Characterization of Coconut Oil Fractions Obtained from Solvent

Home , Lauric acid, Stearin

Characterization of Coconut Oil Fractions Obtained from Solvent Fractionation Using Acetone Sopark Sonwai＊ , Poonyawee Rungprasertphol, Nantinee Nantipipat, Satinee Tungvongcharoan and Nantikan Laiyangkoon Department of Food Technology, Faculty of Engineering and Industrial Technology, Silpakorn University, THAILAND

Abstract: This work was aimed to study the solvent fraction of coconut oil (CNO). The fatty acid and triacylglycerol compositions, solid fat content (SFC) and the crystallization properties of CNO and its solid and liquid fractions obtained from fractionation at different conditions were investigated using various techniques. CNO was dissolved in acetone (1:1 w/v) and left to crystallize isothermally at 10℃ for 0.5, 1 and 2 h and at 12℃ for 2, 3 and 6 h. The solid fractions contained significantly lower contents of saturated fatty acids of ≤ 10 carbon atoms but considerably higher contents of saturated fatty acids with > 12 carbon atoms with respect to those of CNO and the liquid fractions. They also contained higher contents of high-melting triacylglycerol species with carbon number ≥ 38. Because of this, the DSC crystallization onset temperatures and the crystallization peak temperatures of the solid fractions were higher than CNO and the liquid fractions. The SFC values of the solid fractions were significantly higher than CNO at all measuring temperatures before reaching 0% just below the body temperature with the fraction obtained at 12℃ for 2 h exhibiting the highest SFC. On the contrary, the SFC values of the liquid fractions were lower than CNO. The crystallization duration exhibited strong influence on the solid fractions. There was no effect on the crystal polymorphic structure possibly because CNO has β’-2 as a stable polymorph. The enhanced SFC of the solid fractions would allow them to find use in food applications where a specific melting temperature is desired such as sophisticated confectionery fats, and the decreased SFC of the liquid fractions would provide them with a higher cold stability which would be useful during extended storage time.

Key words: coconut oil, fractionation, crystallization, solid fat content, polymorphism

1 Introduction food industry5）. Therefore, these fats are usually modified Coconut oil（CNO）is one of the most important oil crops via processes such as fractionation and hydrogenation to in tropical regions. It is an edible oil obtained from matured improve their quality. Due to its low melting temperature coconuts. CNO is thought to be beneficial to health and lack of plasticity and hardness, CNO cannot be used in because it contains high amount of medium-chain triacylg- food products such as margarines, chocolate coatings and lycerols（TAGs）, which are composed mainly of saturated coffee whiteners without modification. fatty acids with chain length from 6 to 12 carbon atoms1）. Fractionation or fractional crystallization is a common The main fatty acid components in CNO are lauric acid technique for fat and oil modification that allows the sepa- （42.6％）and myristic acid（21％）2）. Medium-chain TAGs ration of TAGs in fractions with different melting ranges are hydrolyzed faster and more completely than long-chain and physical properties that are suitable for a variety of TAGs3）, hence, they are used immediately as energy food products6）. Fractionation consists of a controlled crys- sources in the body and avoid being stored in the adipose tallization in bulk crystallizers followed by a physical sepa- tissue2）. Moreover, it has been shown that medium-chain ration of the liquid fraction（olein）from the crystalline TAGs may reduce the incorporation and storage of dietary fraction（stearin）7）. For the fractionation of natural fats and fats and oil in adipose tissue4）. However, fats rich in satu- oils, which are complex mixtures of numerous TAG rated fatty acids usually contain a wider variety of TAG species, olein fraction is enriched in triunsaturated and species that are associated with broader melting ranges monosaturated TAGs hence exhibiting a higher cold stabili- and which are not appropriate for many applications in the ty. On the other hand, disaturated and trisaturated TAGs

＊Correspondence to: Sopark Sonwai, Department of Food Technology, Faculty of Engineering and Industrial Technology, Silpakorn University, THAILAND E-mail: [email protected] Accepted April 21, 2017 (received for review December 1, 2016) Journal of Oleo Science ISSN 1345-8957 print / ISSN 1347-3352 online http://www.jstage.jst.go.jp/browse/jos/ http://mc.manusriptcentral.com/jjocs

