Extraction of rapeseed cake oil using subcritical R134a/butane: Process optimization and quality evaluation

Abstract The optimal extraction conditions of rapeseed cake oil using subcritical R134a/butane were established by response surface methodology. The quality of subcritical R134a/butane extraction oil (SRBEO) was compared with supercritical CO2 extraction oil (SCO2EO) and hexane extraction oil (HXEO). The results showed the highest extraction yield obtained by subcritical R134a/butane in the condition of R134a‐butane ratio of 1.5 kg/kg, at 45°C for 50 min. Compared with SCO2EO and HXEO, the extraction yield and β‐carotene content of SRBEO (87.76%, 357.21 μg/100g) were the highest. The content of phospholipids and canolol in SRBEO (3.01 mg/g, 118.51 mg/100 g) was higher than SCO2EO (not detected, 95.82 mg/100 g) and less than HXEO (25.78 mg/g, 131.85 mg/100 g). The tocopherols in SRBEO were equivalent to SCO2EO but phytosterol content of SRBEO (560.19 mg/100 g) was less than SCO2EO (591.40 mg/100 g). For fatty acids, the three extraction oils had slight difference. Thus, subcritical R134a/butane extraction appeared to be feasible for rapeseed cake oil extraction.


| INTRODUC TI ON
Rapeseed is a cruciferous seed, appearing round or oval. Rapeseed is planted wildly in China, and the planting area accounts for more than 40% in all oil crops and the oil yield about 30% in total oil yield.
At present, approximately 90% of rapeseed varieties are double-low (low erucic acid and low glucosinolates) oilseed rape (Yang et al., 2014). Double-low rapeseed oil is high-quality edible oil with low erucic acid content and reasonable fatty acid composition (Li et al., 2016). Rapeseed meal as by-product after oil extraction remains much oil, with approximately 12%-15%, which is unbenefited for manufacturers. Moreover, rapeseed cake is rich in tocopherols, polyphenols, phytosterols, and phospholipids (SzydlOwska-Czerniak, Amarowicz, & Szlyk, 22010), which possess important biological and chemical properties. Therefore, it is of great significance to the further development and utilization of oil and minor components in rapeseed cake.
Generally, residue oil in cake is extracted by hexane (Xu, Han, Zhou, Wu, & Ding, 2016). Hexane extraction has the advantages of large processing capacity and low production cost. However, the oil extraction contains much residual solvent. In addition, the high-temperature extraction and desolvation process will destroy the thermal sensitive substances in the oil. At the same time, it also brings environmental problems such as waste heat dissipation and emission (Campbell et al., 2011). Supercritical CO 2 extraction is also commonly used in the extraction of oils and fats, which has the advantages of low cost, nontoxicity, and environmental friendliness. The density, viscosity, and diffusion coefficient of CO 2 can be controlled by adjusting the temperature and pressure of SCO 2 , achieving the aim of improving the extraction rate or selectivity of oil and fat (Cvjetko et al., 2012;Jokić, Bijuk, Aladić, Bilić, & Maja, 2016;Li, Xia, Vazquez, & Song, 2017). However, the pressure of supercritical extraction is high and the equipment is expensive (Sahena et al., 2009), which have kept the technology from being widely used in industries. Subcritical fluid extraction (SFE) is a new separation technology developed after supercritical fluid extraction. It guarantees quality and productivity, and enables industrialization. Thus, for the rapeseed meal, SFE is a better choice to maintain oil quality (Russin, Boye, Arcand, & Rajumohamed, 2011). Several solvents can be used as subcritical fluids to extract plant oil or animal lipid, such as propane, butane, dimethyl ether (DEM), and 1,1,1,2-tetrafluoroethane (R134a). Teixeira, Ghazani, Corazza, Marangoni, and Ribani (2018) investigated the extraction of sapucaia oil with subcritical propane and obtained 93.38% oil with good oxidative stability. Sun et al. (2018) studied antarctic krill lipid using subcritical butane extraction. Results indicated that the lipid recovery rate was 81.2%, and most astaxanthin (248.4 mg/ kg) remained in product. Noteworthily, R134a presents low toxicity, inertia, nonflammability, and mild critical conditions. R134a is suitable and effective to extracting components such as procyanidins (Tan, Zhu, & Feng, 2018), polyphenols, β-carotene (Mohd Setapar, Khatoon, Ahmad, Yunus, & Zaini, 2014;Mustapa, Manan, Azizi, Norulaini, & Omar, 2009), and polycyclic aromatic hydrocarbons (Shi, Qiao, & Feng, 2016). Nevertheless, some solutes (such as fatty acids and triglycerides) habited a slight solubility even in hot R134a.
Modification by adding a small amount of n-butane or dimethyl ether as gaseous cosolvents with boiling points similar to that of R134a could improve the ability for nonpolar solutes. At present, there are few reports about the binary mixtures of R134a and butane as extraction solvents.
Compared with rapeseed cake, large quantities of solvents are required during the subcritical extraction of rapeseed, which increases the cost. Therefore, most of the oil can be extracted firstly by low-temperature pressing, and the residual oil in rapeseed cake can be obtained by subcritical extraction. In this experiment, we adopted the binary mixture of R134a and butane as an extractant, and Box-Behnken response surface method was used to design and optimize the subcritical extraction process parameters. Furthermore, the quality characteristics of the extraction oil were compared with supercritical CO 2 extraction oil and hexane extraction oil.

