Thermal Behavior, Microstructure, Polymorphism, and Crystallization Properties of Zero Trans Fats from Soybean Oil and Fully Hydrogenated Soybean Oil
- First Online:
- Cite this article as:
- Ribeiro, A.P.B., Grimaldi, R., Gioielli, L.A. et al. Food Biophysics (2009) 4: 106. doi:10.1007/s11483-009-9106-y
- 318 Views
Blends of soybean oil (SO) and fully hydrogenated soybean oil (FHSBO), with 10, 20, 30, 40, and 50% (w/w) FHSBO content were interesterified under the following conditions: 20 min reaction time, 0.4% sodium methoxide catalyst, and 500 rpm stirring speed, at 100 °C. The original and interesterified blends were examined for triacylglycerol composition, thermal behavior, microstructure, crystallization kinetics, and polymorphism. Interesterification produced substantial rearrangement of the triacylglycerol species in all the blends, reduction of trisaturated triacylglycerol content and increase in monounsaturated–disaturated and diunsaturated–monosaturated triacylglycerols. Evaluation of thermal behavior parameters showed linear relations with FHSBO content in the original blends. Blend melting and crystallization thermograms were significantly modified by the randomization. Interesterification caused significant reductions in maximum crystal diameter in all blends, in addition to modifying crystal morphology. Characterization of crystallization kinetics revealed that crystal formation induction period (τSFC) and maximum solid fat content (SFCmáx) were altered according to FHSBO content in the original blends and as a result of the random rearrangement. Changes in Avrami constant (k) and exponent (n) indicated, respectively, that—as compared with the original blends—interesterification decreased crystallization velocities and modified crystallization processes, altering crystalline morphology and nucleation mechanism. X-ray diffraction analyses revealed that interesterification altered crystalline polymorphism. The interesterified blends showed a predominance of the β′ polymorph, which is of more interest for food applications.
KeywordsChemical interesterificationLow trans fatsThermal behaviorMicrostructureCrystallization kineticsPolymorphism
Chemical interesterification of blends of fully hydrogenated vegetable oils with liquid oils is currently the most versatile option for producing low trans fats for a diversity of industrial purposes1,2. Given its economic importance and wide availability, soybean oil (SO) is a suitable raw material for manufacturing trans fatty acid-free fat fractions. In order to raise the melting point of these fractions, fully hydrogenated soybean oil (FHSBO), which is also zero trans, has proven highly useful3,4. As hardstock FHSBO offers the advantage, in addition to its economic feasibility, of its high (approximately 85%) content of stearic acid, which is not atherogenic and therefore has no adverse effect on risk of cardiovascular disease5–7.
The main factors that influence a fat’s crystallization process are its chemical composition and the conditions of crystallization. Natural lipids are composed of a variety of triacylglycerol groups with specific requirements as to activation energy for molecular diffusion and formation of stable crystalline nuclei8. When these groups are modified by interesterification, the related energy requirements are altered, and changes occur in crystal growth velocity and size9. Interesterification thus produces significant modifications in the crystallization properties of oils and fats by increasing the number of triacylglycerol species present and altering intersolubility among the triacylglycerol molecules10,11.
Lipid crystallization behavior has very important implications, mainly for industrial processing of products whose physical properties depend largely on fat crystals, such as chocolates, margarines, and shortenings. The crystallization process is divided into nucleation and crystal growth phases. Nucleation involves the formation of molecule clusters that exceed a critical size and are therefore stable. Once a crystalline nucleus has formed, it begins to grow by incorporating other molecules. Crystal growth velocity is proportional to degree of supercooling and inversely proportional to solution viscosity. Crystallization kinetics has a profound influence on fats’ final structures and is intrinsically related to their rheological and plasticity properties12–15.
