Effect of Sucrose Esterified Fatty Acid Moieties on the Crystal Nanostructure and Physical Properties of Water-in-oil Palm-based Fat Blends

Effect of Sucrose Esterified Fatty Acid Moieties on the Crystal Nanostructure and Physical Properties of Water-in-oil Palm-based Fat Blends Ryota Wakui, Takamichi Kamigaki, Yuri Nishino, Yoshiko Ito, Atsuo Miyazawa, and Makoto Shiota 1 Milk Science Institute, Megmilk Snow Brand, Co., Ltd., 1-1-2 Minamidai, Kawagoe, Saitama, JAPAN 2 Graduate School of Life Science, University of Hyogo, 3-2-1 Kouto, Kamigori-cho, Ako-gun, Hyogo, JAPAN

fatty acid, are widely used in the food manufacturing industry to improve the properties and morphology of fat crystals 14,15 . Sucrose ester of fatty acid SEF is employed in a wide range of products because it is odorless and tasteless. SEF is also used in the pharmaceutical and cosmetic industries, where its nonionic and biodegradable properties are desired. Since SEF is a manufactured additive, the type and degree of esterification of the fatty acid moiety, and thus, the fat product s hydrophobic character, can be adjusted to suit the intended use. Several studies have reported on the effects of fatty acid esterification on the fat crystallization behavior of SEF by using bulk phase and oil-in-water emulsified systems 16 24 ; however, very few studies have been conducted on the effects of the nanostructure on the physical properties of the fat product in W/O emulsion systems, which model fat spreads, or margarine.
Transmission electron microscopy TEM is an effective technique for observing nanoscale-sized fat crystals. Acevedo and Marangoni developed a practical method for quantifying the smallest fat crystal unit structure. This technique incorporated a nanoplatelet extraction method using cold isobutanol prior to cryo-TEM analysis, so that the effects exerted by the nanoplatelet structure on the physical properties of semisolid fats can be observed 10,11,25 .
This study aimed to investigate the effect of nanostructure fat crystals caused by SEF on the physical properties of W/O emulsions, including the influence exerted on semisolid fats due to rapid cooling crystallization. We examined how the type of fatty acid moieties esterified to sucrose could affect the size of the nanoplatelets using SFC, oil migration, and the storage elastic modulus as the measured parameters. The addition of SEF was observed to influence aggregated microstructures, and the smaller platelets provided higher SFC in the fat phase, relating to higher elastic modulus values in W/O emulsion fat blends at 30 . The present study attempted to show that the relationship between the nanostructure and the physical properties of fat products is controlled by the nature of SEF. The results of this study will enable manufacturers to improve food formulations for the production of various kinds of fat spreads and margarine.

Materials
Soybean oil, palm oil, and fully hydrogenated rapeseed oil were purchased from the Nisshin OilliO Group, Ltd.
Tokyo, Japan . The fatty acid composition of the lecithin was reported to be 14.8 C16:0 , 12.3 C18:1 , 62.3 C18:2 , and 6.9 C18:3 26 . The effect of the different base structures of lecithin and SEF on the fat microstructure was determined to obtain comparative data.
The formulations of the model W/O-type emulsified samples consisted of 48.0 soybean oil, 20.0 palm oil, 2.0 fully hydrogenated rapeseed oil, 0.5 monoacylglycerol, and water. The water content was 29.0 and 29.5 in the samples with and without 0.5 SEF, respectively. The fatty acid compositions of the fat blends were 19.5 palmitic acids, 5.3 stearic acids, 26.7 oleic acid, 38.9 linoleic acid, 4.7 linolenic acid, and 1.6 behenic acid. Semisolid fat blends were examined to determine their nanostructure and physical properties. These included fat blends without SEF and lecithin control , with L-195 L-SE , P-170 P-SE , S-170 S-SE , O-170 O-SE , ER-190 ER-SE , S-370 SH-SE , and with fractionated soybean lecithin LE .

