Effects of Low-melting-point Fractions of Cocoa Butter on Rice Bran Wax-corn Oil Mixtures: Thermal, Crystallization and Rheological Properties.

The fatty acid compositions, polymorphism, solid fat content (SFC), thermal properties, microstructure and rheological properties of fat blends of rice bran wax and corn oil (RWC) with low-melting-point fractions of cocoa butter (LFCB) in the range of 20-50% were investigated. With the raising content of LFCB, the hardness, SFC, storage modulus (G') and loss modulus (G'') of blend samples increased. The unsaturated fatty acids of blend samples with different LFCB proportion were in the range of 60.42% to 71.25%. Two kinds of polymorphism were observed in blend samples, which were β'-Form and β-Form. During the crystallization process, the rice bran wax was first crystallized, and then induced a part of LFCB formed β'-Form crystals and another LFCB formed the β-Form crystals. The results show that the addition of LFCB could improve the plasticity of fat blends and reduce the difference in properties between them and commercial shortening.

pearance and texture. Therefore, direct replacement of shortening with unsaturated fatty acid-rich fat blends in bakery products has been presented a challenge 9 .
Cocoa butter, as a natural solid fat, has complex chemical compounds and excellent plasticity, applying widely production of food with special appearence 10 . Thus, cocoa butter could be regarded as a proper material for improving the plasticity and structure of the fat products. Because of the difference structure of triglyceride molecular, however, cocoa butter exhibits various polymorphisms under different processing conditions 11 . But the fractionation technology could efficiently decrease the complexity of crystallization behavior of cocoa butter by simplifying its chemical compounds. By controlling the crystallization time and temperature the fractions having different melting points can be separated from cocoa butter. These fractions are generally classified into a high-melting-point fraction and a low-melting-point fraction. Researchers have focused on using high-melting-point fractions to improve the qualities of chocolate products, especially for preventing the occurrence of fat bloom on there 12 14 . But the lowmelting-point fractions generally were seen as the by-product in the process of fractionation and rarely reported its application of the food industry. And there is still a lack of information on applying underutilized the low-meltingpoint fractions from cocoa butter materials to improve physical properties of such fat blends.
Thus, this study added low-melting-point fractions of cocoa butter to rice bran wax and corn oil mixtures and investigated its crystal structure, thermal behavior, and rheological properties. The similarities in crystallization properties of the prepared samples were evaluated by comparing them with commercial shortenings. Our research will provide insight into improve the plasticity of fat blends and reduce the difference of physical properties between fat blends and commercial shortening.

Materials
Corn oil, refined rice bran wax and cocoa butter were generously provided by Kerry Specialty Fats Ltd. Tianjin, China . Commercial shortening was purchased from local market. Supelco 37 Component FAME mixture was purchased from Sigma-Aldrich China Shanghai, China . All other chemicals and organic solvents were purchased from Sinopharm Chemical Reagent Co. Ltd. Beijing, China . The fatty acid compositions of corn oil, commercial shortening and cocoa butter are presented in Table 1 17 . Cocoa butter samples were melted in glass beakers at 80 for 30 min to erase all crystal memory. Then the samples transferred to a DWB-20S water bath with temperature controller for cooling from 80 to 24 at 5 /min. In order to master the level of crystallization, solid fat content SFC of cocoa butter samples then were measurements every 1 h measured by pulsed nuclear magnetic resonance pNMR . Once the SFC of processed cocoa butter reached to 20 , samples were centrifuged 10000 rpm, 2 min . Immediately following centrifugation, the liquid cocoa butter in centrifuge tubes was poured out. The solid cocoa butter was placed on a filter paper for absorbing the liquid lipid. The liquid cocoa butter was regarded as low-melting-point fractions of cocoa butter LFCB , and the solid cocoa butter is regarded as high-melting-point fractions of cocoa butter HFCB . The melting point of HFCB is 5.1 higher than that of LFCB, but their polymorphisms are both β-V Form. The fatty acid compositions of the HFCB and LFCB are given in Table 1, and their thermal properties and polymorphisms are shown in Supplementary Fig. 1.

