Effect of Fatty Acid Chain Length on the Crystallization Behavior Of

Effect of Fatty Acid Chain Length on the Crystallization Behavior of Trans-free Margarine Basestocks during Storage Peng Hu1, Xuebing Xu2 and Liangli (Lucy) Yu3＊ 1 Institute of Food and Nutraceutical Science, School of Agriculture and Biology, Shanghai Jiao Tong University, Shanghai, 200240, CHINA 2 Wilmar (Shanghai) Biotechnology Research & Development Center Co., Ltd, Shanghai, 200137, CHINA 3 Department of Nutrition and Food Science, University of Maryland, College Park, MD, 20742, UNITED STATES

Abstract: In order to obtain margarine free of trans-fatty acids, four interesterified basestocks were prepared by chemical interesterification (CIE) of oil blends. Different ratios of palm stearin, palm olein and soybean oil were mixed without and with 1) fully hydrogenated Acer truncatum oil (FHATO), 2) fully hydrogenated rapeseed oil or 3) palm kernel oil containing a similar amount of saturated, monounsaturated and polyunsaturated fatty acids, but different saturated fatty acid length for CIE. Compared to the physical blends, the CIE samples demonstrated lower slip melting points and decreased solid fat contents, especially at high temperatures, indicating that the CIE samples might have improved mouthfeel. In all CIE samples, the β crystal form disappeared and only the β’ crystal form was observed, except for sample 2, which contained a mixed β and β’ forms. Furthermore, in all CIE samples, except sample 1, the β' crystal forms began transforming to β form after only two cycles of higher temperature treatments indicating that the CIE sample with FHATO had the most resistance to temperature fluctuation during storage which may be attributed to its longer saturated chains. In conclusion, the CIE basestocks containing longer saturated fatty acids could be more suitable for margarine use.

Key words: fully hydrogenated Acer truncatum oil, fatty acids, trans-free, interesterification, crystallization

1 Introduction （SFC）, triacylglycerol（TAG）composition, and the polymor- Bakery fats including margarine and shortening provide phic behavior of their crystals5）. Also, fat crystals play a key food with desirable functions such as tender texture, lubri- role in defining the functions and properties of fats. They cation, and creaming properties. However, considering the exhibit monotropic polymorphism with various crystal limited source of butter, several commonly used processing types under different conditions which can result in methods including hydrogenation, interesterification, and significantly different physical properties. There are mainly fractionation were applied to obtain bakery fats in the past three types of fat crystals, namely: α, β' and β, in an in- years. Among them, the fats produced by partial hydroge- creasing order of thermodynamic stability. Among them, nation are reported to contain trans fatty acids which can the β′ form is the most desirable crystal form due to its lead to several health problems1）. On the contrary, better plasticity which is attributed to its small needle-like interesterification has become a more popular method crystals less than 5 μm capable of immobilizing a large which produces fats low in trans fats or trans-free. Inter- amount of liquid oil. Conversely, the most stable β crystal esterification of rice bran oil, shea olein, and palm stearin2）; form is high-melting, self-occluding and coarse, and has rice bran oil, palm stearin, and coconut oil3）; and camellia large platelets; it has the most stable arrangement of the seed oil, palm stearin, and coconut oil4） have been per- molecules which is responsible for the appearance of formed, and the resulted fats showed more desirable func- visibly grainy fats, leading to the separation of oil portion6）. tionalities in foods. Hence, the stable β' form is desirable for bakery fats. In general, the physiochemical properties of bakery fats Interesterification can also alter TAG composition as well are altered by several factors, such as solid fat content as the crystal form and stability of fats, causing a change in

＊Correspondence to: Liangli (Lucy) Yu, Department of Nutrition and Food Science, University of Maryland, College Park, MD, 20742, UNITED STATES E-mail: [email protected] Accepted December 15, 2016 (received for review November 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

353 P. Hu, X. Xu and L. (L.) Yu

the physicochemical properties of physical blends. Most 105℃ for 1 h under reduced pressure to remove the resid- previous work focused on the effect of fatty acid diversity, ual water. The final product was obtained after purification TAG carbon number and diversity, and the effect of TAG via bleaching（performed with 1.5％（ w/w）bleaching earth structures on the stability of β' form in the CIE fats7, 8）. under reduced pressure for 25 min at 105℃ and filtrated Little is known about the effect of long-chain fatty acids at 70℃）and deodorization at 240℃ for 1.5 h. and long-carbon TAG compositions on polymorphic behavior of the fats, especially on the β' stability during storage. 2.3 Gas chromatography analysis of fatty acids The objective of this research was to investigate whether Fatty acid composition was determined according to the and how fatty acids chain lengths might alter the physico- AOCS method Ce 1f-96 as reported by Adhikari & Hu2）. chemical properties and crystal stability of trans-free mar- Fatty acid methyl esters（FAME）were prepared by trans- garine basestocks during storage. Four trans-free marga- esterification of oil with 0.5 mol/L KOH in methanol and n- rine basestocks with various fatty acid chain lengths were hexane at 60℃ for 0.5 h. Gas chromatography（GC）analysis prepared by the interesteriﬁcation of fully hydrogenated of FAME was performed using an Agilent 7820A GC system Acer truncatum oil（FHATO）with high long-chain fatty equipped with a Agilent GC capillary column（CP-sil 88 100 acid content（C24:0）, fully hydrogenated rapeseed oil m×0.25 mm; Agilent Technologies Inc., Middelburg, Neth- （FHRSO）, palm oil（including palm stearin（PS）, and palm erlands）. Hydrogen was used as a carrier gas. The initial olein（OL））, and palm kernel oil（PKO）with PS, OL, and temperature of the oven was 80℃ and was held for 2 min. soybean oil（SBO）as the basic stock, respectively. Fatty Then, the oven temperature was increased to 120℃ at the acid compositions（FAC）, TAG, slip melting point（SMP）, rate of 10℃/min. Oven temperature was further increased SFC and differential scanning calorimetry（DSC）were ex- to 180℃ at a rate of 5℃/min and held for 2 min followed amined for their potential use as margarine basestocks. by increasing to 206℃ at the rate of 2℃/min. Finally the The polymorphic behavior and crystal morphology stability temperature was increased to 230℃ at the rate of 25℃/ under temperature fluctuation during storage were also min and held for 5 min. The injector and detector tempera- studied. The information obtained from this study may be tures were set at 250℃ and 280℃, respectively. Fatty acids used to design and produce desirable bakery fats. were identiﬁed by comparison with relative retention times of the standard mixtures. Results are expressed as relative percentages of peak area.