Preparation of the W/O emulsions and the bulk fat
blends Fat blends with added monoacylglycerol were melted and maintained at 80 for 30 min to erase all crystal memory from the structure. Then, the water phase was added to the fat blend and mixed using a homogenizer Ultra-Turrax T25 disperser, Ika Werke GmbH & Co. KG, Staufen, Germany at 50 . The pre-emulsified mixes 300 g were added to a stainless container 8 cm diameter and 8 cm deep , equipped with a batch-type scraped surface chiller system in which the walls of the container in direct contact with the sample mixes were cooled to 20 using a circulating liquid refrigerant. The mixture was placed in the container with 2-blade propellers rotating at 600 rpm and was rapidly cooled from 50 to 10 in 300 s. The temperature of the mixture was monitored by a thermometer dredged through the mixture and the cooling temperature curve obtained was similar for all samples. Immediately after achieving 10 , the crystallized W/O emulsions were placed in polypropylene containers and kept at 5 until the relevant analysis had been conducted. The average size of the water droplets in the SEF was between 2.08 μm minimum, SH-SE and 3.02 μm maximum, O-SE . The size of the LE droplets was 1.60 μm, which was relatively small when compared with the other samples.

XRD measurements
We conducted XRD measurements Rint-Ultima 2000, CuKα: λ 1.54 Å; Rigaku Corp., Tokyo, Japan to determine the small-angle diffraction patterns 2θ 0 -6 and to calculate the long spacing values of the crystal polymorphs. The XRD spectra profiles were calculated by subtracting the blank data. The W/O emulsion fat samples were maintained at 5 during the measurement experiments. The crystallite size d was calculated by applying the Scherrer equation to the full width at half maximum FWHM of the XRD peak 27,28 . The d values appeared to agree with the nanoplatelet thickness and lamellar height taken from side view observations via cryo-TEM 10 .
The chain packing subcell structures, which are polymorphic forms of the relevant fat blends, were determined by wide-angle XRD. Quantitative estimations of the proclivity for the β-form were based on the ratio of the peak height of the β-form 0.46 nm to the β -form 0.42 nm ; these values were obtained from the XRD peak intensity 29, 30 .

SFC measurements
The SFC of the fat phases of the samples was determined with an NMR minispec mq20 Bruker Optik GmbH, Ettlingen, Germany . Here, melted bulk fat blends 4 mL were placed in NMR tubes ID 0.8 cm, L 18 cm and then immersed in a water bath at 80 to erase all crystal memory from the sample. Next, these tubes were kept at 5 for 30 min in the water bath, and the SFC value was measured. The sample tubes were then transferred to a water bath set at the crystallization temperature 10 , and another round of SFC readings was obtained after 30 min. These steps were repeated at intervals of 5 , for the temperature range from 5 to 35 . All SFC measurements were done in triplicate. In this study, the SFC from the solid-state of SEF was presumed sufficiently small compared to that of fat 0.7 SEF . To elucidate the effect of surfactants of the SFC of saturated fatty acids solidified via cocrystallization with fat, further investigation using DSC is required.
The melting property of the SFC of the samples was confirmed to be adequate for measurements taken at 30 min intervals in contrast to crystallization by preliminary determination.

Dynamic rheological measurements
The storage modulus, G , obtained from constant frequency dynamic compression experiments, was selected as the rheological parameter to be investigated, as solid fatlike behavior is described well by the storage modulus. Dynamic experiments were performed between parallel plates using a rheometer Rheometrics, TA Instruments, New Castle, DE, USA . The sample thickness and radius were 2 and 12.5 mm, respectively. A small frequency 1.0 rad/s and strain 0.5 , which have linear viscoelastic be-havior, were applied while the sample was heated from 5 to 30 at the rate of 0.5 /min. Representative profiles of W/ O emulsion samples are shown in the figure, and their repeatability was confirmed.

