Rice bran was obtained from Bernas Rice Mills, Co. Tiram, Selangor, Malaysia. The rice bran samples were stabilized before the extraction process for 2 min at 2450MHz, using a national microwave/convection oven (model NN-C2003S, Matsushita Electric Industrial Co., Japan) (Mariod et al., 2010).

Methanol, acetonitrile, dicholoromethane, isopropanol, acetic acid, chloroform, isooctane, p-anisidine, 1-butanol, butylated hydroxyanisole (BHA), hexane, thiobarbituric acid, ethanol, sodium thiosulfate, ether, and phenolphthalein were of analytical and HPLC grade obtained from Fisher Scientific (Loughborough, LE, UK) and Merck (Darmstad, Germany).

Chemicals such as potassium hydroxide, sodium chloride and sodium methoxide were obtained from Sigma - Aldrich (St. Louis, MO, USA). The standard for unsaponifiable matter (tocopherol) and neutral lipids were purchased from Sigma - Aldrich (St. Louis, MO, USA). The FAME reference standard was purchased from AccuStandard (New Haven, CT, USA). A pure standard of gamma-oryzanol (Standard compounds of oryzanols, cycloartenyl ferulate and 24-methylene cycloartanyl ferulate) was obtained from Sigma Chemical Co. Ltd, Japan.

The supercritical fluid extraction system was used (Thar Technologies, Pittsburgh, PA, USA, Laboratory Scale) with a rotary evaporator (Buchi, Flawil, Switzerland), universal and micro type centrifuge machine (Hettich Zentrifugen, Tuttlingen, Germany), ultrasonic water bath (HanShin, USA), and analytical balance (Shimadzu, Nakagyo Ku, Kyoto, Japan). Each component of lipids in rice bran oil was determined using Agilent Technologies HPLC series 1200 (Santa Clara, CA, USA).

Supercritical fluid high oryzanol rice bran (SFE HO) and supercritical fluid low oryzanol rice bran (SFE LO) oils were extracted using SFE with different pressures and temperatures. In brief, carbon dioxide (CO₂) from a cylinder was pressurized and then fed into the solvent reservoir. Pressurized CO₂ becomes the supercritical fluid by preheating in an equipped air-bath. After rice bran (180 g) was placed in the extractor, the supercritical fluid flowed into the extractor and mixed with rice bran. The extractor was connected to the collector vessel by a micro-metering valve. The temperature and pressure were controlled to obtain the SFE oil. After that the mixed solution of RBO and supercritical fluid from the extractor is depressurized and the RBO was collected (Bruhl and Matthaus, 1999). To obtain the HO from SFE the pressure was 600 bars and the temperature was 60 °C. Meanwhile, to obtain the LO the pressure was 200 bars and temperature was 40 °C. The flow rate and duration of time to extract HO and LO was similar with 25g·min^-1 of flow rate and 150 min duration of time.

In brief, 100 grams of ground rice bran were weighed and equally divided into 4 extraction thimbles. Each thimble was then transferred into a soxhlet extractor. Approximately, 300 mL of n-hexane were added to the extraction flask, which were connected to the extractor and condenser. After extraction began, the solvent flow rate was manually adjusted to 7 min·cycle^-1. Finally, the extraction process was terminated after 100 cycles. After extraction was complete, n-hexane was removed at 50 °C under reduced pressure using a rotary evaporator. The flasks were placed in a desiccator chamber for 1 hour. The obtained soxhlet oil (SOX) was weighed and the yield was calculated.