951 S. Sonwai, P. Rungprasertphol, N. Nantipipat et al.

concentrate in the stearin fraction giving it a specific controlled double jacket crystallizer. Fractionation was melting behavior suitable for a specific application. The carried out at 10℃ for 0.5, 1 and 2 h and at 12℃ for 2, 3 basis of such fractionation resides in the solubility of the and 6 h with a constant agitation rate of 25 rpm. After frac- TAG species in the liquid phase at controlled temperatures, tionation, samples were separated by vacuum filtration which is dependent on the molecular weight and degree of using a Whatman No.1 filter paper to obtain the liquid and unsaturation8）. Industrial-scale fractionation is classified solid fractions of CNO. Acetone was removed from the into three main groups, such as dry, detergent, and solvent liquid fractions using a rotary vacuum evaporator with fractionation9）. The main advantage of solvent fractionation moderate heating. Then, remaining solvent in all fractions is the great separation efficiency and the enhanced yield of was further removed using nitrogen flushing. All samples the targeted phase compared to the other methods10）. were kept at －18℃ prior to analysis. Different abbrevia- The aim of this study was to characterize TAG and fatty tions were designated to all solid and liquid fractions ac- acid compositions, solid fat content and crystallization cording to their crystallization temperature and duration as properties of the solid and liquid fractions of CNO obtained shown in Table 1. The yields of the solid fractions obtained from solvent-fractionation with different conditions（crys- at different fractionation conditions are also given in Table tallization temperature and cluration）. This was hoped to 1. create different fractions of CNO that would meet specific requirements for different food applications. 2.3 Fatty acid compositions CNO and its solid and liquid fractions were converted into fatty acid methyl esters using AOAC official method 969.3311）. The fatty acid methyl esters analysis was per- 2 Materials and Methods formed in a Shimadzu GC with flame ionization detector 2.1 Materials （GC-FID）. The system had an VertiBondTM wax capillary Refined, bleached and deodorized CNO was purchased column（50 m long, 0.25 mm internal diameter and 0.20 μm from Katevanich Industry Co., Ltd.（Nakhonpathom, Thai- film thickness）. Compound identification was carried out land）and used without further treatment. Its free fatty acid using external standards of fatty acids methyl esters. content（as oleic acid）was ＜0.02％. The standard fatty Helium was used as a carrier gas with a flow rate of 1 mL/ acid methyl esters for fatty acid analysis using gas chroma- min and with a controlled initial pressure of 93.2 kPa at

tography（GC）were purchased from AccuStandard, Inc. 120℃. N2 and air were makeup gases. The injection tem- （USA）. The standards for TAG analysis using high perfor- perature was 210℃, and the oven temperature program mance liquid chromatography（HPLC）were purchased from was holding at 120℃ for 3 min before increasing at a rate Sigma Chemical Co.（St. Louis, MO, USA）. Acetone and of 10℃/min to 220℃, holding at this temperature for 30 acetonitrile were of HPLC grade from Burdick and Jackson min, increasing at a rate of 5℃/min to 240℃, followed by （Muskegon, MI, USA）. All other chemicals and solvents holding at 240℃ for 30 min. The split ratio was 100:1, the used were obtained commercially and were of the highest injection volume was 1 μL, and the detector temperature purity available. was 280℃. After the samples were analyzed, their chromatograms were acquired and the fatty acid contents were 2.2 Fractionation of coconut oil calculated based on percentage of peak area. A mass of 100 g of CNO was melted at 80℃ for 10 min and cooled to 50℃ before being mixed with 100 ml of 2.4 Triacylglycerol compositions warm acetone（50℃）and transferred to a temperature- The TAG compositions of the mixtures were determined

Table 1 Abbreviations for all solid and liquid fractions and the yield of all solid fractions obtained from fractionation at different conditions.

Abbreviations Fractionation conditions Yield of solid fractions Solid fractions Liquid fractions ℃ Temperature ( ) Duration (h) (％ wt.) S10C0.5h L10C0.5h 0.5 17.9±1.9 S10C1h L10C1h 10 1 33.1±3.5 S10C2h L10C2h 2 47.2±3.7 S12C2h L12C2h 2 11.0±0.7 S12C3h L12C3h 12 3 13.6±0.8 S12C6h L12C6h 6 18.4±1.7

952 J. Oleo Sci. 66, (9) 951-961 (2017) Solvent fractionation of coconut oil

by HPLC（Shimadzu LC-20 AD, Shimadzu Corp, Kyoto, tinuously recorded for 6 h. Japan）with system controller CBM-20A and diode array detector SPD-M20A. Two C-18 columns（Inertsil ODS-3; 4.6 2.8 Crystal morphology ×250 mm; 5 μm particle size; by GL Sciences Inc., Japan） Crystal network microstructure of the all samples was were used in series. The mobile phase consisted of acetone observed by polarized light microscopy（PLM）（ Olympus and acetonitrile（70:30, v/v）with a flow rate of 0.72 mL/min. BX51, Olympus Optical Co., Ltd., Tokyo, Japan）equipped The column temperature was set at 35℃ with a column with a digital camera（Olympus C- 7070, Olympus Optical heater（Shimadzu CTO-10AS column oven）. The injection Co., Ltd., Tokyo, Japan）. All fat samples were melted at volume was 20 μL. Peak identification for TAGs were per- 80℃ for 10 min to totally eliminate the memory effect. formed by comparing their retention times with those of Twenty microliters of each molten sample was placed on a authentic standards and/or by comparison with the HPLC glass slide, which was heated to 80℃ prior use, and pattern of CNO reported by Marikkar et al.12）. Calculation covered by a cover slip14）. Then, the samples were trans- of TAG compositions was based on area percents from the ferred to and stored in a temperature-controlled cabinet, chromatograms. the temperature of which was maintained at 18±0.2℃, for 24 h. A 10×lens was employed to image the gray scale 2.5 Crystallization thermograms photographs of the fat crystals. The crystallization thermograms of CNO and its solid and liquid fractions were determined with a Perkin-Elmer 2.9 Polymorphism differential scanning calorimeter（DSC）（ model DSC 8000, Temperature/time-dependent small and wide-angle x-ray PerkinElmer Co., Norwalk, CT, USA）following AOCS pro- diffraction（SAXD and WAXD）of CNO and the CNO-SE cedure Cj 1-9413）. The instrument was calibrated with blends was investigated using a Hecus S3-MICRO high flux indium（mp 156.6℃）as a reference standard. A fat sample system（Hecus, Graz, Austria）. The unit uses a 50 W, high of 3-5 mg was placed in an aluminum pan（20 μL capacity） brilliance GeniX microfocus source and customized and hermetically sealed with a sample press. An empty pan FOX-3D multi-layer point focusing optics（Xenocs SA, served as reference and was used to obtain baseline set- Grenoble, France）with a 100×250 μm2 FWHM（vertical× tings. Samples were heated from room temperature to horizontal）at focus. The x-ray beam was generated by a 50