Fatty acid analysis
The fatty acid composition of the fat blends and monoacylglycerol were determined according to the AOCS Official Ce 1c-89 protocol 31 . GLC analysis was conducted using an HP5890 Agilent Technologies, Inc., Palo Alto, CA, USA equipped with an FID operating at 300 . The SP-2560 column Supelco, Inc., Bellefonte, PA, USA was used as the capillary column and the injector port temperature was set at 250 . The analyses were conducted by ramping the temperature from 180 to 200 at a rate of 2 /min after an initial holding time of 45 min at 180 .

Observation of water droplets in emulsion
The average sizes of the water droplets in the samples were determined using a confocal laser microscope, FLUOVIEW FV1000 Olympus Co., Tokyo, Japan , and an optical scanner GT-X820 Seiko Epson Co., Suwa, Japan .

Pretreatment for the extraction of the nanoplatelets
The sample pretreatment procedure for the extraction of nanocrystals was conducted as previously reported 10 . Briefly, the nanocrystals were separated from the fat samples using a cold solvent-based extraction method, followed by cryo-TEM imaging. The fat blend samples 0.8 g were suspended in cold isobutanol 10 mL . The mixture was then homogenized at 24,000 rpm for 10 min using a homogenizer attached to the disperser element S10N-10G Ultra-Turrax T25 disperser, IKA Werke GmbH & Co. KG, Staufen, Germany . The mixture was filtered under vacuum through a membrane filter with a pore size of 1.0 μm, and the resulting crystals were collected using a spatula. The isolated crystals were suspended in a cold mixture of isobutanol 10 mL and water 10 mL before being homogenized for 10 min and then subjected to vacuum filtration again. The recovered crystals were suspended in cold isobutanol and were sonicated for 20 min using a Branson 3510J-MTH sonicator Yamato Scientific Co., Ltd., Tokyo, Japan . During sonication, ice-cold water was used in the water bath to prevent a temperature increase. This sonication step helps to disperse nanoplatelets that may have become aggregated during the filtration steps. The temperature of the suspended samples remained under 10 during this step.

Cryo-TEM observation
Approximately 4 μl of each sample mixture was placed on a carbon grid with a perforated carbon film C-flat CF-2/2-4C, Protochips, Morrisville, NC, USA at 10 . The excess liquid was blotted off using a filter paper for 4 s. A drop of 2 uranyl acetate solution was then added to enhance the contrast of the crystals against the solvent, after which the excess solution was again blotted off using a filter paper for 4 s. After a further 2 s, the grid was dipped in liquid ethane EM GP, Leica Microsysteme GmbH, Wetzlar, Germany and transferred to a cryo holder JEOL Ltd., Tokyo, Japan for direct observation at 269 in a JEM3000EFC JEOL Ltd., Tokyo, Japan containing a topentry liquid-helium-cooled stage and operated at 300 kV using an MDS Minimum Dose System, JEOL Ltd., Tokyo, Japan . Micrographs were obtained using a sensitive charge-coupled device camera Temcam-F214, TVIPS GmbH, Gauting, Germany . By employing this observation method, the platelet structure could be observed from above, and the side views of the nanoplatelet stacks of the internal layered structure could be visualized as previously reported 10 . The size distribution of the platelets, i.e., the length and width as measured from above, was also determined.

Measurements of oil migration
The extent of oil migration, which is one of the factors used to ascertain quality defects of fat spread products, was determined. Here, W/O emulsion samples were molded into a steel cylinder 10 mm in diameter and 5 mm thick . Then, the specimens were kept at 0 for 30 min before a scaled filter paper 5 mm 40 mm was inserted into the center of the mix. The vertical oil penetration distance into the filter paper at 28 was recorded after 90 min.

Observation by polarized light microscopy
Approximately 50 mg of the fat blend was placed on a glass plate heated to 80 and covered with a glass coverslip. Glass coverslips that were ca. 30 μm thick were used as spacers to ensure uniformity of the sample thickness 32 . Then, the samples were cooled and stored in an incubator at 10 until analysis. A Leica DM2700P polarized light microscope PLM Leica Microsystems, Tokyo, Japan was employed to observe the microstructures of the bulk fat blends. The mean diameters of the fat crystals were calculated using image analysis software 33 . When measuring the size of the spherulite, all pictures were converted to binary images to allow characteristic features to be distinguished from background interference in the software.