| Materials
Rapeseed cake (moisture content, 4.80% and residual oil content, 12.79%) was obtained by low-temperature pressing of Zhongshuang 11 double-low rapeseed in pilot plant of Oil Crop Research, Chinese Academy of Agricultural Sciences.

| Subcritical R134a/butane extraction
Supercritical fluid extraction was executed using a PLE-5L subcritical fluid extraction system (Henan Subcritical Extraction Biological Technology Co., Ltd). Rapeseed cake was placed in the extraction vessel and sealed with cover. At first, the vacuum pump was turned on to reduce the pressure in extraction vessel, separation vessel, and measuring vessel to −0.01 MPa. Subsequently, the static extraction ran. The ratio of material to solvent (the mixture of R134a and butane, and the mass ratio was 6:1, 6:2, 6:3, 6:4, 6:6, kg: kg) was 1:8 (w/v). Following that, the oily solution was introduced into the separation vessel. The R134a/butane was compressed by compressor, and the vacuum pump was opened for further recovery until the pressure of the three vessels to −0.01 MPa. Extract oil was collected from the separator vessel and centrifuged at 4,863 g for 10 min, and the upper oil was stored at 4°C for analysis.
Response surface methodology with Box-Behnken design (BBD) was used to optimize the process parameters of subcritical fluid extraction. The mass ratio of R134a and butane (A), extraction temperature (B), and extraction time (C) were independent variables, and extraction yield was dependent variable. According to single-factor experiment results, the extraction temperature is set at 25-45°C and the extraction time is 10-50 min. The codes and level values of independent variable are shown in Table 1. The experiment was carried out in a random order. Response surface plots were made by Design Expert (DE), and the relationship between the factors was analyzed.
The extraction yield was calculated as: where Y was the extraction yield (%), and m 0 and m 1 were masses of oil in raw materials and extracted cake (g), respectively.

| Hexane extraction
Hexane extraction of rapeseed cake was carried out in two steps.
150 g of cake and 1,200 ml of hexane solvent were homogenized and held for 50 min in water bath at 45°C. Then, the extracted oil was collected by solvent recover by vacuum evaporation. The extracted oil was collected and stored at 4°C for analysis.

| Residual oil content
The residual oil content in rapeseed cake was determined according to the ISO659.2009, which involves the gravimetric analysis with analytical grade petroleum ether in a Soxhlet apparatus (B-811; Buchi Labortechnik AG) for 8 hr.

| Fatty acid
Oil samples were methylated using sodium methoxide as reported by Teixeira et al. (2018) and determined by Agilent 7890A gas chromatography system (Agilent) equipped with HP-INNOWAX capillary column (30 m × 0.32 mm, 0.25 μm, Agilent Corp). The injector and detector temperatures were both 260°C. The flow rate of the carrier gas N 2 with a split ratio of 80:1 was set at 1.5 ml/min. The oven temperature was programmed as follows: helding the initial temperature of 210°C for 9 min, followed by an increase to 350°C at the rate of 20°C/min for 10 min. The injection volume was 1 μL.
The fatty acid composition of samples was analyzed by comparing the retention time of standard fatty acids. The relative ratio of fatty acids was calculated by normalization method.