Preparation of fat blends
Samples were prepared firstly by heating corn oil up to 90 under stirring and rice bran wax was slowly added 5 w/w and mixed up to its complete dissolution. Then, the mixture was kept under agitation without heating for 3 min in order to form the rice bran wax-corn oil mixture RWC . Then LFCB and RWC were used to prepare the fat blend samples with the content of LFCB 20 ,25 ,30 ,35 ,40 ,45 and 50 w/w . With the aim of inducing the correct crystallization of LFCB, the fat blend samples were tempered according to the method reported by Afoakwa et al. 18 . All samples were cooled from 90 to 40 at a rate of 5 /min. The temperature of samples then were dropped to 21 at a rate of 2 /min and held for 20 min. The unstable crystals were erased by increasing the temperature to 25 for 1 min. The above process was always accompanied by stirring at 300 rpm. All fat blend samples were stored at 20 for at least 7 days before performing analysis Fig. 1 .

Fatty acid composition
The fatty acid compositions of samples were determined by gas chromatography after esterification. The fatty acid methyl esters FAMEs were prepared according to the reported method 19 . The analysis was performed on an Agilent GC-2010 gas chromatograph equipped with a HP-88 capillary column 0.25 μm, 100 m 0.25 mm, Agilent, USA and a flame ionization detector FID . The inlet temperature was 270 and the detector temperature was 280 . The initial column temperature was 130 ; then it was raised to 180 at a rate of 7 /min; then it was increased to 215 at a rate of 3 /min and maintained for 15 min; finally heated to 230 at 4 /min and hold for 3 min.

Solid fat content
SFC was monitored by pulsed nuclear magnetic resonance pNMR Minispec-mq20, Bruker, Germany according to the AOCS Official Method Cd 16b-93 20 . Approximately 3 g sample was placed in the NMR glass tube. Then SFC of these samples was measured after storage at 5, 10, 20, 30, and 40 for 30 min.

Texture analysis
The hardness of the samples was determined using a Texture analyzer TA-XT2 Stable Micro Systems, Surrey, England according to the method reported by Afoakwa et al. 18 . The sample was placed on the flat plate, followed by applying a uniform stress onto a stainless steel probe height: 8.3 cm, inner diameter: 1 cm with a needleshaped tip height: 1.3 cm, inner diameter: 0.2 cm , which moved downward at a constant rate of 2 mm/s, and displacement of 3 mm. The maximum values of penetration force measured by a 30 kg dynamometer was converted to a hardness g force , using XT.RA Dimension, Exponent 32 software Stable Micro Systems, Godalming, Surrey, UK . For each fat blend sample, two independently prepared replicates were made and three testing were repeated for different positions of each sample.

Rheological properties
The rheological properties of samples were characterized by using a rheometer MCR301, Anton Paar, Austria with a water circulator for temperature control 7 . The samples were prepared at parallel plate geometry 40 mm diameter and 2 mm thickness and then the edges were carefully trimmed with a spatula. Samples were placed in the parallel plate for 15 min at 20 before testing to reduce the effect of sampling process on the rheological properties of the samples. The viscosity of samples was measured under steady-shear conditions with a shear rate ranging from 1 to 40 s 1 . For the dynamic oscillatory experiments, frequency sweep tests were performed between 0.1 and 10 Hz to obtain the storage G and the loss modulus G . All rheological experiments were performed in triplicate and their averages were reported in the study.

Polymorphism analysis
The polymorphism of the samples was analyzed by X-ray powder diffractometry XRD Bruker D8 Advance, Karl- sruhe, Germany . The Cu-Ka radiation k 1.54 Å was set to 40 mA and 40 kV. The samples were extruded into a quartz cell having a diameter of 2 cm and a depth of 0.5 mm, and kept at 20 for 24 h before analysis. Samples were monitored at 2θ scale of 10-30 with a rate of 2 /min. The characteristics of d-spacing and diffraction intensity were analyzed using MID Jade 6.5 software Rigaku MSC Inc., the Woodlands, TX and these data would be used to determine the polymorphism of the samples.

Thermal properties
Melting thermograms of samples were measured using differential scanning calorimetry DSC 200F3, Netzsch, Germany , after indium calibration 21 . A portion 5 0.2 mg from samples was used for analysis. The portions were placed in aluminum pan and covered hermetically. All aluminum pans were stored at 20 for 12 h in order to stabilize the property of the samples. The applied melting process was heated from 20 to 90 at the rate of 5 / min. Melting behavior of samples was evaluated in terms of peak melting temperature Tm and melting enthalpy En .