80℃ and held for 10 min and then cooled at 5℃/min to kV, 1 mA Cu Kα anode. The sample-detector distance was －60℃. The crystallization onset temperatures and the ～280 mm（SAXD）and ～300 mm（WAXD）. Spectra were crystallization peak temperature were determined with the captured with dedicated Hecus 1-D position-sensitive de- built-in DSC software. tectors（model PSD-50M）. X-ray ranges were 2000 Å＞d＞ 11 Å and 4.9＞d＞3.3 Å for the SAXD and WAXD regions. 2.6 Solid fat content For sample preparation, ～20 μL of melted sample were Changes in the solid fat content（SFC）as a function of placed in 1.5 mm O.D. quartz capillaries（Charles Supper temperature between 15℃ and 35℃ and the melting be- Company, Inc., Natick, MA, USA）using a long needle and havior of CNO and its solid and liquid fractions were deter- syringe. The sample was then transferred to the unit’s mined by pulse-nuclear magnetic resonance（p-NMR）spec- sample port and kept at 80℃ for 10 min and then cooled trometer（Minispec-mq20, BRUKER, Karlsruhe, Germany） from 80 to 20℃ at 5℃/min followed by cooling from 20 to following AOCS method Cd 16-8113）. Briefly, the fat samples －5℃ at 1℃/min after which it was being hold at －5℃ for were heated to 80℃ for at least 10 min, cooled to 0℃ and 15 min15）. Calibration was performed using silver behenate held for 60 min, then heated to the measuring temperature for the SAXD region and bromobenzoic acid and and held for 30-35 min prior to measurement. β-tripalmitin for the WAXD region.

2.7 Isothermal crystallization 2.10 Statistical analysis Changes in SFC during isothermal crystallization at 17℃ All experiments were performed either in duplicate or for CNO and its solid fractions and at 15℃ for CNO and its triplicate and the results were analyzed by ANOVA with liquid fractions were determined using the p-NMR. Samples Least Significant Difference at a 95％ confidence interval（p were pipetted into 10 mm O.D. p-NMR tubes to a height of ＜0.05）. ～4 cm, tempered at 80℃ for 10 min in a waterbath and then transferred to a cooling bath set to 15 and 13℃ for the crystallization at 17 and 15℃, respectively. Once the sample temperature reached 0.5℃ above the crystalliza- 3 Results & Discussion tion temperature, the tube was removed from the cooling 3.1 Yield and fatty acid compositions bath, rapidly wiped dry and put into the p-NMR sample The yields of all the solid fractions obtained after frac- port set at the crystallization temperature. SFC was con- tionations are given in Table 1 where it can be seen that

953 J. Oleo Sci. 66, (9) 951-961 (2017) S. Sonwai, P. Rungprasertphol, N. Nantipipat et al.

the yields increased as the crystallization duration in- propanol. The solid fractions obtained from 12℃ exhibited

creased at both crystallization temperatures. The results the same trend as 10℃ except that the content of C12:0 in corresponded with a previous work by Kang et al.10）who all fractions from 12℃ was significantly lower than CNO（p reported the increase in the yield of the solid fraction of ＜0.05）. The increase in the contents of long-chain fatty palm stearin upon increasing fractionation time. When acids in the solid fraction from that of the original fat has comparing the yields of the solid fractions obtained at the been reported in a number of works concerning fraction- same crystallization time（2 h）but different temperatures ation of fats12, 18, 19）.

（10 and 12℃）, it was found that the fraction obtained at For the liquid fractions obtained at 10 and 12℃, C8:0, C10:0

higher temperature（S12C2h）gave lower yield than the one and C18:1 did not experience any significant change with obtained at lower temperature（S10C2h）. crystallization time but all was higher than in CNO（p＜ Table 2 exhibits the fatty acid compositions of CNO and 0.05）（ data not shown）. This is opposite to what was ob-

its solid fractions. CNO contained mainly lauric acid（C12:0） served with the solid fractions obtained from the same

（～48％）and myristic acid（C14:0）（～ 18％）. The result was temperatures where the contents of these fatty acids were 12, 14, 16） in good agreement with previous reports . S10C0.5h significantly less than CNO. The content of C12:0 did not showed significantly less contents of saturated fatty acids differ from that of CNO in almost all fractions（p＞0.05）.

of ≤ 10 carbon atoms（caproic（C6:0）, caprylic（C8:0）and On the other hand, the contents of C14:0 and C16:0 were less

capric（C10:0）acids）and less contents of unsaturated fatty than those of CNO in all fractions（p＜0.05）but did not

acids（oleic（C18:1）and linoleic（C18:2）acids）than CNO（p＜ change with increasing crystallization time（p＞0.05）. 0.05）. The contents of these fatty acids increased as the Overall, the distribution of individual fatty acids in the crystallization time increased but did not reach the same liquids fractions did not change from CNO as much as in

level as those in CNO. For example, the content of C10:0 de- the solid fractions. This was in good agreement with a pre- creased from 5.8％ in CNO to 4.5％ in S10C0.5h and as the vious report by Marikkar et al.12）who studied the effect of crystallization time increased to 1 and 2 h, it increased to fractional crystallization on the composition and the 4.7 and 4.9％ in S10C1h and S10C2h, respectively（p＜ thermal behavior of CNO from Sri Lanka.

