 15/September/2009. The ripe seeds were collected and the damaged seeds were discarded. The seeds were cleaned, deshelled and dried in an oven at 105°C for 30 min. The seeds were ground to powder using a grinder prior to oil extraction. All chemicals used in the study were analytical grade and used without further purification.
Oil extraction: The extraction of Jatropha curcas seed oil was carried out using hexane as a solvent for 6 h.
Hydrolysis: Free fatty acid (FFA) was obtained by the saponification of Jatropha curcas seed oil. Typically, a KOH solution was prepared by dissolving 25 g KOH in 90% aqueous ethanol. A mixture containing 50 g oil and 300 mL KOH solution was heated at different temperatures and times. One hundred milliliter of hexane and water 200 mL were then added. A 6 N hydrochloric acid was then added until the solution was at pH 1. The resulting lower layer was removed using a separating funnel and discarded. The FFAcontaining upper layer was dried with anhydrous magnesium sulfate and solvent was evaporated in a vacuum rotary evaporator at 35°C. The FFA% was measured.
Urea complex fractionation: The separation of unsaturated fatty acid
from the hydrolyzed fatty acids of Jatropha curcas seed oil was carried
out by Wu et al. (2008). FFA (10 g) were mixed
with urea in 95% aqueous ethanol and heated at 60°C with stirring until
the mixture was turned into a clear homogeneous solution. The ratio of urea/FFA
was changed by using different amounts of urea. Initially, the ureafatty acid
adduct was allowed to crystallize at room temperature but colder temperatures
were maintained later for different periods for further crystallization. The
crystals formed (ureafatty acid adducts, also referred to as the urea complexing
fraction; UCF) were separated from the liquid (nonurea complexing fraction,
NUCF) by fast filtration. The liquid (NUCF) was diluted with an equal volume
of water and acidified to pH 23 with 6 N HCl; an equal volume of petroleum
ether was subsequently added and the FFA were extracted. The top phase, containing
liberated fatty acids, was separated from the aqueous layer containing urea.
The petroleum ether layer was washed with 5% NaCl solution to remove any remaining
urea and then dried over anhydrous Na_{2}SO_{4} and the solvent
was then removed at 65°C using a rotary evaporator.
Fatty acids analysis by GCFID: The identification of the fatty acids was accomplished by gas chromatography. FFAs were converted to FAME for GC analysis according to PORIM Method. Fatty acid profile was analyzed with Shimadzu GC17A with a BPX70 column (30 m x 0.25 mm x 0.25 μm films). Injection and detection (FID) temperatures were set at 260 and 280°C, respectively and nitrogen was used as the carrier gas with flow rate of 0.3 mL min^{1}. split ratio was 1: 39.
RESULTS AND DISCUSSION
Experimental design and statistical analysis: A threefactor Doptimal
design (Box, 1954; Cornell, 1992)
was employed to study the responses, namely after urea inclusion fractionation
the yield of NUCF (Y1 in % by wt., Eq. 2) and percentage of
linoleic acid (Y2 in %, Eq. 3). An initial screening step
was carried out to select the major response factors and their values.
The independent variables were X1, X2 and X3 representing the ureatofatty acid ratio (w/w), crystallization temperature (°C) and crystallization time (h), respectively. The settings for the independent variables were as follows (low and high values): ureatofatty acid ratio of 1 and 5; crystallization temperature of 10 and 10, crystallization time of 8 and 24. Each variable to be optimized was coded at four levels: 1, 0 and +1, the doptimal design is shown in Table 1.
A quadratic polynomial regression model was assumed for predicting individual Y variables. The model proposed for each response of Y was:
where, β_{0}; β_{i}; β_{ii} and β_{ij}
are constant, linear, square and interaction regression coefficient terms, respectively
and xi and xj are independent variables. The Minitab software version 14 (Minitab
Inc., USA) was used for multiple regression analysis, Analysis of Variance (ANOVA)
and analysis of ridge maximum of data in the response surface regression (RSREG)
procedure. The goodness of fit of the model was evaluated by the coefficient
of determination R2 and the Analysis of Variance (ANOVA).
Table 1: 
Independent variables and their levels for Doptimal design 