2 Materials and Methods 2.4 Triacylglycerol analysis by UPC2-Q-TOF-MS 2.1 Materials TAG composition analysis of physical blends and inter- FHRSO, PS, OL, PKO, and SBO were kindly provided by esterified fats was conducted by ultra-performance conver- Kerry Specialty Fats（Shanghai）Co. Ltd. FHATO was pre- gence chromatography quadrupole time-of-ﬂight mass pared through the full hydrogenation of Acer truncatum spectrometry（UPC2-Q-TOF-MS）using a Waters Xevo G2 oil. Sodium methoxide（dry powder）was purchased from Q-TOF mass spectrometer（Milford, MA, USA）following a Aladdin Reagent（Shanghai）Co. Ltd. The standard mixtures laboratory protocol9）. One gram of the sample was dis- （fatty acid methyl esters MIX, C4-C24）for fatty acids analy- solved in 10 mL hexane/isopropanol（7:3, v/v）and filtered sis were purchased from SUPELCO（Bellefonte, PA, USA）. through a 0.22 μm filter membrane before UPC2-Q-TOF- All other chemicals were of analytical grade. MS analysis. A Waters Acquity UPC2 system, which stands for Acquity UltraPerformance Convergence Chromatogra- 2.2 Chemical interesteri cation phy（Milford, MA, USA）was equipped with a binary solvent Four different formula with the weight ratio of delivery pump, an autosampler, a column oven, and a back PS:OL:SBO:FHATO＝14:56:15:15（ w/w/w/w）, pressure regulator. The qualitative analysis was performed PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, PS:OL:SBO＝ at 50℃ using an Acquity UPC2 BEH 2-EP column（150 mm 45:40:15（w/w/w）and PS:OL:SBO:PKO＝50:20:15:15（w/w/ ×3.0 mm ID; 1.7 μm; Waters, Milford, MA, USA）. The

w/w）（ labeled as 1, 2, 3 and 4, respectively）were prepared elution gradient（eluent A, CO2; eluent B, acetonitrile: for CIE. Each CIE combination was mixed in a flat bottom ethanol at 1:1, v/v）was: started at 0.2％ B; increased lin- flask and dried at 105℃ under reduced pressure for 30 early to 0.7％ B at 5 min, 0.8％ B at 10 min, 1.2％ B at 15 min. After cooling to 80℃, 0.4％（ w/w）sodium methoxide min, 2.0％ B at 20 min, and 12％ B at 25 min. The back was added as a catalyst. Interesterification was conducted pressure was set at 1600 psi. The ﬂow rate was 1.2 mL/min, under reduced pressure for 50 min at 105℃. 10％（ w/w） and the injection volume was 1.0 μL. The data acquisition aqueous citric acid was added to inactivate the catalyst, was in a positive ion electrospray ionization（ESI）mode. and the mixture was stirred for another 10 min. The reac- The desolvation gas was nitrogen, and the collision gas was tion solution was washed with hot water until it reached a argon. The capillary voltage was 2.5 kV, the cone voltage pH of 7. Then, the oil phase was collected and dried at was 40.0 V, and the source offset was 80 V. The source

354 J. Oleo Sci. 66, (4) 353-362 (2017) Effect of Fatty Acid Chain Length on the Crystallization Behavior of Trans-free Margarine Basestocks during Storage