Crystal morphology
A small droplet approximately 10 μL of the completely melted sample held at 90 for 15 min was placed on a preheated glass slide and subsequently covered the coverslip. The samples were stored at 20 for 24 h before testing. The crystal morphology of the samples was observed with a polarized light micrograph PLM equipped with a Leica DM4000M vertical optical microscope Leica Microsystem, Wetzlar, Germany and a digital monochrome camera Q Imaging Retiga 1300, Vancouver, BC, Canada .

Data analysis
All experiments were performed in triplicate and the results were reported as mean standard deviation. Statistical analysis of the results was completed using SPSS 19.0. Data were graphed using Origin 2017 with error bars indicating standard error of the mean. Data were compared to identify significant differences between samples using one-way analysis of variance ANOVA with Turkey posttest p 0.05 . The data with different letters were regarded to have statistically significant differences of their means.

Fatty acid composition
The fatty acid profiles are the key to describe the physiochemical properties of a solid fat 22 . The fatty acid compositions of blend samples are shown in Table 2. The saturated fatty acid increased from 28.75 to 39.58 as the content of LFCB. The changes of palmitic acid C16:0 and stearic acid C18:0 were the most significant, which rose from 18.59 to 22.33 and from 8.66 to 15.89 respectively. The content of unsaturated fatty acid represented a downward trend, from 71.25 to 60.42 . As the main two unsaturated fatty acids in the blend samples, the contents of oleic acid C18:1 and linoleic acid C18:2 showed an opposite trend with increasing of the LFCB. The difference in fatty acid composition between RWC and LFCB is the main cause of this phenomenon. C18:2 is the highest content of fatty acid in RWC 58.05 , while in the LFCB is only 2.10 Table 1 . Therefore, the C18:2 in blend samples decreased from 46.86 to 30.49 with the  Table 1 , the blend samples were observed lower saturated fatty acid content. This situation has been conferred as one of the main nutritional advantages of fat blend contained liquid oil against hydrogenated shortening products 2, 23 . In the shortening, the trans oleic acid exhibited a higher level Table 1 , but its content was hardly observed in the blend samples. Trans fatty acids are a substance that is highly toxic to human health during the production process 24 . The results showed that the blend samples had a higher level of unsaturated fatty acids and an extremely low level of trans oleic acids compared to the commercial shortening. It means that these samples have significant advantages in nutrition and food safety.

Solid fat content
SFC is the feature that defines the percentage of the solid parts of fats at certain temperature and hence reflecting the changes in consistency and plasticity of products at different temperatures 25 . The SFC of blend samples with and without LFCB and commercial shortening are shown in Fig. 2. With the exception of Sample 6 and 7, the other samples had a lower SFC at different temperatures than the shortening. For the bakery products, solid fat is crucial for a well-aerated, tender, lubricated and processable viscoelastic dough structure, thus relatively high SFC values are required during the mixing process 25,26 . Vallerio et al. 27 believed that SFC for superior shortening products should be above 20 at 20 and not less than 5 at 40 .
As expectation, the SFC of blend samples increased with the content of LFCB at the same temperature. It is worth noting that the SFC of Sample 1 and Sample 2 rapidly declined above 15 . The SFC of Sample 1 was only 10 at 20 , while the SFC of other samples Sample 3-7 and shortening fats fallen between 20 and 30 . The result indicated that samples with LFCB content fewer than 25 only keep liquid or semi-fluid state at room temperature 20 , just like the appearances of the sample we observed at room temperature Fig. 1 . Samples with content of LFCB greater than 20 could remain well plasticity. This could be explained by the fact that the saturated fatty acid content of the blend samples increased with the proportion of LFCB, resulting in more fat crystals formed. Thus, the higher proportions of fat crystals in blend samples provide a enough SFC and plasticity.

Texture analysis
Hardness is important for fat-rich food texture property. The hardness values of the blend samples with and without LFCB and commercial shortening are displayed in Fig. 3. Consistent with the results of SFC, the hardness value of the blend samples increased as the content of LFCB. The hardness value of the sample contained 30 of LFCB Sample 3 was the closest to that of shortening. The  hardness of fat products has direct relationship to their fatty acid compositions 28 . The results of fatty acid compositions Table 2 indicated that the saturated fatty acids of blend samples increased with the content of LFCB. The raising of saturated fatty acids suggested forming more solid fat in blend samples, exhibiting higher SFC content. As mentioned, an increase in SFC means the samples contain more fat crystals, which result in higher hardness of the samples 29 .