Response surfaces and contour plots were developed using the fitted quadratic
polynomial equations obtained from RSREG analysis and holding the independent
variables with the least effect on the response at two constant values and changing
the levels of the other two variables.
RESULTS AND DISCUSSION
The initial fatty acid mixture was composed of 13.19% palmitic (16:0), 0.40%
palmitoleic (16:0), 6.36% stearic (18:0), 43.32% oleic (18:1) and 36.70% linoleic
(18:2). Average molecular weight of the fatty acids was 203.36 as obtained from
saponification test of the original oil. The results agree with those of (Salimon
and Abdullah (2008). In this chapter, the PUFA (linoleic acid) concentration
has been done by using urea complex fractionation (GuilGuerrero
and Belarbi, 2001), using the FFA that was previously obtained.
The purpose of this procedure is to obtain a PUFA concentrate as enriched in
linoleic acid as possible and, simultaneously, to maintain the highest yields
of linoleic acid. The crystallization process with urea preferentially selects
saturated and monounsaturated fatty acids and the tendency of fatty acids to
combine with urea decreases with increasing chain length (AbuNasr
et al., 1954).
Experimental values obtained for response, the yield percentage of NUCF and the percentage of linoleic acid for twenty five design points are given in Table 2. The results show that LA had been purified in the filtrate, while monounsaturated fatty acid (oleic acid) and saturated fatty acid (Palmitic, Palmitoleic and stearic acids) were enriched in the crystal phase. Thus these results demonstrate that oleic, palmitic, palmitoleic and stearic acids have more tendency to form urea adducts than LA.
Hayes (2006) and Hayes et al.
(1998) have reported similar results for urea complex fractionation experiments
carried out for low erucic acid rapeseed oil, canola and vegetable and fish
oil. In certain experimental conditions the % of LA derived from the NUCF phase
was relatively high and some even greater than 90% (Table 2).
This showed that the experimental conditions should be suitable for the preparation
of high purity LA. However, it is difficult to completely remove all the saturated
fatty acids to obtain 100% purity of unsaturated fatty acids in the concentrate.
Ratnayake et al. (1988) has reported that complete
removal of saturated fatty acids by urea complexation may be impossible since
some of the saturated fatty acids do not bind with urea during crystallization.
Model fitting: The quadratic regression coefficient obtained by employing a least squares method technique to predict quadratic polynomial models for the yield percentage of NUCF (Y1) are given in Table 3. Examination of these coefficients with a Ttest shows that for the yield percentage of NUCF (Y1) ), the linear and square terms of ureatofatty acid ratio (X1) were highly significant (p<0.01) and the quadratic terms of the ureafatty acid ratio (X1) of the yield percentage of the NUCF (Y1) was significant (p<0.05).
The ureatofatty acid ratio (X1) and the crystallization temperature (X2) for the yield percentage of linoleic acid (Y1) in the concentrate were significant at p<0.05 and others between any two of the four factors were not.
Table 2: 
Doptimal design arrangement and responses for nonureacomplexed
fraction of Jatropha curcas seed oil 

Table 3: 
Regression coefficients of the predicted quadratic polynomial
model for response variables (the yield percentage of linoleic acid) in
urea inclusion fractionation experiment of Jatropha curcas seed oil 

** p<0.05; ***p<0.01. T: F test value. See table 2 for
a description of the abbreviations 
Table 4: 
Regression coefficients of the predicted quadratic polynomial
model for response variables (the percentage of linoleic acid) in urea inclusion
fractionation experiment of Jatropha curcas seed oil 