temperature was 120℃ and the desolvation temperature cycle. After each two cycles, the crystal polymorphism of was 500℃ with a desolvation gas flow rate of 900.0 L/h. samples were determined by X-ray diffraction（XRD）spec- The cone gas flow was 80.0 L/h. troscopy（D8 Advance, Bruker Corporation, Billerica, MA, To collect data in TOF MS experiments, MSE technology USA）with a fine copper X-ray tube, operating at 40 kV and was used in which two separate scan functions were pro- 40 mA. The short spacing was observed in the 2θ range of grammed for the MS acquisition. The first scan function 12-30°, and the scan rate was set at 2.0°/min with a step was set at a low collision energy（6 eV）to provided parent increment of 0.02°. ions, and the second scan function was set at a higher collision energy（ramped from 30 to 40 eV）to provided frag- 2.9 Crystal morphology by polarized light microscope ment ions. The scan time for each function was set at 0.3 s. An Eclipse E400 POL polarized light microscope（PLM） Data were acquired and analyzed with Waters MassLynx （Nikon, Japan）was used to observe the crystal morphology v4.1 software. The sample analysis was performed in tripli- of the physical blends and interesterified fats. A small cate. droplet（approximately 10 μL）of melted fat at 80℃ was placed on a glass slide and covered with a glass slip. Then, 2.5 Solid fat content it was chilled at －40℃ for 30 min before being tempered Solid fat content（SFC）was determined according to the at ambient temperature for 10 min for observation. The AOCS method Cd 16-93b on a Bruker minispec MQ20 NMR photomicrograph of the crystal was taken at 200×magnifi- analyzer（Bruker Optik, Ettlingen, Germany）. The fats were cation. After the first observation, the samples were kept melted completely in a 60℃ water bath for 30 min and in the temperature chamber with the same temperature then placed in 0℃ bath for 60 min. Afterwards, the program as described in 2.8. The temperature program was samples were raised to temperatures of 10, 20, 25, 30, 35 repeated every two days as a cycle. After each two cycles, and 40℃ and remained at those temperatures for 30 min, the crystal photomicrographs of these samples were ob- and were then placed in the NMR analyzer to determine served. the SFC values. 2.10 Statistical analysis 2.6 Determination of slip melting point Data were reported as the mean±standard deviation The slip melting point（SMP）of the samples was deter- （SD）for triplicate measurements. Statistical analysis was mined according to the AOCS method Cc. 3.25（AOCS performed using Statistical Analysis System Software, 1990b）. Capillary tubes filled with one-cm high column of version 9（SAS Institute, Inc., Cary, NC, USA）and the level fats were cooled at －40℃ for 4 h before being immersed of significance was set at p＜0.05. in a beaker full of cold water. The water was stirred and heated gradually. The temperature was recorded when the fat in the tube started to rise due to hydrostatic pressure. That temperature was recorded as the SMP. 3 Results and Discussion 3.1 FAC contents 2.7 Differential scanning calorimetry The FAC contents of the individual oils and the inter- Melting profiles of samples were studied using a differ- esterified fats are presented in Table 1. PS, OL, and SBO ential scanning calorimetry（DSC）method（DSC Q2000, TA were all used in CIE 1-4, while FHATO, FHRSO, and PKO Inc., New Castle, DE, USA.）. The DSC instrument was cali- were used in CIE 1, CIE 2, and CIE 4, respectively. FHATO brated using indium（m.p. 156.6℃, Hf＝28.45 J/g）and n-do- was rich in stearic acid（C18:0, 63.0％）and other long-chain decane（m.p. 9.65℃, Hf＝216.73 J/g）. An empty aluminum fatty acids including arachidic acid（C20:0, 8.7％）, behenic pan used as the reference sample（5 mg）was accurately acid（C22:0, 17.9％）, and lignoceric acid（24:0, 6.2％） weighed for DSC analysis. The sample was heated to 80℃ （Table 1）. FHRSO had a high content of behenic acid at 20℃/min and held for 10 min. Thereafter, the tempera- （C22:0, 44.8％）, while PKO mainly consisted of lauric acid ture was decreased to －40℃ at 5℃/min. After holding at （C12:0, 46.7％）and myristic acid（C14:0, 16.1％）. Thus, －40℃ for 10 min, the melting curve was obtained by after interesterification, CIE 1 showed a little higher heating to 80℃ at 5℃/min2）. content of C20:0 and C24:0 relative to CIE 2 which was rich in C22:0. CIE 3 and CIE 4 exhibited lower C18:0 2.8 Polymorphism by X-ray diffraction spectroscopy content but was higher in C16:0 level than that of CIE 1 The melted（80℃ for 30 min）samples before and after and CIE 2. CIE 4 also contained 6.3％ C12:0 and 3.1％ interesterification were chilled at －40℃ for 30 min before C14:0, which were much greater than that of the other stored at a temperature chamber in which the temperature three interesterified fats. Through rational designing of was held at 10℃ for 24 h and at 25℃ for another 24 h. The formula, CIE 1-4 showed similar content of oleic acid temperature program was repeated every two days as a （C18:1）, linoleic acid（C18:2）and linolenic acid（C18:3）, re-