Statistical analyses
One-way analysis of variance was used to compare the mean values obtained for the groups. Post hoc multiple comparisons were conducted using Tukey s test, with p 0.05 considered statistically significant. The statistical analyses were performed using PASW Statistic 18 SPSS Japan Inc., Tokyo, Japan .

The nanoplatelet structure of fat-functionalized SEFs
The cryo-TEM images and estimations of the crystal size of the nanoplatelets are shown in Fig. 1. The length and width of the crystals, as well as the length/width ratio, are shown in Table 1. The length of the L-SE crystals was not different from that of the control, whereas the length of the P-SE and S-SE crystals was shorter than that of the control. For the three esterified saturated fatty acids namely, lauric, palmitic, and stearic acids , long carbon chains of the fatty acid resulted in short nanoplatelets. The length of the crystals for both O-SE and the control sample was similar, whereas the crystal length of ER-SE was smaller than that of the control sample. These observations indicated that longer carbon chains on the fatty acids esterified to sucrose resulted in shorter nanoplatelets for both the saturated and unsaturated fatty acids. Furthermore, the addition of the SH-SE blend caused a decrease in the crystal length; however, the extent of the decrease was less than that for the S-SE blend. The ratio of mono-, di-, tri-, and polyester of S-170 was 1:99, while that of S-370 was 20:80, indicating that the SH-SE blend was more hydrophilic than the S-SE blend due to its lower esterified rate of fatty acids. Therefore, the differences in the crystal sizes were attributed to the higher affinity of the SH-SE blend to the interface of the W/O emulsion than the S-SE blend. As such, the contribution of the SH-SE blend to the overall crystallization process was lower.
The differences in the width of the nanoplatelets were not more marked than the variations observed in the length of the nanoplatelets. The L-SE blend contained wider nanoplatelets than the control sample, whereas the S-SE blend contained narrower nanoplatelets. The addition of S-SE decreased both the length and the width of the resulting nanoplatelets. In contrast, the addition of lecithin did not influence the size of the nanoplatelets. The ratio of length/width for each blend is shown in Table 1. The length/width ratio decreased in the following order: control L-SE P-SE S-SE for the saturated fatty acids, and control O-SE ER-SE for the unsaturated fatty acids. Thus, the addition of SEFs with long chain fatty acids produced a lower length/width ratio for the nanoplatelets.
3.2 Microstructure of bulk fat systems via polarized microscopy Fat crystal networks are arranged hierarchically with characteristic nanoscale and aggregated mesoscale structures 10 . These polycrystals are created by the agglomeration of TG nanoplatelets that constitute the primary crystals formed upon nucleation. To investigate these structures, bulk phase blends without any added water phase were prepared according to the requisite formulation, and these crystals were observed using PLM. The respective diameters of the spherulite were measured. SEFs were added to the fat phase and significant changes were observed in the crystal morphology Fig. 2 . The size of the spherulite from bulk fats was much larger than that of the SEF-containing nanoplatelet samples prepared via W/O emulsion and subjected to rapid cooling Table 1 . The crystals of semisolid fat blends containing SEFs with saturated and long-chain fatty acids namely, P-SE, S-SE, and SH-SE were smaller than those associated with the control sample. While the smallest crystals were obtained for fats containing SEFs with a stearic acid moiety. No size difference was observed between the S-SE and the SH-SE blends. Next, the spherulite size was considered. The spherulite size in the L-SE blend was bigger than that of the control sample. Differences in spherulite size between the control and the O-SE blend were also observed. These results suggested that SEFs with long chain and saturated fatty acids reduced the size of the spherulites in the bulk fat system by directly affecting the triacylglycerol TG crystal formation. The crystal size of LE was smaller than that of the control sample, which was in accordance with previous studies conducted with a blend of palm oil and palm olein 7,22 , where it was reported that palmitic and stearic sucrose ester with HLB 1 delayed nucleation, ultimately leading to smaller crystals. Figures 3A and 3B show the relationship between the nanocrystal sizes as observed using cryo-TEM and the diameter of the microscale spherulite as observed using PLM . The determination coefficient of the nanocrystal length and width relative to the spherulite diameter was calculated to be R 2 0.573 and R 2 0.625, respectively. A similar interdependent relationship existed between the size of the nanoplatelets in the W/O emulsion sample, which was subjected to rapid cooling, and the diameter of the spherulite in bulk fats. These results suggested that the molecules of SEFs with hydrophobic nature used in this study tend to exist in the oil phase in W/O emulsions. Therefore, in this study, it was clear that the size of the nanoplatelets influenced the degree of aggregation in the structures.