**p<0.05; ***p<0.01. T: F test value. See table 2 for
a description of the abbreviations 
Table 4 shows the percentage of linoleic acid (Y2) for the quadratic regression coefficient obtained by employing a least squares method technique to predict quadratic polynomial models. Examination of these coefficients with a Ttest shows that for the percentage of linoleic acid (Y2) the linear, square and quadratic terms of ureatofatty acid ratio (X1) were highly significant (p<0.01). Among six interactions, the ureatofatty acid ratio (X1) for the percentage of linoleic acid (Y2) were highly significant.
The coefficients of independent variables (ureatofatty acid ratio; X1, crystallization temperature; X2 and crystallization time; X3) determined for the quadratic polynomial models (Table 3) for the yield percentage of NUCF (Y1) and percentage of linoleic acid (Y2) are given below:
Diagnostic checking of the fitted models: ANOVAs for the fitted models are summarized in Table 5. Examinations of the two models with an Ftest and Ptest indicate a nonsignificant lackoffit at p>0.05 and pure error was 37.64%. The regression coefficient (R2) of the yield percentage of NUCF (Y1) was 0.94 (Table 3).
Table 6 shows the ANOVAs for the fitted models. Examinations
of the two models with an Ftest indicate a nonsignificant lackoffit at p>0.05
but the Ptest indicate a significant lackoffit was 0.0486 at p<0.05 and
pure error was 0.14%. The regression coefficients (R2) of the percentage of
linoleic acid (Y2) were 0.99 (Table 4).
Table 5: 
Analysis of variance, showing the effect of the variables
as linear, square and interactions on the response Y1 (the yield % of NUCF)
of the Doptimal design 

Table 6: 
Analysis of variance, showing the effect of the variables
as linear, square and interactions on the response Y2 (the % of linoleic
acid) of the Doptimal design 

These indicate that the generated models adequately explained the data variation
and represented the actual relationships among the reaction parameters.
Response surface plotting and optimization in the linear weighting method:
Equation 2 and 3 showed that the yield percentage
of NUCF and percentage of linoleic acid have a complex relationship with independent
variables that encompass both first and secondorder polynomials.
RSM is one of the best ways of evaluating the relationships between responses, variables and interactions that exist. Significant interaction variables in the fitted models (Table 3, 4) were chosen as the axes (ureatofatty acids ratio X1 and crystallization temperature X2) for the response surface plots. The relationships between independent and dependent variables are shown in the threedimensional representation as response surfaces. The response surfaces for the yield percentage NUCF (Y1) were given in Fig. 1a and b.
The response surfaces of the percentage of linoleic acid (Y2) is shown in Fig.
2a and b. In a contour plot, curves of equal response values are drawn on
a plane whose coordinates represent the levels of the independent factors. Each
contour represents a specific value for the height of the surface above the
plane defined for combination of the levels of the factors. Therefore, different
surface height values enable one to focus attention on the levels of the factors
at which changes in the surface height occur (Wanasundara
and Shahidi, 1999).
The contour plots (Fig. 1b, 2b) show the
combination of levels of the ureatofatty ratio and crystallization temperature
that can afford the same level of the yield percentage of NUCF and percentage
of linoleic acid.

Fig. 1: 
(a) Response surface and (b) contour plots for the effect
of the urea/FFA ratio (X1, w/w) and temperature (X2, °C) on the yield
% of NUCF (Y1, %) in the NUCF 

Fig. 2: 
(a) Response surface and (b) contour plots for the effect
of the urea/FFA ratio (X1, w/w) and temperature (X2, °C) on the % of
linoleic acid (Y2, %) in the NUCF 
Canonical analysis was performed on the predicted quadratic polynomial models
to examine the overall shape of the response surface curves and used to characterize
the nature of the stationary points. Canonical analysis is a mathematical approach
used to locate the stationary point of the response surface and to determine
whether it represents a maximum, minimum or saddle point (Wanasundara
and Shahidi, 1999; Mason et al., 1989).
The model of separation of oleic acid and saturated fatty acid was developed on the basis of the analysis of RSM. The ureatofatty acid ratio was the most important parameter for the yield percentage of NUCF, percentage of oleic acid and percentage of saturated fatty acid. The process may be help produce highly pure of linoleic acid and yield percentage of NUCF from an economic point of view, as well as being a promising measure for further utilization of agriculture products.
CONCLUSIONS
From these results, we demonstratd that urea/FFA ratio was the most important parameter for the yield percentage of NUCF and the percentage of linoleic acid. In addition, the urea method can be used when the goal is to obtain percentage of linoleic acid. The process may be help produce highly percentage linoleic acid from an econmic point of view, as well as being a promising measure for further utilization of agriculture products.
ACKNOWLEDGMENT
We would like to thank UKM and the Ministry of Science and Technology for research grant UKMGUPNBT0827113 and UKMOUPNBT29150/2010.