355 J. Oleo Sci. 66, (4) 353-362 (2017) P. Hu, X. Xu and L. (L.) Yu , （ w/w/ w/w/w/w ） 3.1f ± 0.1 4.7c ± 0.2 0.2b ± 0.0 0.3b ± 0.0 0.5b ± 0.1 0.3b ± 0.0 0.3b ± 0.0 6.3b ± 0.1 0.3b ± 0.0 1.1b ± 0.0 （ 45:40:15 ＝ 14:56:15:15 ＝ 5.0c ± 0.2 0.2b ± 0.0 0.3b ± 0.1 0.6b ± 0.1 0.2b ± 0.1 0.9d ± 0.1 0.4b ± 0.0 1.2b ± 0.0 , PS:OL:SBO 1.4c ± 0.2 5.9c ± 1.0 ND ND 0.1a ± 0.1 ND ND 0.2b ± 0.0 0.6b ± 0.1 0.3b ± 0.1 0.8d ± 0.1 0.4b ± 0.0 1.1b ± 0.0 （ w/w/w/w ） , PS:OL:SBO:FHRSO , 14:56:15:15 1.5c ± 0.2 0.7c ± 0.0 0.2b ± 0.0 2.4b ± 0.4 0.7b ± 0.2 0.6b ± 0.2 0.2b ± 0.1 0.4b ± 0.0 1.1b ± 0.0 ＝ w/w/w/w ）（ 3.0f ± 0.1 4.5c ± 0.2 13.9e ± 0.2 10.7d ± 0.2 0.2b ± 0.0 0.3b ± 0.0 0.5b ± 0.1 0.2b ± 0.1 0.3b ± 0.0 ND ND ND ND ND ND 6.2b ± 0.2 0.3b ± 0.0 1.1b ± 0.0 14:56:15:15 ＝ , PS:OL:SBO:FHRSO 5.1c ± 0.2 0.2b ± 0.0 0.3b ± 0.1 0.6b ± 0.0 0.2b ± 0.0 0.9d ± 0.2 0.4b ± 0.0 1.2b ± 0.0 （ w/w/w/w ） 1.5c ± 0.1 6.2c ± 0.7 ND ND 0.2a ± 0.1 ND ND 0.7c ± 0.1 0.2b ± 0.0 0.6b ± 0.1 0.3b ± 0.0 0.4b ± 0.0 1.1b ± 0.0 14:56:15:15 ＝ 1.3c ± 0.2 0.7c ± 0.0 0.2b ± 0.0 2.5b ± 0.4 0.8b ± 0.2 0.5b ± 0.1 0.1b ± 0.1 0.3b ± 0.1 1.0b ± 0.1 9.0e ± 0.5 1.2c ± 0.1 0.1a ± 0.1 31.3d ± 0.7± 0.8 30.9d 31.4d ± 0.1 28.1c ± 0.9± 0.5 31.1d 31.2d ± 0.8± 0.1 31.5d 28.1c ± 0.9 0.1a ± 0.1 16.1f ± 0.4 16.2f ± 0.3 16.9f ± 0.1 15.2e ± 0.1 16.0f ± 0.4± 0.3 16.2f 16.9f ± 0.4 15.2e ± 0.1 0.8c ± 0.1 4.2a ± 0.4 33.0d ± 0.5 33.4d ± 0.6 44.6g ± 0.2 41.1f ± 0.3 33.2d ± 0.5± 0.7 33.2d 44.8g ± 0.2 41.4f ± 0.3 0.1a ± 0.0 31.3f ± 0.4 30.9f ± 0.4 31.4f ± 0.1 28.0e ± 0.1 31.1f ± 0.4± 0.4 31.2f 31.5f ± 0.1 28.1e ± 0.1 0.2b ± 0.0 0.4b ± 0.1 of the individual oils, blends and four chemical interesterified fats. ） , respectively. , respectively. （ FAC 0.1a ± 0.1 4.1a ± 0.2 0.1a ± 0.0 8.7d ± 0.0 6.2d ± 0.0 （ w/w/w/w ） 1.0c ± 0.4 ND 0.3a ± 0.2 17.9d ± 0.0± 0.4 44.8e 1.7c ± 0.6 ND ND 0.1a ± 0.1 ND 0.7c ± 0.2 ND ND 5.5c ± 0.5 ND ND 0.4b ± 0.0 4.1b ± 0.1 63.0g ± 0.0 39.2f ± 0.6 13.7e ± 0.2 10.9d ± 0.2 50:20:15:15 ＝ ± 0.1 , respectively. , respectively. 0.1a 0.1a ± 0.0 0.2a ± 0.1 0.4b ± 0.0 0.2b ± 0.1 ND ND 1.0d ± 0.1 4.2b ± 0.1 0.3b ± 0.2 0.4ab ± 0.3 Fatty acid compositions （ w/w/w/w ） Table 1 Table oil; FHRSO stands for fully hydrogenated rapeseed PKO palm kernel CIE chemical interesterifica tion; ND not detected. 6.1c ± 0.4± 0.4 11.5d 58.4g ± 0.5 ND 0.2a ± 0.1 1.2e ± 0.1 5.0c ± 0.2 5.9c ± 0.3 10.8d ± 0.2 51.3i ± 0.4 ND ND 14.6f ± 0.3± 0.4 14.5f 15.1h ± 0.1 13.6e ± 0.1 14.3f ± 0.3± 0.4 14.5f 15.1h ± 0.1 13.6e ± 0.1 0.1a ± 0.1 0.1a ± 0.1 0.2b ± 0.1 , and PS:OL:SBO:PKO 0.3b ± 0.1b 50:20:15:15 standard deviation; the same letter in the same row means there was no significant difference. PS stands for palm stearin; OL stands for palm olein; SBO stands for soybean oil; FHATO stands for stands for palm olein; SBO soybean oil; FHATO PS stands for palm stearin; OL standard deviation; the same letter in row means there was no significant difference. ＝ ± （ w/w/w ） ND ND ND ND ND ND ND ND ND ND ND PKO PS OL SBO FHATO FHRSO Blend 1 Blend 2 Blend 3 Blend 4 CIE 1 CIE 2 CIE 3 CIE 4 80.8f ± 0.2 66.6e ± 1.6± 0.5 45.6b 16.0a ± 0.2 99.8g ± 0.2± 0.1 99.5g 52.1c ± 1.0 52.5c ± 1.0 51.1c ± 0.4± 0.4 55.6d 52.6c ± 1.0 52.4c ± 0.9± 0.8 51.2c 55.8d ± 1.0 Acer truncatum 0.1a ± 0.0 3.7c ± 0.4 ND ND ND ND ND ND ND ND 3.4c ± 0.2 ND ND ND ND ND ND ND ND 2.3a ± 0.1 2.7b ± 0.1 8.6b ± 0.2 59.9h ± 1.6 39.8e ± 0.6 10.9c ± 0.2 16.0b ± 0.3 26.8d ± 0.4± 0.5 42.3h 24.7c ± 0.4 2.6b ± 0.0 45:40:15 ）＝（ n-6 ）（ n-3 （ n-9 ）（ n-6 ）（ n-3 ） , and PS:OL:SBO:PKO C20:0 C22:0 C24:0 SFA MUFA 16.1b ± 0.3 26.9c ± 1.3 42.5e ± 0.5± 0.4 25.1c PUFA TFA C8:0 C10:0 C12:0 46.7d ± 0.4 C14:0 16.1g ± 0.3 C16:0 C18:0 C18:1 C18:2 C18:2T C18:3 C18:3T PS:OL:SBO Values are shown as mean Values fully hydrogenated PS:OL:SBO:FHATO of ratio weight the with interesterification chemical randomized by prepared were samples 4 CIE and 3 CIE 2, CIE 1, CIE Blend 1, 2, 3 and 4 samples were physical mixtures with the weight ratio of PS:OL:SBO:FHATO w ）

356 J. Oleo Sci. 66, (4) 353-362 (2017) Effect of Fatty Acid Chain Length on the Crystallization Behavior of Trans-free Margarine Basestocks during Storage

Table 2 Triacylglycerol（TAG）compositions of interesterified fats and physical blends.