| Phospholipids
Phospholipid content of oil samples was analyzed according to the method of Xu et al. (2016). 3 g of oil was accurately weighed and 0.1 g of zinc hydroxide was added to a crucible. The sample was carbonized before being completely ashed in a muffle furnace at 575°C for about 2 hr. The ash was dissolved in 10 ml of 50% HCl solution and heated to a slight boiling point and held for 5 min. The solution was cooled and filtered into a 100-mL volumetric flask. 50% (w/v) KOH solution was used to neutralize to turbidity. 50% (v/v) HCl solution was slowly added to completely dissolve the precipitation. The solution was diluted with distilled water to the scale. 10 ml of test solution was pipetted into a 50-mL colorimetric tube. 8 ml of hydrazine sulfate solution (0.0015%, w/v) and 2 ml of sodium molybdate solution (2.5%, w/v) were added. The tube was heated in a boiling water bath for 10 min and cooled to room temperature. The solution was diluted with distilled water to the scale and determined against a blank using a DU 800 UV/Visible Spectrophotometer (Beckman Coulter Inc., USA) at 650 nm.

| Phospholipid compositions
A 4 g (exact to 0.0001 g) of oil was weighed into a 100-mL test tube, and 50 ml of trichloromethane was added. 10 ml solution was transferred into an activated amino silica gel solid-phase extraction column. The column was then eluted with 2.0 ml of trichloromethane-isopropanol mixed solution and 3 ml of acetic acid-ether mixed solution, respectively. Phospholipids were eluted with 3 ml methanol. The eluent was evaporated by rotary evaporator at 45°C. The residue was dissolved in 10 ml hexane-isopropanol-acetic acid

| β-Carotene
β-Carotene content in oil was measured according to the method of Gimeno et al. (2000) with a few modifications. A 2 g of oil was sapon-

| Tocopherols
The content of tocopherols in oil was determined according to AOCS

| Phytosterols
The content of phytosterols in oils was quantified according to Xie's method (2017) with the same modifications. Oil sample (0.2 g) (accuracy 0.0001g) and 0.5 ml of 0.5 mg/ml 5α-cholestane(internal stand) were saponified with 10 ml of 2 mol/L KOH in ethanol at 60°C for 60 min. The hexane layer was dried with anhydrous sodium sulfate and silylated using 100 µl N, O-Bis (trimethylsilyl) trifluoroacetamide + 1% trimethylchlorosilane (BSTFA + TMCS) at 105°C for 15 min. Then, the mixture was dissolved in 1 ml hexane for further analysis by an Agilent 6890A gas chromatography system (Agilent) equipped with a DB-5HT column (30 m × 0.32 mm, 0.1 μm; Agilent). The flow rate of nitrogen (carrier gas) was 2.0 ml/min. The detector and injection temperature were 320°C. The oven temperature was programmed by keeping at 60°C for 1 min and increased to 310°C at a rate of 40°C/ min. The temperature of 310°C was held for 10 min. The injection volume was 1 μl, and the split ratio was 25:1.

| Statistical analysis
The results were reported as mean ± standard error in triplicates.
Statistical analysis was performed using a SPSS program (SPSS 20.0 for Windows, SPSS Inc.). Significances of difference between the means were determined by one-way analysis of variance (ANOVA), using Duncan's test at a significance level of 5% (p < .05).

| Effect of the ratio of R134a and butane on extraction yield
The impact of the ratio of R134a and butane on the extraction yield was conducted with five ratios (6:1, 6:2, 6:2, 6:4, and 6:6 (kg/kg)) and is shown in Figure 1. The results indicated that the extraction yield obviously depended on the ratio. It rose with an increase in the proportion of butane in the mixed solvent. Afterward, the yield tended to be stable when R134a and butane were in the mass ratio of 6:3 (kg/kg), that is, the ratio was 2 kg/kg. Furthermore, there were no significant differences in the yield between the ratio at 6:4 and 6:3 (kg/kg) (p > .05), nor between 6:4 and 6:6 (kg/kg) (p > .05).
Therefore, the ratio at 6:3 (kg/kg) was selected as the central point of extraction ratio in the optimization experiments.