Lipids extracted by SFE and Soxhlet were fractionated on an Solid Phase Extraction (SPE) system using a pre-packed 1-g silica gel cartridge column (Sigma - Aldrich St. Louis, MO, USA). Although vacuum could be applied to increase the flow rate of the eluent, generally gravity alone was sufficient to provide adequate flow. Initially, the cartridges were conditioned with 5 mL of hexane, followed by 5 mL of CHCl₃. The neutral lipid (NL), glycolipid (GL), and phospholipid (PL) fractions were successively eluted with 10 mL of CHCl₃ and acetone (4:1, v/v), 15 mL of acetone and methanol (9:1, v/v), and 10 mL of methanol. The purity of the fractions was checked by TLC. Solvents from the fractions were rotary evaporated, transferred to recovery vials using the eluting solvents, and evaporated to dryness at 37 °C. The recovery of the lipids was 100%. The amount of polar lipids was the sum of GL and PL (Hubbard et al., 2004).

Unsaponifiable and wax contents were measured following AOCS Da 11-42, and AOCS Ch 8-02 Methods, respectively (AOCS,1998).

The fatty acid compositions of the RBOs were determined by GLC. The oils were converted to their corresponding methyl esters according to the AOCS Official Methods (1998). BF₃/methanol was used for methylation. The GLC analysis of the fatty acid methyl esters FAME was performed using a Hewlett-Packard HP-5890 Series II gas chromatograph equipped with an Ultra 2 (25 m×0.32 mm×0.5 µm film thickness of 5% biphenyl and 95% dimethyl polysiloxane; Hewlett-Packard, Waldron, Germany) equipped with a capillary column, a split injector (split ratio 88:1), and an FID. The column temperature program was 5 min at 150 °C, 10 °C·min^-1 to 275 °C, and 10 min at 275 °C. The injector and detector temperatures were 250 and 280 °C, respectively. The carrier gas was nitrogen at a flow rate of 1.6 mL·min^-1. Air and hydrogen flow rates were 460 and 33 mL·min^-1, respectively. The peaks of fatty acids were identified by comparing their retention times with those of a mixture of standard fatty acid methyl esters (Sigma Chemicals Co., Deisenhofen, Germany). Each fatty acid methyl ester sample was analyzed in duplicate.

Gamma oryzanol was quantified using a reversed-phase HPLC. A C18 column and a diode-array ultraviolet-visible detector (Hewlett-Packard, San Fernando, CA) were employed in the HPLC system. The mobile phase consists of methanol, acetonitrile, dichloromethane, and acetic acid (50:44:3:3) by volume. The flow rate was controlled at 1.4 mL·min^-1. The analytes were detected at 330 nm. The samples were diluted with the mobile phase and analyzed. Peak identification was based on the comparison of retention time (RT) values with the authentic standards of cycloartenyl ferulate, 24-methylene cycloartanyl ferulate and a mixture of oryzanol standards. The total oryzanols were quantified based upon the peak area of 24-methylene cycloartanyl ferulate, the major oryzanol component in rice bran. The concentrations of individual components of γ-oryzanol were calculated and the concentration of γ-oryzanol was obtained by adding all individual components (Shahid et al., 2005, Renuka Devi and Arumughan, 2007).

Tocopherols were measured by a reversed-phase HPLC system connected to a fluorescent detector at Ex = 298 nm and Em = 328 nm. A 2.1 mm narrow-bore analytical column packed with 5-m ODS (C18) hypersil silica used for the separation of tocols. The mobile phase acetonitrile/methanol/isopropanol/water (45:45:5:5 by volume), was programed to acetonitrile/methanol/isopropanol (50:45:5) for 10 min. These conditions were maintained for 15 min before returning to the original conditions. The total HPLC run time was 26 min. A fine separation was achieved for tocopherols but bands were broad for tocotrienols. The samples were quantified by comparing the retention times/peak areas with those of the obtained standards. The total tocopherol and tocotrienol contents were determined by summing up the Α, Β, γ and δ isomers (Shahid et al., 2005).