Determining the crystal structures using small-and
wide-angle X-ray analysis The long spacing, as determined using XRD analysis, corresponds to the thickness of the lamellae 10 . In this study, the thickness d was calculated using small-angle X-ray analysis Table 1 . Figures 3C and 3D show the correlation between the nanocrystal length and d R 2 0.187 , as well as the correlation between the nanocrystal width and d R 2 0.188 . The results suggest that the nanocrystal thickness was not influenced by the nanoplatelet length or width.
Palm oil is widely used as a hardstock ingredient in the manufacture of margarine. Palm oil-containing fat blends used in this study tend to cause polymorphic transitions Table 1 Crystal sizes of samples added sucrose ester of fatty acid and fractionated lecithin.  from the β to the β form that are due to the segregation of the TG, 1,3-dipalmitoyl 2-oleoyl glycerol POP . Fat blends containing fully hydrogenated rapeseed fat and palm oil tended to form β-type polymorphs after long storage periods 34 . In addition, the presence of saturated monoacylglycerols and tripalmitoylglycerol in palm based fat-induced the formation of a polymorphic β form 35 . Therefore, polymorphs β/β of samples were compared using wideangle X-ray analysis. The β/β ratio of the SH-SE blend and LE was higher than that of the other samples containing SEFs Table 2 , suggesting that palm-based fat, which contains a hydrophilic emulsifier, promoted transformation to a stable β form.