Exact Mass Calculated + CIE 1 CIE 2 CIE 3 CIE 4 Blend 1 Blend 2 Blend 3 Blend 4 ([M+NH4]) Mass PPP 824.7713 824.7707 6.0a±0.1 6.6b±0.1 12.4d±0.1 15.8e±0.2 6.5b±0.1 7.0c±0.1 15.4e±0.1 17.1f±0.2 PSP 852.8030 852.8020 5.1f±0.1 2.9e±0.1 2.3d±0.1 1.9d±0.1 0.7a±0.0 1.3b±0.1 1.6c±0.1 2.0d±0.1 POP 850.7892 850.7864 12.4a±0.1 16.5c±0.1 20.9d±0.1 14.1b±0.1 21.9e±0.2 22.5ef±0.2 26.2g±0.2 23.9f±0.2 PLP/ 848.7702 848.7707 5.9a±0.1 6.4b±0.1 10.2f±0.2 7.4e±0.1 6.7c±0.1 7.1d±0.1 7.2d±0.1 6.3b±0.1 PSS 880.8329 880.8333 2.4d±0.1 1.5c±0.1 0.5a±0.0 0.4a±0.0 1.4c±0.1 0.9b±0.1 0.4a±0.0 0.4a±0.0 POS 878.8185 878.8177 8.5e±0.2 5.5d±0.1 3.5b±0.1 2.6a±0.1 3.6b±0.1 4.0c±0.1 3.9c±0.1 3.6b±0.1 POO 876.8030 876.8020 9.3b±0.2 12.3c±0.2 12.3c±0.2 7.6a±0.1a 15.6e±0.1 15.8e±0.1 14.8d±0.1 12.0c±0.1 PLS 876.8039 876.8021 4.4e±0.1 2.7d±0.1 2.0c±0.1 1.5a±0.1 1.8b±0.1 1.8b±0.1 1.5a±0.1 1.5a±0.1 PLO 874.7855 874.7864 9.0d±0.1 8.8d±0.1 12.1e±0.2 7.6bc±0.1 7.6bc±0.1 7.7c±0.1 7.4b±0.1 5.9a±0.1 PLL 872.7711 872.7708 2.8b±0.1 3.1c±0.1 3.8d±0.1 2.4a±0.1 3.0c±0.1 3.1c±0.1 3.1c±0.1 2.4a±0.1 SSS 908.8622 908.8646 1.7b±0.0 2.6c±0.1 0.5a±0.1 0.4a±0.0 3.9d±0.1 2.7c±0.1 0.5a±0.0 0.4a±0.0 SOS 906.8484 906.8490 2.5d±0.1 1.3c±0.0 0.4a±0.1 0.4a±0.0 0.6b±0.1 0.5ab±0.0 0.5ab±0.0 0.4a±0.1 SOO 904.8326 904.8333 3.3f±0.1 2.1e±0.1 1.1b±0.0 0.7a±0.0 1.8d±0.0 1.8d±0.1 1.7d±0.1 1.3d±0.0 OOO 902.8165 902.8177 1.8b±0.1 2.5de±0.1 2.2c±0.1 1.3a±0.1 2.7f±0.1 2.6ef±0.1 2.4d±0.1 2.0b±0.0 OSL 902.8166 902.8177 3.8d±0.1 1.8c±0.1 1.2b±0.1 0.7a±0.0 1.3b±0.1 1.3b±0.1 1.2b±0.0 0.9a±0.1 OLO 900.8002 900.8020 3.4d±0.1 2.8c±0.1 3.3d±0.1 1.8b±0.0 1.8b±0.1 1.8b±0.1 1.8b±0.1 1.4a±0.1 OLL 898.7849 898.7864 1.7b±0.1 1.8b±0.1 1.6b±0.1 1.0a±0.1 3.9d±0.1 2.0c±0.1 2.0c±0.1 1.6b±0.1 others －－ 6.6c±0.2 11.1e±0.3 0.1a±0.0 0.3a±0.1 9.8d±0.3 10.1d±0.3 2.0b±0.4 1.7b±0.2 Values are shown as mean±standard deviation; the same letter in the same row means there was no significant difference. M stands for myristic; P stands for palmitic acid; S stands for stearic acid; O stands for oleic acid; L＝linoleic acid. All the TAG positional isomers are included. For example, POP＝POP + PPO + OPP; POS＝POS + PSO + SPO; POO＝POO + OPO + OOP; SOS＝SOS + SSO + OSS; SOO＝SOO + OOS; OLL＝OLL + LLO + LOL. CIE 1, CIE 2, CIE 3 and CIE 4 samples were prepared by randomized chemical interesterification with the weight ratio of PS:OL:SBO:FHATO＝14:56:15:15 （w/w/w/w）, PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively. Blend 1, Blend 2, Blend 3 and Blend 4 samples were mixtures with the weight ratio of PS:OL:SBO:FHATO＝14:56:15:15（w/w/w/w）, PS:OL:SBO:FHRSO＝ 14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively.