0.05）. The contents of lauric acid（C12:0）in all solid fractions obtained at 10℃ did not changed significantly from CNO（p 3.2 Triacylglycerol compositions ＞0.05）. On the contrary, the saturated fatty acids with＞ Triacylglycerol（TAG）compositions of CNO and its solid 12 carbon atoms exhibited the opposite behavior. The con- and liquid fractions are given in Table 3. CNO contained tents of these fatty acids were significantly higher than mainly LaLaLa（22.2％）, CLaLa（18.6％）, LaLaM（15.5％）, CNO at 0.5 h of crystallization（p＜0.05）before decreasing CCLa（14.2％）and LaMM（10.0％）, where La, C and M are as the crystallization time increased to 1 and 2 h. The in- lauric, capric and myristic acids, respectively. The results crease in saturated fatty acids with＞12 carbon atoms in were consistent with previous reports12, 20）. For the solid the solid fraction of CNO was observed previously by Kar fractions, CCLa, CLaLa and LaLaLa, which are low melting et al.17）who studied solvent fractionation of CNO using iso- TAG components in CNO12）with carbon number ≤ 36, expe-

Table 2 Fatty acid compositions of CNO and its solid and liquid fractions. Fatty acid content (％) Fatty acid CNO 10℃ 12℃ S10C0.5h S10C1h S10C2h S12C2h S12C3h S12C6h d b b c a b b Caproic acid (C6:0) 0.5±0.0 0.3±0.0 0.3±0.0 0.4±0.0 0.2±0.0 0.3±0.0 0.3±0.0 f c d e a b d Caprylic acid (C8:0) 7.2±0.2 4.6±0.0 5.0±0.1 5.3±0.1 3.5±0.0 4.2±0.0 4.9±0.1 f c d e a b cd Capric acid (C10:0) 5.8±0.1 4.5±0.0 4.7±0.1 4.9±0.1 3.9±0.0 4.3±0.0 4.6±0.0 d b bc cd a a b Lauric acid (C12:0) 48.2±0.2 47.0±0.3 47.4±0.1 47.9±0.4 45.5±0.4 45.8±0.3 46.7±0.5 a c b b e d c Myristic acid (C14:0) 18.2±0.3 23.4±0.2 22.3±0.1 22.1±0.3 25.5±0.1 24.0±0.1 23.0±0.0 a c b b e d b Palmitic acid (C16:0) 9.1±0.2 12.3±0.3 11.4±0.1 11.2±0.2 13.9±0.1 13.0±0.1 11.6±0.2 a cd bc b e de bc Stearic acid (C18:0) 2.6±0.0 3.2±0.1 3.1±0.0 3.0±0.1 3.5±0.1 3.3±0.1 3.1±0.1 c ab b ab a ab b Oleic acid (C18:1) 6.9±0.3 4.1±0.3 4.9±0.2 4.2±1.0 3.4±0.4 4.4±0.2 4.8±0.3 b a ab ab a ab ab Linoleic acid (C18:2) 1.4±0.3 0.5±0.6 1.0±0.1 1.0±0.2 0.6±0.1 0.8±0.2 0.9±0.1 All data are mean values±standard deviations of triplicate measurements. Values with the same letter in each row are not significantly different (p＞0.05).

954 J. Oleo Sci. 66, (9) 951-961 (2017) Solvent fractionation of coconut oil

Table 3 Triacylglycerol compositions of CNO and its solid and liquid fractions. Triacylglycerol content (％) Triacylglycerol CNO 10℃ 12℃ S10C0.5h S10C1h S10C2h S12C2h S12C3h S12C6h CCLa 14.2±0.1e 8.7±0.1c 8.9±0.3c 10.3±0.1d 6.2±0.2a 7.8±0.0b 9.0±0.1c CLaLa 18.6±0.4e 12.9±0.3c 13.5±0.3c 14.7±0.3d 10.2±0.4a 12.0±0.1b 13.0±0.1c LaLaLa 22.2±0.3d 18.5±0.4bc 18.7±0.6c 19.1±0.1c 16.7±0.3a 17.6±0.4ab 18.2±0.9bc LaLaM 15.5±0.3a 19.2±1.2cd 19.8±0.1cd 18.8±0.4bc 20.6±1.0d 20.5±0.1d 19.4±0.5cd LaLaO 1.9±0.0ab 1.3±0.4a 1.4±0.2a 1.9±0.5ab 1.4±0.5a 2.1±0.3b 1.4±0.4a LaMM 10.0±0.1a 17.4±1.0c 15.6±1.4b 14.8±0.3b 19.7±0.8e 17.4±0.2d 16.4±0.3bc LaLaP 0.8±0.1ab 0.5±0.2a 1.2±0.7b 1.0±0.3ab 0.5±0.2a 0.6±0.1a 0.9±0.3ab LaMO 2.1±0.2c 1.0±0.0ab 1.2±0.1ab 1.1±0.2ab 0.9±0.1a 0.9±0.2a 1.1±0. 1ab LaMP 4.7±0.2a 11.0±0.0d 10.3±0.1c 9.4±0.2b 13.9±0.2e 11.4±0.2d 10.0±0.8bc LaOO 2.0±0.1d 1.0±0.1ab 1.3±0.3bc 1.3±0.3bc 0.8±0.1a 0.9±0.3a 1.4±0.1bc LaPO 1.2±0.2c 0.7±0.1ab 0.6±0.1ab 0.5±0.0ab 0.5±0.3ab 0.3±0.1a 0.5±0.2ab LaPP＋MMO 1.3±0.2a 5.1±0.2cd 4.5±0.1bc 4.1±0.3b 6.9±0.2e 5.7±0. 5d 4.9±0.3c MMP 1.9±0.2e 0.8±0.2ab 1.0±0.2bc 0.9±0.1ab 0.6±0.1a 1.1±0. 2bc 1.3±0.1cd MPO＋POL 2.6±0.2c 0.8±0.3a 1.0±0.1a 1.0±0.2a 0.6±0.1a 0.9±0.2a 1.8±0.0b OOO 1.1±0.1bc 1.1±0.4bc 1.1±0.1bc 1.2±0.2c 0.7±0.1ab 0.7±0.0a 0.8±0.1ab All data are mean values±standard deviations of triplicate measurements. La = lauric acid, C = capric acid, M = myristic acid, O = oleic acid, P = palmitic acid and L = linoleic acid. Values with the same letter in each row are not significantly different (p＞0.05).