Effect of fatty acids moieties on the SFC
The effect of additives on the SFC was found to be dependent on certain base fat formulations 19 . To understand how the presence of certain fatty acid moieties on SEFs could affect the degree of crystal formation, the effect of SEFs on the SFC of semisolid fat mixtures containing fat blend systems with monoacylglycerol was determined in this study Table 2 . We found that the SFC of the P-SE and S-SE blends was higher than that of the control sample between 5 and 35 . The SFC of L-SE, LE, and SEFs with unsaturated fatty acids O-SE and ER-SE was the same as that of the control sample, but lower than that of the samples in which the SEFs were esterified with saturated fatty acids. Thus, the SFC increased in the following order: L-SE P-SE S-SE from 20 to 30 , indicating that the presence of SEFs with long fatty acid chains accelerated fat crystallization. While the SEFs with unsaturated fatty acids did not affect the SFC. Using SFC measurements, Herrera et al. reported that the addition of SEF with saturated fatty acids to O/W emulsion systems accelerated crystallization for the high-melting fractions of milk fat and its mixtures with sunflower oil 16,18 . For Domingues et al., accelerated crystallization was noted in the esterified fat of soybean oil and the fully hydrogenated soybean oil upon the addition of SEFs with stearic acid 23 . Garbolino et al. showed that fatty acid residues of sucrose palmitate could interact to cause co-crystallization of the palmitic acid residue that resulted in higher SFC values; this study was conducted using a fat blend containing palm oil and fat esterified with palm kernel oil and sunflower oil 17 .
3.5 The effect of the fatty acid moieties of SEF on the storage elastic modulus, G The physical properties of the semisolid fat W/O emulsion are affected by the gelation of fat during the crystallization process. The temperature dependence of G was determined for the semisolid emulsions containing SEF that had been prepare using the rapid cooling method Fig. 4 . The G value of the samples was measured between 5 and 30 , with an incremental increase of 0.5 /min. The Table 2 Physical properties of samples added sucrose ester of fatty acid and fractionated lecithin.
samples showed different G values depending on the type of fatty acid moiety attached to the SEF. G of the control sample was the highest of all the samples tested between 5 and 20 . The G values of P-SE, S-SE, and SH-SE were higher than those of LE, O-SE, and ER-SE, suggesting that the fat blends with SEF esterified with saturated fatty acids exhibited different interactions between the dispersed crystals to their unsaturated counterparts. SH-SE exhibited higher hydrophilicity when compared to S-SE due to the differences in the esterification rate between sucrose and the respective fatty acids. This result showed that the hydrophilicity of SEF exerted almost negligible effects on the physical properties of the fat blend, suggesting that the cross-interaction processes between the dispersed water droplets and the SEF were almost inconsequential. The trend in the temperature dependence of the G value of L-SE was similar to that observed for the fat blend containing SEF esterified with unsaturated fatty acids O-SE and ER-SE , whereas the G value of LE mirrored the trend seen for the SEF samples esterified with unsaturated fatty acids. Mid-length chain fatty acids had similar physical properties to the semisolid fat blends regardless of the associated saturated fatty acid moiety. Additionally, LE consisted of fractionated soybean lecithin that possessed unsaturated fatty acids; which could account for the similarity in the physical properties of SEF esterified with unsaturated fatty acids. For the saturated fatty acid samples, higher G values were noted above 25 when compared to the values seen for the control sample; in contrast, when the temperature was below 25 , the G value for all samples was lower than that of the control.
The G value of most samples decreased with increasing temperature from 5 to 15 and from 15 to 20 and then decreased again above 23 . This characteristic temperature dependence was observed in the control sample. Lopez and colleagues 36 reported an increase in G at approximately 20 for dairy products; however, the cause was not identified. We suggest that this behavior in the control sample was due to a change in the polymorph of the TGs or reconstitution of solid fat crystals between 15 to 20 , which is fixed in a 2 mm gap parallel plate in the rheometer. A combined investigation observing the microstructure of the solid fat network and monitoring the polymorph could verify our hypothesis 37 . Furthermore, the addition of emulsifiers with a high melting point to the fats was reported to increase the rate and extent of fat crystallization 16 . In this study, a similar trend was observed in the W/O emulsion containing semisolid fats; thus, resulting in higher G values in the fat blends containing SEF esterified with saturated fatty acid moieties. This suggests that the physical properties are influenced by the interaction between the fat crystals.
3.6 Effect of the fatty acid moieties of SEF on oil migration The excess oil from fatty food is of great concern to food manufacturers as it causes undesirable changes in the properties and function of fat-containing foods. Therefore, the effect of SEF esterified with various fatty acid moieties on oil migration for W/O emulsion semisolid samples was investigated Table 2 . We established that P-SE and S-SE had low oil migration tendencies when compared to the other samples. Additionally, there were no significant differences between S-SE and SH-SE regarding oil migration. Rouseau showed that the crystal network was attached to the water droplets in the W/O-type semisolid emulsion and that this behavior ultimately influenced the temperaturedependent character of oil migration 9 . However, in this study, this influence was not observed statistically due to the low detection accuracy of our procedure.