sulting in comparable content of saturated fatty acids （9.8％ and 10.1％, respectively）in comparison with Blend （SFA）, monounsaturated fatty acids（MUFA）, and polyun- 3 and Blend 4（2.0％ and 1.7％, respectively）（ Table 2）. saturated fatty acids（PUFA）. Although the hydrogenated After interesterification, the TAG compositions clearly fats of FHATO and FHRSO were used in CIE 1 and CIE 2, varied in POP, PSP, PPS, and SSS（Table 2）. The content of respectively, the trans fatty acid content of both samples POP in CIE 1 decreased nearly 50％ after interesterifica- was less than 0.7％. Thus, FHATO and FHRSO can be used tion. The content of PSP and PPS in CIE 1 increased sig- to prepare low trans margarine fats. nificantly in comparison with Blend 1, whereas there was little change for other three CIE samples. Also, SSS content 3.2 TAG composition in CIE 1 decreased dramatically while it had little change The TAG composition of the interesterified fats and in the other CIE samples. The content of PPP in the four physical blends are summarized in Table 2. Due to the dif- interesterification samples was lower（p＜0.05）than that of ference in fatty acid compositions of Formula 1-4, the TAG their physical blends. However, CIE 1 and 2 showed much compositions of Blends 1-4 are different. As Blends 3 and 4 lower（p＜0.05）PPP content than CIE 3 and 4 for the dif- were rich in C16:0, they showed greater content of PPP ference in their fat composition. The changes in the TAG （15.4 and 17.1％, respectively）and POP（26.2 and 23.9％, profiles were responsible for the variation in physical char- respectively）relative to Blends 1 and 2（6.5 and 7.0％, re- acteristics of the samples after interesterification. spectively for PPP; 21.9 and 22.5％ respectively for POP）（Table 2）. As Blends 1 and 2 contained FHATO and 3.3 SFC and SMP determination FHRSO rich in C18:0, their SSS content（3.9 and 2.7％, re- The SMPs of the interesterified fats ranged from 39.2 to spectively）were higher than that in Blends 3 and 4（0.5 and 43.6℃, which were much lower than those of the physical 0.4％, respectively）（ Table 2）. In addition, the content of blends at 48.3 to 54.0℃（ Table 3）. In comparison with the long-chain fatty acids（C20:0, C22:0, and C24:0）in Blends 1 physical blends, the interesterified fats showed lower SFCs and 2（4.6 and 7.9％, respectively）were also greater than at high temperatures（30, 35 and 40℃）. Importantly, the that in Blends 3 and 4（Table 1）. Thus, higher contents of SFC content between 1.99％ and 8.47％ at 40℃ could undefined long TAGs were detected in Blend 1 and Blend 2 improve the mouth melting taste. On the contrary, the

357 J. Oleo Sci. 66, (4) 353-362 (2017) P. Hu, X. Xu and L. (L.) Yu

Table 3 Solid fat contents（SFC）and slip melting points（SMP）of interesterified fats and physical blends.

CIE 1 CIE 2 CIE 3 CIE 4 Blend 1 Blend 2 Blend 3 Blend 4 10℃ 48.08a±1.18 50.85b±0.29 54.68d±0.14 57.68e±0.23 51.15b±0.28 47.39a±0.40 53.76c±0.25 58.52e±0.31 20℃ 33.4 b±0.23 34.47c±0.28 33.09b±0.52 33.43b±0.37 36.93d±0.21 31.48a±0.33 36.96d±0.22 40.01e±0.16 SFC 25℃ 23.32b±0.33 25.11d±0.18 23.28b±0.47 22.55a±0.45 28.8 f±0.09 23.95c±0.14 27.71e±0.28 29.67g±0.29 （％） 30℃ 16.66c±0.17 18.58d±0.15 15.81b±0.14 14.28a±0.15 23.92h±0.11 20.24e±0.19 21.06f±0.08 21.79g±0.14 35℃ 10.43b±0.43 13.22c±0.23 9.76b±0.55 7.44a±0.49 20.39f±0.08 17.23e±0.32 15.7 d±0.41 15.56d±0.55 40℃ 3.88b±0.11 8.47d±0.11 6.09c±0.33 1.99a±0.22 16.31h±0.10 13.87g±0.31 11.3 f±0.31 10.79e±0.28 SMP（℃） 40.6b±0.20 43.3 c±0.60 43.6 c±0.10 39.2 a±0.40 51.9 e±0.10 54.0 f±0.00 48.3 d±0.10 48.8 d±0.20 Values are shown as mean±standard deviation; the same letter in the same column means there was no significant difference. CIE 1, CIE 2, CIE 3 and CIE 4 samples were prepared by randomized chemical interesterification with the weight ratio of PS:OL:SBO:FHATO＝14:56:15:15 （w/w/w/w）, PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively. Blend 1, Blend 2, Blend 3 and Blend 4 samples were mixtures with the weight ratio of PS:OL:SBO:FHATO＝14:56:15:15（w/w/w/w）, PS:OL:SBO:FHRSO＝ 14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively.

SFCs of all physical blends were greater than 10％ at 40℃, leading to a greasy mouthfeel. The greater SFC might limit their further application in shortening and margarine. In general, to achieve a desirable spreadability for bakery fats, the desirable SFCs should be between 15％ to 35％ at ambient temperature（25℃）10）. To avoid oil leakage, the SFC should also exceed 10％ at 20℃11）. As expected, the interesterified fats showed SFC values of 22.55-25.11％ at 25℃ and 33.09-34.47％ at 20℃, which fulfill the require- ments of bakery fats, indicating their potential application in margarine basestocks.