rienced decreases from that of CNO（p＜0.05）at the crystallization times of 0.5 h for 10℃ and 2 h for 12℃. The contents of these TAGs increased upon increasing crystallization time（p＜0.05）at both crystallization temperatures. On the contrary, LaLaM, LaMM, LaMP and LaPP＋MMO （where P and O are palmitic and oleic acids, respectively）, which are high melting TAGs12）with carbon number ≥ 38, increased from that of CNO at the crystallization times of 0.5 h for 10℃ and 2 h for 12℃（ p＜0.05）but decreased as the crystallization time increased（p＜0.05）. The rest of TAGs remained relatively unchanged with crystallization temperature and time. The decrease in low-melting TAGs （or TAGs with short and medium chain fatty acids）and the increase in high-melting TAGs in all solid fractions with respect to the original sample, CNO, agreed with a previous report12）. The content of CCLa and CLaLa in all liquid fractions ob- Fig. 1 DSC crystallization thermograms of a CNO and its tained at 10 and 12℃ increased slightly from that of CNO （） whereas the content of LaMM and LaMP decreased from solid fractions and（b）CNO and its liquid fractions. that of CNO（data not shown）. The crystallization time exhibited no significant effect on the contents of these TAGs 3.3 Crystallization thermograms at both crystallization temperatures. Finally, the crystalliza- The DSC crystallization thermogram of CNO exhibited tion temperature and time did not have any effect on the two distinct exothermic peaks at ～6 and －0.9℃（ a and b, rest of TAGs in the liquid fractions. respectively, in Fig. 1a）, corresponding to the crystallization of high- and low-melting TAGs. The crystallization

onset temperature（Tonset）of CNO was 9.3℃（ Table 4）. The

955 J. Oleo Sci. 66, (9) 951-961 (2017) S. Sonwai, P. Rungprasertphol, N. Nantipipat et al.

Table 4 Crystallization onset temperatures（Tonset）and temperatures of phase transition points based on DSC crystallization thermograms shown in Fig. 1.

Temperature of transition point (℃) Tonset (℃) High-melting peak Low-melting peak CNO 9.3 5.9 －0.9 Solid fractions S10C0.5h 15.3 12.0 3.3 S10C1h 14.1 11.7 1.7 S10C2h 12.7 10.3 1.0 S12C2h 16.7 13.9 3.8 S12C3h 16.4 13.7 3.7 S12C6h 13.7 11.3 0.4 Liquid fractions L10C0.5h 6.1 3.8 －2.1 L10C1h 5.7 2.9 －2.5 L10C2h 3.5 0.6 －4.0 L12C2h 6.4 2.1 －3.7 L12C3h 5.9 1.8 －3.7 L12C6h 4.0 1.2 －3.8 crystallization thermograms of the solid and liquid fractions The crystallization peaks of all liquid fractions were appeared to be similar to that of CNO, however, there were located at the temperatures lower than those of CNO（Fig. significant differences with respect to Tonset and the crystal- 1b）. Tonset of all liquid fractions were also lower than CNO lization peak temperatures. The crystallization thermogram （Table 4）. With increasing crystallization time, the crystal- of S10C0.5h displayed two partly-merged exothermic lization peak temperatures for both high- and low-melting peaks at ～12 and ～3.3℃（ Fig. 1a）, which were signifi- TAGs as well as Tonset decreased for all fractions. The effect cantly higher than those of CNO. Tonset of the sample was ～ of the crystallization temperature was similar to what hap- 15.3℃. The solid fractions obtained at longer crystalliza- pened with the solid fractions. tion times, S10C1h and S10C2h, showed crystallization When comparing the solid and liquid fractions obtained thermograms similar to S10C0.5h. However, the locations at the same crystallization time and temperature, it was of both exothermic peaks shifted towards lower tempera- found that Tonset and the crystallization peak temperatures tures with a decrease in Tonset to ～14.1℃ for S10C1h and of all solid fractions were higher than those of CNO and ～12.7℃ for S10C2h. In addition, the crystallization peak their respective liquid fractions. This, as already pointed representing the low-melting fraction became more domi- out, was likely attributed to the fact that the solid fractions nant at longer crystallization time. At 12℃, the solid frac- contained higher quantity of saturated fatty acids with ≥ 14 tions show a similar trend as 10℃. carbon atoms（Table 2）and higher contents of high-melting When comparing the solid fractions obtained from the TAGs（Table 3）. same crystallization time but at different crystallization temperatures, it was found that the fraction obtained from 3.4 Solid fat content a higher crystallization temperature started to crystallize Changes in the SFC as a function of temperature for sooner with a higher Tonset and higher crystallization peak CNO and its solid and liquid fractions are given in Fig. 2. temperatures for both high- and low-melting fractions. The measurement of the SFC is a useful tool for determin-