For the oxidative stability studies, the antioxidant BHA (0.02%) was added to separate the samples of SFE HO, SFE LO, and SOX based on the concentration of antioxidant used and represents the maximum added antioxidant levels used in foods (0.02%). The antioxidant (0.02 g) was dissolved in 10 mL of ethanol and then added to 100 g of RBOs following the procedure of Mariod et al. (2006). The ethanol was then evaporated at 45 °C for 15 min to avoid over-oxidation and to enable the antioxidants to thoroughly mix with the sample. Control samples contained ethanol without the addition of antioxidants. The oxidation process was accelerated using the procedure of Mariod et al. (2006) in which all the experimental RBO samples were placed in an oven in an ultrasonic water bath (Bandelin electronic, Berlin, Germany) at 70 °C with a stream of air (flow 1 L·h^-1). Subsamples of the RBO were removed from the oven at 0, 7, 14, 21, 35 and 42 days and stored at 4 °C overnight in a refrigerator until tested for oxidation. The oxidation was evaluated by determining the peroxide value, conjugated dienes, p-anisidine value and thiobarbituric acid reactive substances.

Determination of Peroxide Value. Peroxide value was determined following the AOCS official method Cd 8b-90 (AOCS, 1998).

Determination of p-anisidine Value. p-Anisidine value of the oil was analyzed according to the method of AOCS, (1998).

Determination of Conjugate Dienes. The conjugated dienes were determined following the AOCS official method Ti 1a-64 (1998). 10 mL of each sample were mixed with 10 mL of isooctane and the absorbance was read at 234 and 270 nm using cells of 1 cm. Pure isooctane was used as reference. The conjugated dienes were calculated according to the following equation:

Where CD = conjugated diene, A = absorbance at 234 nm, C = concentration (g·100 mL^-1) and I = path length (cm).

Determination of Thiobarbituric Acid-Reactive Substances (TBARS). TBARS were determined following the method of Shahidi et al. (2003), where each sample (200 mg) was weighed into 25 ml volumetric flask and then made up to the mark with 1-butanol (ACS grade) and mixed thoroughly. A 5.0 mL portion of this solution was transferred into a dry test tube, and a 5.0 mL of fresh TBA reagent (200 mg TBA in 100 mL 1-butanol) were added to it. The tube was placed in a 95 °C water bath for 120 min. The tube was cooled to room temperature and the absorbance of the solution was read at 532 nm. The TBAR value was calculated as follows:

Where A = absorbance at 532 nm and m = sample weight.

Free fatty acids as oleic acid percentages in the oil samples, were determined using an alkali titration method according to AOCS (1998).

PV is a measurement of the amount of peroxides formed in fats and oils through autoxidation and oxidation processes (Stier 2006). Generally, it is a measurement of the degree of initial oxidation of fats and oils. Measuring the peroxide value is well-established method for the determination of primary oxidation products in fats (Kulås and Ackman 2001). The peroxide values of SFE HO, SFA LO and SOX oils with and without BHA during storage for 42 days are shown in Table 3. Since RBO contains relatively high levels of unsaturated fatty acids (more than 77%), oxidation can occur during the extraction and during post-harvest processing and storage. The addition of synthetic butylated hydroxyanisol (BHA) to these oils protects them against oxidation. Table 3 shows that the initial PVs of SOX, SFE HO, SFE LO are 1.8, 0.75 and 0.5 meq O₂·kg^-1 oil, respectively, the initial PV of SOX was higher than SFE, after storage for 42 days under successive heating at 70 °C. The PV changed during storage and increased to 14.0, 11.25 and 10.25 meq O₂·kg^-1 oil, respectively. After the addition of BHA the PVs were 13.3, 10.5 and 8.25 meq O₂·kg^-1 oil respectively, indicating that BHA keeps the oil from oxidation. These results showed that the oxidative stability of RBO could be protected by extracting the oil using the SFE technique which produces stable oils.