The hierarchical organization of structural levels of the fat’s crystalline network is understood better by examining the structural levels formed in the network when the fat crystallizes out of the liquid state. The microstructural level or mesoscale of a fat crystal network may be defined as structures measuring 0.5 to 200 μm. It is quantified mainly by visualizing its geometry. As nanostructural materials (0.4–250 nm), triacylglycerols crystallize into particular polymorphic forms. Most triacylglycerols crystallize as spherulites, meaning that crystal growth occurs radially. The crystals grow to a size of 1 to 4 μm and then form clusters (larger than 100 μm) in a mass and heat transfer-limited process. The clustering process continues until a continuous three-dimensional network is formed by the coupling of these microstructures, trapping the fat’s liquid phase16–19.
Fats’ tendency to crystallize is of fundamental concern to processing techniques. Triacylglycerols generally crystallize initially into the α and β′ polymorphic forms, although the β form is more stable. This phenomenon relates to the fact that the β form has greater activation free energy of nucleation. The polymorphic transformation is an irreversible process from the less stable to the more stable form, and depends on the temperature and time involved. At constant temperature, the α and β′ forms change, with time, into the β form by the liquid–liquid or solid–solid mechanisms20,21. Fats with crystals in the β′ form offer greater functionality, because they are softer, support aeration better, and offer creaming properties. The β′ form is thus generally the polymorph of greatest interest for producing high-fat foods, such as margarines and confectionary and bakery products. However, suitable products can be obtained even using fats with a high propensity towards the β form, such as FHSBO22.
Also, when the triacylglycerol composition of an oil or fat is subjected to change by interesterification, a series of alterations in thermal profiles are observed. Melting and crystallization thermograms are extremely useful tools for ascertaining the changes caused by randomization, while the various thermal phenomena are examined by monitoring changes in enthalpy and phase transitions of the various triacylglycerol blends23,24.
This study evaluates the effect of chemical interesterification on the thermal behavior and crystallization properties of binary SO:FHSBO blends, with a view to studying interesterified bases for food product applications. The modifications resulting from randomization were evaluated by way of triacylglycerol composition, differential scanning calorimetry, crystallization isotherms, polarized light microscopy, and X-ray diffraction.
Materials and Methods
The materials used were refined soybean oil (SO), purchased commercially, and fully hydrogenated soybean oil (FHSBO), kindly provided by a local industry. The catalyst was 99%-pure sodium methoxide powder (Sigma-Aldrich).
The blends were prepared in the proportions 90:10, 80:20, 70:30, 60:40, and 50:50 SO:FHSBO (w/w), melted at 100 °C and homogenized for 10 min at that temperature to melt the crystals completely, prior to each interesterification reaction.
For the reactions, a 500-mL jacketed borosilicate glass reactor with bottom drain port and conical ground glass joints was coupled to a thermostated circulator bath (LAUDA RE 212, −30 to +200 °C, ±0.02 °C), stirring system (universal motor with electronic speed control up to 4,000 rpm—Marconi, BR) with axial flow stirrer, vacuum pump (Vacuubrand model 30 diaphragm pump), and probe-type digital thermometer (−50 to +300 °C, ±1 °C—Incoterm). The samples (200 g) were dried for 20 min at 100 °C in the reactor itself, under vacuum with stirring at 500 rpm. Catalyst content was 0.4% and the reaction was conducted under vacuum at 100 °C, with stirring at 500 rpm, for 20 min, according to the optimization by Grimaldi et al.25. The reaction was terminated by adding distilled water and 5% citric acid solution. The interesterified samples were washed carefully with distilled water (80 °C) to remove any soaps that had formed and were then dried under vacuum at 110 °C for 30 min.
Triacylglycerol composition was analyzed in a CGC Agilent 6850 Series GC System capillary gas chromatograph. An Agilent DB-17HT Catalog: 122–1811 capillary column (50%-phenyl-methylpolysiloxane, 15 m in length × 0.25 mm bore, containing 0.15 μm film). The conditions were: split injection, ratio 1:100; column temperature 250 °C, programmed up to 350 °C at 5 °C/min; carrier gas helium at 1.0 mL/min flow rate; injector temperature 360 °C; detector temperature 375 °C; injection volume 1.0 μL; and sample concentration of 100 mg/5 mL of tetrahydrofurane. Triacylglycerol groups were identified by comparing retention times, following the procedures of Antoniosi Filho et al.26.