From Table 4, Tonset and the crystallization peak tempera- ing the suitability of each fraction for a particular purpose ture for the high melting fraction of S12C2h were ～16.7 in formulated foods6）. SFC of CNO was ～64％ at 15℃ and and ～13.9℃, respectively, whereas the values of these pa- decreased rapidly as the temperature increased before rameters were ～12.7 and ～10.3℃ for S10C2h. This was reaching 0％ at 27.5℃. All of the solid fractions showed mainly because the former fraction contained higher con- SFC values higher than CNO at all temperatures of mea- tents of high-melting TAGs and lower contents of low- surement（Fig. 2a）. The SFC curve of S12C2h was located melting TAGs than the latter fraction（Table 3）. well above the SFC curves of other solid fractions, indicat-

956 J. Oleo Sci. 66, (9) 951-961 (2017) Solvent fractionation of coconut oil

Fig. 2 Solid fat content measured at different Fig. 3 Crystallization curves obtained during isothermal temperatures of（a）CNO and its solid fractions and crystallization of（a）CNO and its solid fractions at （b）CNO and its liquid fractions. 17℃ and（b）CNO and its liquid fractions at 15℃. ing that it was the hardest fraction likely because it contained highest content of high-melting TAGs（LaLaM, LaMM below that of CNO（Fig. 2b）, suggesting that they contained and LaMP）10, 19）and lowest content of low-melting TAGs more liquid oil than CNO at any temperature and hence （CCLa, CLaLa and LaLaLa）（ Table 3）. On the other hand, were softer than CNO. Among the liquid samples, the SFC the SFC curve of S10C2h was positioned above that of of L12C6h was the lowest at all measuring temperatures. CNO but bellow all other solid fractions, demonstrating When comparing the solid fractions with their respective that it was the softest solid fraction. The SFC results liquid fractions, it was found that the SFC values of all solid shown here agreed well with the results obtained from the fractions were higher than the liquid fractions at all tem- DSC studies shown earlier. The increased SFC in the solid peratures of measurement. The SFC increase and decrease fractions would provide them with enhanced texture and in the solid and liquid fractions, respectively, from the orig- specific melting behaviors suitable for specific applications inal fat is what to be expected after fractional crystalliza- which would normally be impossible with CNO due to its tion as reported previously6, 10, 21）. This is because the solid low SFC and melting temperature. Particularly, the in- fraction usually contains higher content of high-melting creased SFC between 25 and 30℃ would make the solid TAGs than the native fat and the liquid fraction. fractions resist high temperature better than the original CNO, thereby allowing them to be used in food applications 3.5 Isothermal crystallization such as margarines and cocoa butter substitutes. At 17℃, CNO exhibited sigmoidal growth in SFC that The SFC curves of all liquid fractions were stationed started at ～3 min followed by a rapid rise towards a

957 J. Oleo Sci. 66, (9) 951-961 (2017) S. Sonwai, P. Rungprasertphol, N. Nantipipat et al.

plateau SFC（SFCmax）of ～49％ at ～1.5 h（Fig. 3a）. All the highest SFCmax（～75％）due to its highest contents of solid fractions showed faster crystallization than CNO with saturated fatty acids（C14:0, C16:0 and C18:0）and high-melting a more rapid increase in SFC during initial crystallization TAGs, whereas S10C2h displayed the slowest crystalliza- before reaching plateaus at ～0.5-1 h. Among the solid tion and the lowest SFCmax（～53％）. The results were in fractions, S12C2h exhibited the fastest crystallization and good agreement with the DSC results（Fig. 1）and the SFC

Fig. 4 Polarized-light microscopy images of（a）CNO,（ b）S10C0.5h,（ c）S10C1h,（ d）S10C2h,（ e）S12C2h,（ f）S12C3h,（ g） S12cC6h,（ h）L10C0.5h,（ i）L10C1h,（ j）L10C2h,（ k）L12C2h,（ l）L12C3h and（m）L12C6h obtained after static crystallization at 18℃ for 24 h.

958 J. Oleo Sci. 66, (9) 951-961 (2017) Solvent fractionation of coconut oil

results（Fig. 2）which showed that S12C2h was the first the fat ranged from 50 to 150 μm. All solid fractions ob- solid fraction to crystallize and with the highest SFC values tained at 10℃ showed crystal morphology similar to CNO whereas the opposite was true for S10C2h. but with varying sizes（Fig. 4b-d）. The average crystal size At 15℃, CNO crystallized faster than at 17℃ and of S10C0.5h was slightly larger than CNO crystals（Fig. 4b） reached a plateau at ～1 h with SFCmax of ～52％（ Fig. 3b）. whereas the crystal sizes of S10C1h（Fig. 4c）and S10C2h All liquid fractions crystallized more slowly than CNO. （Fig. 4d）were smaller. The solid fractions obtained at 12℃ L12C2h showed the fastest crystallization and highest showed different crystal morphology than the original