The free fatty acids (%) of different RBOs during storage are shown in Table 4. The FFA level of SOX (4.125±0.06) was significantly higher than (3.1 and 3.57) that of the SFE HO, SFE LO samples at zero time, respectively. The FFA level increased to 23.04, 21.99 and 23.0 in SOX, SFE HO and SFE LO during 42 days of storage, respectively. These amounts decreased after the addition of BHA. The oils extracted by SFE showed low amounts of% FFAs. These oils seem to be more stable than the oils extracted using Soxhlet method. The presence of natural antioxidants such as gamma oryzanol and tocopherol besides a high amount of unsaponifiable matters in the SFE oil can explain its high stability compared to the SOX oil.

The measurement of conjugated dienes is a useful index of the early stages of peroxidation in studies with pure lipids and isolated lipoproteins (Dekkers et al., 1996). Table 5 shows the conjugated dienes of the different RBOs extracted using the two different methods. In this table it can be seen that conjugated diene levels increased rapidly in RBOs between 0 and 42 days at both 234 and 270 nm. SFE HO and SFE LO contained lower amounts of CD (0.55 and 0.675) than that of SOX (0.921) at 234 nm whose amounts were 0.602, 0.603 and 0.713 at 270 nm, respectively. After storage for 42 days the conjugated dienes were significantly increased in SOX, SFE HO, and SFE LO to 1.143, 1.101 and 1.13 respectively at 234nm and 1.438, 1.255 and 1.27 at 270 nm, respectively. From Table 5 it is clear that SFE HO and SFE LO showed better stability than SOX under successive heating for 42 days and the addition of BHA decreased conjugated diene levels in all samples.

The change in TBAR levels (µmol·g^-1)² in the different RBOs during storage is shown in Table 6. The TBAR level increased significantly (P < 0.001) from 0.37 to 1.32 µmol·g^-1 between day 0 and 42 days in SOX. These levels increased from 0.20 to 0.88 in SFE HO and from 0.32 to 0.86 (µmol·g^-1)² in SFE LO, respectively. These results show that the oils extracted by SFE have better stability than SOX. The addition of BHA antioxidants to all samples was effective in reducing the formation of secondary oxidation products, and thus increasing the oxidative stability.

The p-anisidine values (p-AV) of rice bran oil extracted by the SFE and Soxhlet methods during storage are shown in Table 7. The successive heating of RBOs (at 70 °C for 42 days) with and without BHA leads to autoxidation and formation of primary products that decomposed readily and formed aldehydes, ketones and alcohols as secondary products. As p-AV is a more meaningful test for the assessment of the oil's quality during heating than the PV, it was used here because it measures the secondary products of oxidation reactions (Mariod et al., 2010). Using SFE as a method of extraction gave products rich in natural antioxidants that inhibited the formation of the secondary products in comparison with the Soxhlet method; the amount of secondary products formed seems to be smaller than that formed in the SOX samples (Table 7). The P-AVs of SOX, SFE HO and SFE LO at day zero without and with BHA were 0.107, 0.094, 0.039, 0.03, 0.042 and 0.032 respectively; these amounts increased rapidly to 0.489, 0.476, 0.121, 0.11, 0.10, and 0.10 respectively.

Table 8 illustrates the decrease in the amounts of the total tocopherol contents (mg·g^-1) of SOX and SFE HO and SFE LO with increasing storage time. Within 42 days more than 46% of the initial amount of tocopherols was degraded in SOX, while in SFE HO and SFE LO only 31% and 42% of the initial tocopherols was degraded, respectively. The addition of synthetic BHA normally retards this degradation in all samples. In spite of the higher amount of total tocopherols at the beginning of the storage period (zero day), the degradation of tocopherols occurred significantly (P<0.05) faster in SOX than in SFE, where part of the initial tocopherols was still available after 42 days of storage. It is possible that, in addition to oryzanol and tocopherols, RBO contains other anti-oxidative active components that retard the degradation of the tocopherols and increase the shelf-life of the oil. Previous research has shown the degradation of tocopherols in the storage of edible oils (Chotimarkorn et al., 2005, Player et al., 2006) due to autoxidation.