| Optimization of subcritical R134a-n-butane extraction conditions
Response surface methodology (RSM) is an excellent experimental design and analysis method. It is widely used in the optimization of extraction process of oil and plant functional substances (Bezerra, Santelli, Oliveira, Villar, & Escaleira, 2008 (Solanki, Parikh, & Parikh, 2007).
The order was R134a-butane ratio > extraction time > extraction temperature.
The shape of contour plot can reflect the strength of interaction effect. Elliptical shape shows that two factors interact significantly, while circular shape shows the opposite (Porto, Porro, & Decorti, 2013). Contour and response surface plots are shown in Figure 2.
The interaction effects between R134a-butane ratio and extraction temperature, extraction time, and extraction temperature on oil extraction yield were significant, while that of R134a-butane ratio and extraction time had no significant difference. In order to verify the optimum conditions, the extraction parameters were corrected appropriately. In practice, the optimum conditions were adjusted as follows: R134a-butane ratio was 1.5 kg/kg, the extraction time was 50 min, and the extraction temperature was 45°C.
The corresponding yield was 87.76% and close to the predicted value.
The corresponding yield was 87.76% and close to the predicted value.
The results suggested that model was reliable and could be used to predict actual yield of rapeseed cake oil by subcritical extraction.

| Extraction yield of different methods
The extraction yield of oil from rapeseed cake obtained by supercritical CO 2 extraction (SCO 2 E), subcritical R134a/butane extrac-  Sun et al. (2018) also reported that the lipid extraction capacity of subcritical fluid was better than SCO 2 and HX, and SCO 2 and HX had no significant difference. Noteworthily, the working pressure of SCO 2 E was 28 MPa, which was much higher than that of SRBE. The severe operating conditions in this process necessitate high-duty equipment to withstand the high pressures and the investment is expensive (Sparks, Hernandez, Zappi, Blackwell, & Fleming, 2006). Compared with SRBE, HXE has less investment in disposable fixed asset. However, its steam consumption for desolvation is 11 times higher than that of subcritical process. Therefore, subcritical extraction was more economical in terms of long-term investment.

| Physicochemical characteristics of oils extracted by different methods
Physicochemical characteristics of rapeseed cake oils extracted by different methods are shown in Table 5. The color and viscosity of oils extracted by different methods revealed significant difference (p < .05). Red values of SCO 2 EO and SRBEO were 2.2 and 2.4, respectively, lower than that of nHXEO, while yellow values were higher than that of the latter, and so too of viscosity values. Acid value of nHXEO was 2.08 mg/g, which was 0.94 mg/g and 0.86 mg/g more than SCO 2 EO and SRBEO, respectively.
It was due to lipase produced by endogenous or external microorganisms in rapeseed cake that has stronger hydrolytic activity in the presence of trace water. Under high-pressure conditions of SCO 2 E and SRBE, much water was lost in rapeseed cake, and the lipase activity is inhibited.
Comparatively, lipase was active in nHXE, which led to the hydrolysis of some rapeseed oil, thus increasing the acid value. The peroxide value of

| Fatty acid composition of oils extracted by different methods
Fatty acid (FA) composition is a major index of oil quality. The fatty acid composition of the extraction oil is shown in TA B L E 6 Fatty acid composition of oils extracted by different methods to be the highest in oils (63.15%-63.84%) and it in SCO 2 EO was somewhat higher than others.
Oleic acid is an important monounsaturated fatty acid (MUFA) which exerts important effect on the prevention and treatment of cardiovascular diseases (Gu, Liu-X, Liu-H, Pang, & Qin, 2017).
Next in much were linoleic acid (C18:2; 19.22%-19.66%) and linolenic acid (C18:3; 7.81%-8.00%), which were essential 18-carbon unsaturated fatty acid (UFA). These fatty acids were of great significances to growth and development of human body. The results were in accordance with Jenab, Rezaei, and Emamdjomeh (2010), who also described that oleic acid was the primary fatty acid in rapeseed oil (over 60% of total fatty acids), followed by linoleic acid and linolenic acid at approximate percentages (17.2% and 8.0%, respectively).
Statistical analysis showed that it had significant differences in fatty acid composition (p < .05) but no significant difference in the total content of UFA (p > .05) among the three extraction oils.

| Phospholipids and Phospholipid compositions of oils extracted by different methods
The presence of phosphatides (PLs) is undesirable because they are harmful to color, flavor, foaming of the edible oil, which is not conducive to the consumption and sale. In addition, phospholipids often exist together with metal ions by chelation, which can act as catalysts for lipid oxidation, leading to lipid oxidation and rancidity (Goh, Khor, & Gee, 1982).
The results in Table 7 showed that there were significant differences in the PLs and PL composition content of the oil extracted by the three processes. PLs in SCO 2 EO were lower than the detection limit. The result concurred with Wan, Zhang, Huang, and Guo (2018), who reported extremely low phospholipids were detected in canola oil extracted by supercritical carbon dioxide. It was ascribed to CO 2 is a nonpolar molecule and only suitable for extracting nonpolar or weak polar substances (such as phospholipid) (Zaidul, Norulaini, & Omar, 2006