Rheological properties
Apparent viscosity of the blend samples were measured as a function of shear rate. As shown in Fig. 4, the viscosity of samples had a tendency to decrease in a non-linear way with increasing shear rate. The viscosity of Sample 3 was most similar to that of shortening, consistent with the results of SFC and texture analysis. The viscosities of Sample 1 and Sample 2 were similar with that of RWC while below than other samples. This could be explained by the fact that the low level of LFCB in Sample 1 and Sample 2 could not form sufficient gel networks in the samples so the resulting their less viscosity 7 . The results indicate that there is a threshold between the LFCB of Samples 2 and Sample 3 25 and 30 , respectively , which could cause the fat blend to form a stable gel network. The shear thinning property of samples was observed, which became aggravated increase the slope of the viscosity curves as the increasing content of LFCB. This is a common property of fat blend 30,31 . There are some weak attractive forces such as London Dispersion forces or van der Waals forces in crystallization network of samples, which strengthen as the raising gel content. However, once the crystallization network is deformed by the shear force, the weak attractive force will be destroyed, resulting in a  Shock rheology experiments were used to determine the storage modulus G and loss modulus G of blend samples and commercial shortenings to understand their viscoelastic modulus Fig. 5 . There was no significant difference in G and G between RWC and Sample 1, which was contributed to the fact that less LFCB content could not change the rheology properties of sample. The G and G of all samples increased with the content of the LFCB. It s consistent with the results reported by Khakhanang et al. 33 , the more content of the solid fat were added, the more G and G of the sample were measured. In addition, a crossover point of G and G in each blend sample was observed, where permanent structural deformation starts 3,34 . Beyond and behind the crossover point, the samples exhibited solid-like behavior G G and liquid-like behavior G G respectively. The values of crossover point of G and G increase with the content of LFCB, which was construed as resulting from stronger crystallization networks 3,34 . Combining the results of fatty acids and SFC Table 2 and Fig.  2 , the increase in the content of LFCB provide more saturated fatty acid for blend samples, which not only cause a higher SFC, but also enhance the weak attractive force between the crystals, resulting in a stronger crystallization network in the samples. Therefore, the critical point crossover point of G and G representing the permanent network deformation will increase with the content of the LFCB.

Polymorphism and thermal analysis
The polymorphism of blend samples and RWC and commercial shortening are shown in Fig. 6. Except for RWC and Sample 1, three diffraction peaks were observed in the diffraction patterns of the each blend sample and shortening, which were 4.6 Å, 4.1 Å, and 3.7 Å, respectively. The results indicated that the blend samples, similar to the shortening, contained both β-Form d-spacing 4.6 Å crystals and β -Form d-spacing 4.1 Å and 3.7 Å crystals. According to previous research, the β -Form crystals in blend samples were generally considered to be the crystals of wax materials 33,35 , and the β-Form crystals represented the stable crystals of cocoa butter materials 10 . Therefore, the two crystal forms β-Form and β -Form present in the diffraction pattern of the blend samples were the crystals formed by the LFCB and rice bran wax respectively. In addition, the β-Form crystals d-spacing 4.6 Å were not observed in RWC and Sample 1. The result indicated that low content of LFCB 20 could not crystallize into β-Form crystals in the blend samples, which were in line with the previous results of SFC, hardness and rheology properties.
The melting curves and thermal information of the blend samples and commercial shortening are shown in Fig. 7 and Table 3, respectively. Only one melting peak was observed in RWC and Sample 1 melting point of 66.9 and 62.3 respectively , which was the melting peak of rice bran wax crystals. This is in agreement with the results observed in XRD Fig. 6 , indicating that rice bran wax was the only structured materials what could form crystals in the blend sample with low levels of LFCB. Three melting peaks were observed for other samples Sample 2-7 . Peaks 1 and 2 have melting points around 26 and 31 , which represented the melting points of the β -Form and β-Form crystals of cocoa butter materials, respectively 10 . The results show that the LFCB would not only form β-Form crystals, but also form β -Form crystals in the blend samples under the influence of corn oil and rice bran wax. Based on the result of polymorphism and thermal properties, the β -Form peak d-spacing 4.1 Å and 3.7 Å in the diffraction pattern Fig. 6 of the blend samples was the combination of rice bran wax and LFCB crystals. In addition, the enthalpy of Peak 1 in the blend samples was significantly higher than that of peak 2, indicating that the β -Form crystals are the main component of samples. The difference between the content of two crystal forms in LFCB decreases as increasing its content. The ratio between the enthalpy of the β -Form crystals Peak 1 and the β-Form crystals Peak 2 dropped from 4.6 in Sample 2 to 1.9 in Sample 7. There are two melting peaks in the melting curve of shortening, which represent β -Form crystals and β-Form crystals, respectively. This is the same result as observed by XRD. Unlike the blend samples, the proportion of β -Form crystals in shortening is lower than that of β-Form crystals. In baked products, the β -Form crystals in shortening are regarded as better in air stabilization than other crystal structures 36,37 . The β -Form crystals are relatively small and enable to trap a larger amount of liquid oil in the crystal network and hence create a glossy surface and a smooth texture. Therefore, the high content of the β -Form crystals in the blend samples suggest that they have good baking characteristics.