3.4 DSC analysis The melting thermographs of physical blends and interesterified fats are shown in Fig. 1. For Blend 1 and Blend 2, there was a large sharp melting peak at 51℃, indicating Fig. 1 DSC melting thermographs of interesterified fats the high melting TAGs in PS, FHATO and FHRSO. Blends 3 （CIE 1-4）and physical blends（Blend 1-4）. CIE 1, and 4 showed three main melting peaks including a small CIE 2, CIE 3 and CIE 4 samples were prepared by melting peak at around 49℃ and a broad melting peak at randomized chemical interesterification with the around 20-43℃ relative to Blends 1 and 2 which can be as- weight ratio of PS:OL:SBO:FHATO＝14:56:15:15 cribed to the difference of raw materials. After interesteri- （w/w/w/w）, PS:OL:SBO:FHRSO＝14:56:15:15（w/w/ fication, the large sharp melting peaks in Blends 1 and 2 w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and were replaced by a broad melting peak at low temperature PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, （27-42℃）. In addition, all the lowest melting peaks shifted respectively. Blend 1, Blend 2, Blend 3 and Blend 4 to high temperature indicating the formation of softer samples were mixtures with the weight ratio of products. The variation might be attributed to the rear- PS:OL:SBO:FHATO＝14:56:15:15（w/w/w/w）, rangements of the fatty acids in the triacylglycerol mole- PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, cules2）. PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL: SBO:PKO＝50:20:15:15（w/w/w/w）, respectively. 3.5 Polymorphism Polymorphic form was the most essential criterion for 1 while Blends 2-4 contained a mixture of β and β' crystal the functional properties of margarine and shortening12）. forms（Fig. 2A）. After interesterification, only β’ crystal Short spacing of the major polymorphs was identified as form was observed in CIE 1, CIE 3 and CIE 4 which may be follows: α, a single spacing at 4.15 Å; β', two strong spac- due to the TAG variation induced by interesterification, at ings at 3.7-4.0 Å and 4.2-4.3 Å; β, a very strong spacing at the same time β and β’ were both observed in CIE 2. It is 4.6 Å13, 14）. The polymorphic forms of physical blends and reported that PSP and SPS exist in the β' form, and PPS interesterified fats after different temperature cycles were and PSS exhibit both the β and β' forms whereas PPP, POS, measured by XRD diffraction and are presented in Fig. 2. It and SSS exist in the β form15）. As discussed above, the in- can be seen that only the β crystal form was found in Blend teresterified fats had lower PPP, POP, and SSS content and

358 J. Oleo Sci. 66, (4) 353-362 (2017) Effect of Fatty Acid Chain Length on the Crystallization Behavior of Trans-free Margarine Basestocks during Storage

Fig. 2 X-ray diffraction spectra of the physical blends（Blend 1-4）and interesterified fats（CIE 1-4）with selected temperature cycles. Each temperature cycle involves 24 h at 25℃ followed by 24 h at 10℃. CIE 1, CIE 2, CIE 3 and CIE 4 samples were prepared by randomized chemical interesterification with the weight ratio of PS:OL:SBO:FHATO ＝14:56:15:15（w/w/w/w）, PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively. Blend 1, Blend 2, Blend 3 and Blend 4 samples were mixtures with the weight ratio of PS:OL:SBO:FHATO＝14:56:15:15（w/w/w/w）, PS:OL:SBO:FHRSO＝14:56:15:15（w/w/w/w）, PS:OL:SBO＝45:40:15（w/w/w）, and PS:OL:SBO:PKO＝50:20:15:15（w/w/w/w）, respectively.

359 J. Oleo Sci. 66, (4) 353-362 (2017) P. Hu, X. Xu and L. (L.) Yu

Fig. 3 Polarized light microscopy photomicrographs of physical blends（Blend 1-4）and interesterified fats（CIE 1-4）at the end of 0, 2 and 6 temperature cycles. Scale bar is 10 μm; and magnification is 200. A1-4 stand for polarized light microscopy photomicrographs of Blend 1-4 at the end of 0 cycles. B1-4 stand for polarized light microscopy photomicrographs of CIE 1-4 at the end of 0 cycles. C1-4 stand for polarized light microscopy photomicrographs of CIE 1-4 at the end of 2 cycles. D1-4 stand for polarized light microscopy photomicrographs of CIE 1-4 at the end of 6 cycles. showed higher content of PPS/PSP and PSS/SPS than their 3.6 Microscopic observation corresponding physical blends. However, these CIE The crystal morphology of physical blends and inter- samples had much different stability to temperature fluc- esterified fats stored under 0, 2, and 6 temperature cycles tuation. CIE 1 and CIE 2 showed exactly the same poly- is presented in Fig. 3. Larger crystals were observed in the morphism state even after six temperature cycles. physical blends（A 1-4）, while the interesterified fats（B1, However, a transition from β' to β crystal form was ob- B3-4）showed much smaller crystals with the exception of served after two temperature cycles for CIE 3 and CIE 4, CIE 2（B2）at the beginning of temperature cycle. However, and only the β crystal form remained after six temperature larger crystals were observed in CIE 3 and CIE 4（D3 and cycles. O‘Brien reported that the more diverse the triglyc- D4）due to the aggregation possibly induced by the transi- eride structure of the highest melting portion of the fat tion from β' to β after 6 temperature cycles. As expected, would be, the lower the β forming tendency would be16）. the crystal size was still very small for CIE 1（D1）even after deMan also reported that increasing the diversity of fatty 6 cycles, which fulfills the requirement of margarines and acids and TAG carbon number can increase the stability β’ shortenings19）. Thus, CIE 1 may be suitable for applications polymorph17）;. Thus, the resistant ability of the β' form to in margarine or shortening. temperature fluctuations for CIE 1 and CIE 2 may be attributed to their larger long-chain fatty acid content and diversity of TAG as described above. Moreover, CIE 1 had a higher content of PSP/PPS and PSS/SPS which usually 4 Conclusions remain in the stable β' form. This explained the most stable In summary, four interesterified fats were prepared via polymorph of CIE 1 in the β' form, and resisted tempera- CIE from oil blends of FHATO/PS/OL/SBO, FHRSO/PS/OL/ ture fluctuation18）. SBO, PS/OL/SBO, and PS/OL/PKO/SBO, respectively. Their SMP, SFC, melting profile, polymorphism, and crystalliza-