Effect of the nanoplatelet size on the physical proper-
ties of the semisolid fat blend The nanoplatelets, which were generated during the rapid cooling process through nucleation and crystal growth, immediately aggregated into larger units, leading to the creation of a three-dimensional crystal network structure. The network structure influences a variety of mechanical properties such as the storage modulus and oil migration of the fat products 10 . Figure 5A shows the relationship between the crystal length and the SFC at 10, 25, and 30 . Higher SFC values were seen with lower temperatures, whereas a decrease in the SFC value was observed when a corresponding increase in the crystal size was noted within a specific temperature range. Quantitatively, the relationship between the crystal length and the oil mi- gration value was R 2 0.537 Fig. 5B . Thus, SEFs promoted the formation of many small crystals and significantly influenced the physical properties of the samples, especially SEFs esterified with palmitic and stearic acids, which ultimately had higher SFC values. In some instances, namely, when the acyl groups of the emulsifiers and the TG were similar, a stronger interaction between the functional groups was noted. Both the fat component of palmitic acid, which was mainly derived from palm oil, as well as the fat component of stearic acid, which was derived from fully hydrogenated rapeseed oil, were extremely compatible with the fatty acids esterified to P-SE and S-SE, respectively. The relationship between oil migration and the size of the nanoplatelets was suggested 10 . The observed trend was in line with the results of the present study. Furthermore, the presence of smaller nanostructures led to an increase in the oil-storing capabilities of the matrix of fat products. An increase in the length of the nanoplatelets at 25 and 30 led to a decrease in SFC Fig. 5A ; however, this dependence was not observed at lower temperatures. The correlation between the SFC and the G values at 10 and 30 was R 2 0.049 and R 2 0.884, respectively Figs. 5C and 5D . The G value did not show a linear relationship with the SFC in the samples prepared at 10 . However, a close relationship between the two values was noted at 30 . The lower correlation between the SFC and the G values at 10 was due to the higher G value of the control at 10 compared with the other samples. This result suggested that additives SEF and fractionated lecithin used in this study decreased the strength of the fat crystal network related to G , particularly in the case of abundant amounts of fat crystals. Therefore, the regulation of the physical properties of the semi-SFC by SEF was more effective at ambient conditions than when conducted under cold conditions.
The rheological properties of the semisolid fats were dependent on the ratio of crystallized fat to oil, and the structure of the crystallized fat. In general, higher SFC values promoted more solid-like behavior of fat products. The relationship between the SFC and the storage elastic modulus can be explained theoretically 3 . The oil migration in fatty food products is due to a combination of diffusion and capillary flow processes 38 . The nanoplatelets generated during the rapid cooling stage interacted with each other and aggregated into larger microstructures via van der Waals forces; thus forming a three-dimensional network. To this end, the Marangoni group studied the effects on the nanostructure of TG crystal networks. By comparing the nanostructure of several fat ingredients, they showed that the nanoplatelet size appears to influence the physical properties 3,10 .
In this study, we confirmed that the size of the generated nanocrystals and SFC significantly influenced the physical properties of palm-based W/O fats containing SEF with various fatty acids moieties. To verify the mechanism through which SEF exerts its influence on the physical properties, more detailed work is required in which a mass fractal model can be used to define the relationship between nanostructures and the supersaturation of each sample 4, 6, 10, 39 .

Conclusions
In this study, W/O emulsion semisolid fat blends with added SEF esterified with several fatty acid moieties were prepared via the rapid cooling method. The nature of the nanoplatelet structure was determined by X-ray analysis, and physical properties elastic modulus and oil migration were determined. Changes in the size of the nanoplatelets triggered by the addition of SEF esterified with various fatty acid moieties led to differences in the nanoplate size and physical properties. In addition, analysis of the microstructure of several bulk fat systems was conducted via PLM, and the SFC was also analyzed. The presence and type of SEF impacted both the nanoplate size in the W/O emulsion fat blends and the spherulite size in the fat blends, with similar tendencies in size. Smaller platelets generally resulted in higher SFC in bulk fats, and a high correlation between the SFC and G of W/O emulsion fat blends was observed at 30 . Samples with larger nanocrystals had a higher propensity for oil migration. Thus, the addition of SEF regulated the fat crystal nanostructure during nucleation and crystal growth, ultimately influencing the physical properties. In conclusion, the results in this study could provide useful information for influencing the manufacturing process of fat products that contain less trans-fat products such as margarine and shortenings.