Crystals morphology
The microstructure of blend samples and RWC and commercial shortening are shown in Fig. 8. The similar microstructure Needle-like crystals which represents the β -Form crystals formed by rice bran wax were observed in both RWC Fig. 8A and Sample 1 Fig. 8B 33 . In Sample 2 Fig. 8C , some crystallites were formed around the needle-like crystals. When the content of the LFCB reached 30 Sample 3 , a significant increase in the number of crystallites was observed Fig. 8D , and these crystals were connected to each other to form a crystallization network. The density of crystallization network enhanced with the content of LFCB Figs. 8E, F, G, H , forming a stronger fat structure. These results mean that only rice bran wax could crystallize in Sample 1. In the other word, the low level of LFCB 20 could not crystallize in blend sample, which is in agreement with the results observed in polymorphism and thermal analysis. The LFCB started to crystallize around the needle-like crystals into β -Form crystallites with raising the LFCB content. Thus, the rice bran wax crystals seem like the seed crystals during the crystallization of blend samples, which crystallize first and induce the LFCB to form β -Form crystals. Finally, the crystals of LFCB and rice bran wax together form a crystallization network. The commercial shortening Fig. 8I exhibited a completely different crystalline morphology with that of   the blend samples, as a cluster of spherical crystals. This could be explained by the differences between their chemical compositions. Commercial shortening contains large amounts of palmitic acid and oleic acid Table 1 , while the main fatty acids in blend samples are linoleic acid, palmitic acid and stearic acid. Therefore, blend samples are difficult to aggregate into crystal clusters due to their higher unsaturation.
In addition, the size of needle-like crystals was observed to decrease with increasing LFCB content. Compared with Sample 1 Fig. 8B , the size of the needle-like crystals of Sample 4 Fig. 8E , Sample 5 Fig. 8F and Sample 6 Fig.  8G were significantly reduced. Almost no needle-like crystals were observed in the Sample 7 Fig. 8H , but many spherical crystals were observed. On the one hand, this is because the relative content of rice bran wax in blend samples dropped as adding more LFCB. Researchers reported that the size of rice bran wax crystals in fat blends reduced with declining its content 21,38 . On the other hand, the LFCB gradually dominates the crystallization behaviors of samples with the raising of its content. LFCB contain large amounts of 1,3-saturated-2-unsaturated triglycerides, which tend to form spherical β-Form crystals 39 . The spherical crystals indicated by red circles actually were observed in Sample 7 Fig. 8H . The results indicated that more β-Form crystals would be observed in blend samples with increasing the LFCB content, just like we discussed in the polymorphism and thermal analysis.

Conclusion
In the present study, a fat blend sample was prepared by adding LFCB into rice bran wax-corn oil mixture. The content of unsaturated fatty acids in blend samples was significantly higher than that of commercial shortening and did not contain trans fatty acids. The SFC of the samples increased with the addition of LFCB, resulting in stronger hardness, viscosity, storage modulus and loss modulus. The content of LFCB above 20 providing a better plasticity for blend samples as well as affecting their structure and physical properties. During the crystallization process, the rice bran wax first crystallized formed needle-like crystals , and then induced a part of LFCB to form β -Form crystals, while the content of β-Form crystals formed spherical crystals in the blend samples gradually increased with the content of LFCB. The crystals of rice bran wax and LFCB together formed a crystallization network.

Funding
This work was supported by the Young Scientists Fund in National Natural Science Foundation of China No.