360 J. Oleo Sci. 66, (4) 353-362 (2017) Effect of Fatty Acid Chain Length on the Crystallization Behavior of Trans-free Margarine Basestocks during Storage

tion stability to temperature fluctuation were comprehen- 5） DeMan, J.M. Physical properties of fats, oils and sively investigated. In comparison with physical blends, the emulsiﬁers. AOCS Press: Champaign IL（2000）. CIE samples showed lower SMP and decreased SFC espe- 6） O'Brien, R.D. Fat and oils: formulating and process- cially at high temperatures which might improve mouth- ing for applications. CRC Press: Wangshington, D.C. feel. Only the β' crystal form was observed in the CIE USA（2004）. samples with the exception of the sample containing 7） Zhang, H.; Jacobsen, C.; Adler-Nissen, J. Storage sta- FHRSO. On the other hand, CIE samples with long chain bility study of margarines produced from enzymatical- saturated fatty acids containing FHATO and FHRSO were ly interesterified fats compared to margarines pro- more resistant to temperature fluctuation during storage. duced by conventional methods. I. Physical properties. As a result, the CIE sample based on FHATO was more re- Eur. J. Lipid Sci. Tech. 107, 530-539（2005）. sistant to temperature fluctuation during storage which 8） Yap, P.H.; deMan, J.M.; deMan, L. Polymorphic stability can be attributed to its saturated long chains. The CIE fat of hydrogenated canola oil as affected by addition of with FHATO remained in only the β' form even after long- palm oil. J. Am. Oil Chem. Soc. 66, 1784-1791（1989）. term storage, indicating its great potential for use in mar- 9） Zhou, Q.; Gao, B.; Zhang, X.; Xu, Y.; Shi, H.; Yu, L. garine. Our work demonstrated that the introduction of Chemical profiling of triacylglycerols and diacylglycer- FHATO before interesterification may be a rational way to ols in cow milk fat by ultra-performance convergence obtain suitable trans-free margarine basestocks. chromatography combined with a quadrupole time-of- flight mass spectrometry. Food Chem. 143, 199-204 （2014）. 10） Rao, R.; Sankar, U.K.; Sambaiah, K.; Lokesh, R.B. Dif- Acknowledgments ferential scanning calorimetric studies on structured This work was supported by the National High Technolo- lipids from coconut oil triglycerides containing stearic gy Research and Development Program of China［Grant acid. Eur. Food Res. Tech. 212, 334-343（2001）. Nos. 2013AA102202, 2013AA102207］, and the foundation 11） Laia, O.M.; Ghazalia, H.M.; Cho, F.; Chong, C.L. Physi- program of industry research from the Wilmar（Shanghai） cal and textural properties of an experimental table Biotechnology Research & Development Center Co., Ltd. margarine prepared from lipase-catalysed transesteri- The authors wish to thank the QC group of Wilmar for fied palm stearin: palm kernel olein mixture during technical support. The authors declare no conflict of inter- storage. Food Chem. 71, 173-179（2000）. est. 12） Reddy, S.Y.; Jeyarani, T. Trans-free bakery shortenings from mango kernel and mahua fats by fractionation and blending. J. Am. Oil Chem. Soc. 78, 635-640 （2001）. References 13） Ghotra, B.S.; Dyal, S.D.; Narine, S.S. Lipid shortenings: 1） Willett, W.C.; Stampfer, M.J.; Manson, J.E.; Colditz, a review. Food Res. Int. 35, 1015-1048（2002）. G.A.; Speizer, F.E.; Rosner, B.A.; Hennekens, C.H.; Wil- 14） D'Souza, V.; deMan, J.M.; deMan, L. Short spacings lett, W.C.; Stampfer, M.J.; Colditz, G.A.; Willett, W.C.; and polymorphic forms of natural and commercial sol- Sampson, L.A.; Rosner, B.A. Originally published as id fats: A review. J. Am. Oil Chem. Soc. 67, 835-843 Volume 1, Issue 8845 Intake of trans fatty acids and （1990）. risk of coronary heart disease among women. Lancet 15） Chu, B.S.; Ghazali, H.M.; Lai, O.M.; Che Man, Y.B.; Yu- 341, 581-585（1993）. sof, S. Physical and chemical properties of a lipase- 2） Adhikari, P.; Hu, P. Enzymatic and chemical interester- transesterified palm stearin/palm kernel olein blend ification of rice bran oil, sheaolein, and palm stearin and its isopropanol-solid and high melting triacylglyc- and comparative study of their physicochemical prop- erol fractions. Food Chem. 76, 155-164（2002）. erties. J. Food Sci. 77, C1285-C1292（2012）. 16） O'Brien, R.D. Fat and oils: formulating and process- 3） Adhikari, P.; Shin, J.A.; Lee, J.H.; Hu, J.N.; Zhu, X.M.; ing for applications. CRC Press, Wangshington, D.C. Akoh, C.C.; Lee, K.T. Production of trans-free marga- USA（2004）. rine stock by enzymatic interesterification of rice bran 17） deMan, J.M.; deMan, L. Polymorphism and texture oil, palm stearin and coconut oil. J. Sci. Food Agr. 90, of fats. AOCS Press, Illinois, pp. 225-235（2001）. 703-711（2010）. 18） Meng, Z.; Liu, Y.; Shan, L.; Jin, Q.; Wang, X. Reduction 4） Ruan, X.; Zhu, X.-M.; Xiong, H.; Wang, S.-q.; Bai, C.-q.; of graininess formation in beef tallow-based plastic Zhao, Q. Characterisation of zero-trans margarine fats fats by chemical interesterification of beef tallow and produced from camellia seed oil, palm stearin and co- canola oil. J. Am. Oil Chem. Soc. 87, 1435-1442 conut oil using enzymatic interesterification strategy. （2010）. Int. J. Food Sci. Tech. 49, 91-97（2014）. 19） Lee, J.H.; Akoh, C.C.; Himmelsbach, D.S.; Lee, K.T.

361 J. Oleo Sci. 66, (4) 353-362 (2017) P. Hu, X. Xu and L. (L.) Yu

Preparation of interesterified plastic fats from fats and 4039-4046（2008）. oils free of trans fatty acid. J. Agr. Food Chem. 56,

362 J. Oleo Sci. 66, (4) 353-362 (2017)