| Minor components of oils extracted by different methods
2,6-dimethoxy-4-vinylphenol (vinylsyringol), otherwise known as canolol, is one of the main phenols in rapeseed. It can be formed by decarboxylation of the sinapic acid and accounts for 70%-85% of the total free phenolic acid (Koski. Pekkarinen, Hopia, Kristiina, & Marina, 2003;Wakamatsu et al., 2005). Canolol has been proved to reveal good antioxidative and antimutagenic properties in body (Kuwahara et al., 2004). During rapeseed pressed, only a small proportion of phenol is transferred to the crude oil, and most of them remain in the cake. Therefore, rapeseed cake is a great resource of canolol. Canolol has certain polarity and is easy to combine with proteins or polysaccharides to form stable compounds by hydrogen bonds and hydrophobic bonds. Different extraction solvents would directly affect the break of chemical bond and the release of canolol.
The results of canolol content in oil extracted by three solvents with different polarities are obtained in

| β-Carotene, tocopherols, and phytosterols of oil extracted by different methods
β-Carotene protected lipids from free radical oxidation by reacting with peroxide free radicals, thus inhibiting the proliferation of lipid and promoting the termination of oxidation chain reaction (Sun, Pang, Ye, Lü, & Li, 2009). β-Carotene is also an effective quenching agent while inhibiting singlet oxygen . As shown in Table 8, the amount of β-carotene in SRBEO and HXEO was at an equal level (p > .05) and was significantly higher than that in SCO 2 EO (221.65 μg/100 g). This owed to the lower solubility and/or mass transfer rate of β-carotene in SCO 2. These results indicated that SRBEO had better quality than SCO 2 EO and HXEO due to a higher amount of β-carotene.
Tocopherols, also known as vitamin E, are powerful natural antioxidants that could effectively inhibit lipid oxidation. The tocopherols of the extraction oils by different methods are presented in Table 8. The tocopherols in rapeseed oil were mainly comprised of α-tocopherol (35.88-42.10 mg/100 g) and γ-tocopherol (46.84-54.68 mg/100 g), and the content of latter is comparatively higher.
The total tocopherols in SRBEO and SCO 2 EO had no significant difference (p > .05), which reached the maximum (>90 mg/100 g).
While the α-tocopherol of SRBEO was slightly higher than that of SCO 2 EO, the γ-tocopherol showed lower than SCO 2 EO. Therefore, we inferred that besides solvent polarity, other factors can also affect the extraction of tocopherols.
Phytosterol is a class of chemical compounds with cyclopentane and phenanthrene as its skeleton, which has physiological functions such as reducing the incidence of heart disease, anticancer, and immune regulation . As the results of phytosterols in oils are demonstrated in Table 8, it could be clearly known that the contents of total phytosterols varied from 1,047.80 mg/100 g to 1,179.46 mg/100 g, and the order was SCO 2 EO > SRBEO>HXEO.
Three phytosterols were detected in this study, and sitosterol was the dominant phytosterols (50.14% -51.07% of total phytosterols), followed by campesterol (35.07% -35.36%) and brassicasterol (13.69%-14.73%). These results were consistent with the results reported by Hamama, Bhardwaj, and Starner (2003), where the percent of them were 47.2%-50.6%, 29.8%-34.6%, and 10.1%-17.2%, respectively. Moreover, the contents of brassicasterol, campesterol, and sitosterol in extraction oils were in the same order as those of total phytosterols. The extraction ability of subcritical R134a/butane was weaker than that of SCO 2 , but stronger than that of hexane.

| CON CLUS IONS
Results of response surface methodology indicated that the peak performance of subcritical R134a/butane extraction for rapeseed cake oil was carried out with the parameters as follows: R134a-butane ratio of 1.5 kg/kg, extraction temperature of 45°C, and extraction time of 50 min. The extraction yield was significantly higher than SCO 2 E and HXE. Evaluating the quality of extraction oils found that the fatty acid composition in all oils was almost the same, but the contents of other minor components were significantly different. Subcritical R134a/butane extraction oil owned the highest tocopherols and β-carotene, higher canolol and phytosterols but less phospholipids. Overall, subcritical R134a/butane extraction was an efficient extraction method for rapeseed cake, and it is potential to be an alternative to supercritical CO 2 extraction and traditional hexane extraction.

ACK N OWLED G M ENTS
The authors greatly appreciated the supports of the National Key