SFCmax（50％）whilst L12C6h showed the slowest crystalli- sample and the solid fractions obtained at 10℃. S12C2h zation and lowest SFCmax of 26％. （Fig. 4e）and S12C3h（Fig. 4f）displayed granular crystals due to their high content of high-melting TAGs as men- 3.6 Crystal morphology tioned earlier. S12C6h exhibited densely-populated moder- The polarized-light microscopy images of CNO and its ate size spherulites（Fig. 4g）. solid and liquid fractions obtained after static crystalliza- All liquid fractions displayed the crystal morphology tion at 18℃ for 24 h are given in Fig. 4. The solid phase more closely resemble to that of CNO: spherulites consist- appears white or gray while the liquid phase appears black. ing of needle-like crystals branchinfg out in all directions When a fat is cooled from a molten stage to a temperature （Fig. 4h-m）, with L12C6h showing the largest average below its melting point, the fat will become solid and form crystal size of 300-400 μm（Fig. 4m）. primary crystals which aggregate to form clusters, resulting in the formation of a continuous three dimensional 3.7 Polymorphism network22）. CNO crystals were spherulites consisting of fine CNO first crystallized at 7℃ with a weak SAXD peak at needle-like crystals radiating and branching outward from 40.4 Å closely followed by an additional small peak at 34.5 densly-packed central cores（Fig. 4a）. The crystal size of Å at 6℃（ Fig. 5a）. Soon after that, the diffraction peak at

Fig. 5 Changes in polymorphic structures in SAXD（a-c）and in WAXD（d-f）as a function of temperature during crystallization of（a and d）CNO,（ b and e）L12C2h and（c and f）S12C2h.

959 J. Oleo Sci. 66, (9) 951-961 (2017) S. Sonwai, P. Rungprasertphol, N. Nantipipat et al.

34.5 Å became dominant and as the temperature reduced sample. The solid fractions exhibited increased SFC espe- further the diffraction peak at 40.4 Å disappeared and the cially at higher temperatures between 25 and 30℃ but all peak at 34.5 Å moved slightly to locate at 34.2 Å. The CNO melted just below the body temperature. These fractions WAXD diffraction peaks first appeared at 6℃ at 4.17 Å could be used as specialty fats such as chocolate-type with additional peaks appearing soon after that at 3.82 and coating fats and fats for margarines that original CNO could 4.29 Å（Fig. 5d）. The appearance of the diffraction peaks in not be used without hardening process such as partial hy- SAXD before WAXD suggested the organization of the drogenation. On the contrary, the decrease in the SFC of lattice structure prior to subcell organization. No further the liquid fractions would give them a higher cold stability changes in WAXD and SAXD peak shape or spacing were which would be useful during longer storage time. With a observed during the experimental timeframe. The peaks at good understanding of the effect of process conditions for 3.82 and 4.17 Å were close to 3.80 Å and 4.2 Å which are the fractionation of CNO, the crystallization can be opti- typical diffraction WAXD β′ polymorph patterns23）. The mally controlled to produce required special fractions for 4.29 Å peak was also related to β′ structure of lipid24）. This specific food applications. agreed with a previous report that CNO has one stable polymorph called β’-225）. The polymorphic structure of all solid and liquid fractions did not differ from that of the original sample there- References fore only the diffraction patterns in SAXD and WAXSD of 1） Che Man, Y.B.; Marina, A.M. Medium chain triaclglyc- the solid and liquid fractions obtained from 12℃ for 2 h erol. in Nutraceutical and Specialty Lipids and （S12C2h and L12C2h）are shown in Fig. 5. The only differ- their Co-Products（Shahidi, F. ed.）, Taylor & Francis ence lied in the temperature at which the diffraction peaks Group, New York, pp. 27-56（2006）. started to appear for each sample during the temperature 2） Maruyama, J.M; De Martini Soares, F.A.S.; D’Agostin- decrease. All solid samples began to show the first diffrac- ho, N.R.; Goncalves, M.I.A.; Gioielli, L.A.; da Silva, R.C. tion peaks in both SAXD and WAXD at higher tempera- Effects of emulsifier addition on the crystallization and tures than CNO, and vice-versa, the first diffraction peaks melting behavior of palm olein and coconut oil. J. Ag- in both SAXD and WAXD of the liquid samples started to ric. Food. Chem. 62, 2253-2263（2014）. appear at lower temperatures than CNO. L12C2h started 3） Adhikari, P.; Shin, J.-A.; Lee, J.-H.; Hu, J.-N.; Zhu, X.- to crystallize at 5℃ with a diffraction peak in SAXD at 40.3 M.; Akoh, C.C.; Lee, K.-T. Production of trans-free Å（Fig. 5b）whereas the S12C2h started to crystallize at margarine stock by enzymatic interesterification of 16℃ with diffraction peaks in SAXD at 40.6 and 35.2 Å rice bran oil, palm stearin and coconut oil. J. Sci. （Fig. 5c）. The same trend was observed in WAXD where Food Agric. 90, 703-711（2010）. the diffraction peaks started to emerge at 3℃ for L12C2h 4） Tsuiji, H.; Takeuchi, H.; Nakamura, M.; Okazaki, M.; （Fig. 5e）and at 15℃ for S12C2h（Fig. 5f）. Changes in poly- Kondo, K. Dietary medium-chain triacylglycerols sup- morphic structure after fractionation have been observed press accumulation of body fat in a double-blind, con- with other fats such as palm oil26）and milk fat6）mainly due trolled trial in healthy men and women. J. Nutr. 131, to differences in compositions and configurations of the 2853-2859（2001）. fatty acids and TAGs23）. The lack of influence of the frac- 5） Bootello, M.A.; Garcés, R.; Martínez-Force, E.; Salas, tionation conditions on the polymorphism of CNO fractions J.J. Dry fractionation and crystallization kinetics of reported in this work may have lied in the fact that CNO high-oleic high-stearic sunflower oil. J. Am. Oil Chem. has β’-2 as a stable polymorph as already mentioned. Soc. 88, 1511-1519（2011）. 6） Lopez, C.; Ollivon, M. Triglycerides obtained by dry fractionation of milk fat. 2. Thermal properties and polymorphic evolutions on heating. Chem. Phys. Lip- 4 Conclusion ids. 159, 1-12（2009）. The crystallization of fats and oils is closely related to 7） Calliauw, G.; Fredrick, E.; Gibon, V.; Greyt, W.D.; their chemical and physical properties. In this work, CNO Wouters, J.; Foubert, I.; Dewettinck, K. On the frac- was solvent-fractionated into many solid and liquid fractional crystallization of palm olein: Solid solutions and tions. Depending on the crystallization temperature and eutectic solidification. Food Res. Int. 43, 972-981 duration（10℃ for 0.5, 1 and 2 h, and 12℃ for 2, 3 and 6 h）, （2010）. these fractions showed unique fatty acid and TAG composi- 8） Gibon, V. Fractionation of lipids for use in food. in tions and crystallization behavior. This work has demon- Modifying Lipids for Use in Food（Gunstone, F.D. strated that the liquid fractions of CNO were softer and ed.）, Woodhead Publishing Limited, England, pp. 201- crystallized more slowly whereas the solid fractions were 233（2006）. harder and crystallized faster with respect to the original 9） Kellens, M.; Gibon, V.; Hendrix, M.; De Greyt, W. Palm

960 J. Oleo Sci. 66, (9) 951-961 (2017) Solvent fractionation of coconut oil

oil fractionation. Eur. J. Lipid Sci. Technol. 109, 336- 18） Yella Reddy, S. Improving plasticity of milk fat for use 349（2007）. in baking by fractionation. J. Am. Oil Chem. Soc. 87, 10） Kang, K.K.; Kim, S.; Kim, I.H.; Lee, C.; Kim, B.H. Se- 493-497（2010）. lective enrichment of symmetric monounsaturated tri- 19） Zaliha, O.; Chong, C.L., Cheow, C.S.; Norizzah, A.R.; acylglycerols from palm stearin by double solvent frac- Kellens, M.J. Crystallization properties of palm oil by tionation. LWT-Food Sci. Technol. 51, 242-252 dry fractionation. Food Chem. 86, 245-250（2004）. （2013）. 20） Tan, C.P.; Che Man, Y.B. Differential scanning calori- 11） AOAC. Official Methods of Analysis of AOAC Inter- metric analysis of palm oil, palm oil based products national. AOAC International, Arlington, VA（1995）. and coconut oil: Effects of scanning rate variation. 12） Marikkar, J.M.N.; Saraf, D.; Dzulkifly, M.H. Effect of Food Chem. 76, 89-102（2002）. fractional crystallization on composition and thermal 21） Lee, K.-T.; Foglia, T.A. Fractionation of menhaden oil behavior of coconut oil. Int. J. Food Prop. 16, 1284- and partially hydrogenated menhaden oil: Character- 1292（2013）. ization of triacylglycerol fractions. J. Am. Oil Chem. 13） AOCS. Official Methods and Recommended Practic- Soc. 78, 297-303（2001）. es of the American Oil Chemists’ Society. AOAC In- 22） Martini, S.; Herrera, M.L.; Hartel, R.W. Effect of cool- ternational, Arlington, VA（1998）. ing rate on nucleation behavior of milk fat-Sunflower 14） Sonwai, S.; Ornla-Ied, P.; Aneknun, T. Lauric fat cocoa oil blends. J. Agric. Food. Chem. 49, 3223-3229 butter replacer from krabok（Irvingia malayana）（2001）. seed fat and coconut oil. J. Oleo Sci. 64, 357-365 23） D’Souza, V.; deMan, J.M.; deMan, L. Short spacings （2015）. and polymorphic forms of natural and commercial sol- 15）Sonwai S.; Podchong P.; Rousseau D. Crystallization ki- id fats: A review. J. Am. Oil Chem. Soc. 67, 835-843 netics of cocoa butter in the presence of sorbitan es- （1990）. ters. Food Chem. 214, 497-506（2017）. 24） Ghotra, B.S.; Dyal, S.D.; Narine, S.S. Lipid shortenings: 16） Marina, A.M.; Che Man, Y.B.; Nazimah, S.A.H.; Amin, I. A review. Food Res. Int. 35, 1015-1048（2002）. Chemical properties of virgin coconut oil. J. Am. Oil 25） Timms, R.E. Phase behavior of fats and their mixtures. Chem. Soc. 86, 301-307（2009）. Prog. Lipid Res. 23, 1-38（1984）. 17） Kar, S.; Ghosh, R.K.; Bhattacharyya, D.K.; Ghosh, M. 26） Braipson-Danthine, S.; Gibon, V. Comparative analysis Isopropanol fractionation of coconut oil into its olein of triacylglycerol composition, melting properties and and stearin fractions. Int. J. Sci. Res. Sci. Technol. 1, polymorphic behavior of palm oil and fractions. Eur. J. 83-88,（ 2015）. Lipid Sci. Technol. 109, 359-372（2007）.

961 J. Oleo Sci. 66, (9) 951-961 (2017)