Thermal analysis of the samples was performed by differential scanning calorimetry according to AOCS method Cj 1–9427. The equipment used was a Perkin Elmer DSC 7 thermal analyzer coupled to a TAC 7/DX Thermal Analysis Controller Cooler. The data processing software used was Pyris Series Thermal Analysis System. The conditions of analysis were: sample weight ~10 mg; crystallization curves, 80 °C for 10 min, 80 °C to −40 °C (10 °C/min), and −40 °C for 30 min; and melting curves, −40 °C to 80 °C (5 °C/min). The following parameters were used in evaluating the results: crystallization and melting onset temperatures (Toc and Tof), crystallization and melting peak temperatures (Tpc and Tpf), crystallization and melting enthalpies (ΔHc and ΔHf) and crystallization and melting end temperatures (Tfinal crist and Tfinal fus)28.
Polarized Light Microscopy
The samples were melted at 70 °C in an oven and, with the aid of a capillary tube, one drop of the blend was placed on a glass slide, preheated to a controlled temperature (70 °C), and covered with a slip. Duplicate slides were prepared for each sample. The slides were kept in the oven at the analysis temperature (25 °C) for 24 h. Crystal morphology was examined by polarized light microscope (Olympus, model BX 50) coupled to a digital video camera (Media Cybernetics). The slides were placed on the heating plate support (Mettler Toledo, FP82 Microscope Hot Stage), which was kept at the crystallization temperature. Images were captured under polarized light and ×40 magnification using Image Pro-Plus software version 220.127.116.11 (Media Cybernetics). On each slide, three fields were examined, of which only one was selected to represent the crystals observed. The evaluation parameter chosen for quantitative analysis of the images was maximum crystal diameter14.
The crystal polymorphic form of the fat sample was determined according to AOCS method Cj 2–9527. The analyses were performed on a Philips diffractometer (PW 1710), using Bragg-Bretano geometry (θ:2θ) with Cu-Kα radiation (λ = 1.54056 Å, at 40 KV and 30 mA). Measurements were taken at 0.02° step size at 2θ and 2-s acquisition time with 5 to 40° scans (2θ scale). The samples were melted in a microwave oven at approximately 80 °C and stabilized at 25 °C for 24 h in an oven. Analyses were performed in duplicate at 25 °C. Polymorphic forms were identified from characteristic crystal short spacings. The α form displayed a single diffraction line at 4.15 Å. The β′ form is characterized by two strong diffraction lines at 3.8 and 4.2 Å, while the β form is associated with a series of diffraction lines, one prominent at 4.6 Å and lines of lesser intensity at 3.7 and 3.8 Å27,34,35. β and β′ type crystal contents in the samples were estimated by relative intensity of the short spacings at 4.2 and 4.6 Å23,27,35,36.
Results and Discussion
Interesterification caused significant alteration in the triacylglycerol composition of the blends studied. Increased FHSBO concentration in the blends is associated with higher S3 triacylglycerol content and progressive decline in other triacylglycerol classes. In all blends, interesterification produced a significant decrease in S3 and, to a lesser degree U3, triacylglycerol content with a corresponding increase in the percentages of mainly SU2 and S2U triacylglycerols. These changes appear as a function of blend FHSBO content. With increasing FHSBO concentration in the blends, after-interesterification decreases in S3 and U3 triacylglycerol content ranged, respectively, from 95.64% to 54.72% and 23.50% to 60.38%. After-interesterification increases in S2U and SU2 triacylglycerols ranged, respectively, from 95.10% to 74.52% and 31.53% to 38.30%. Randomization thus caused the largest percentage variations in triacylglycerols of classes S3 and S2U.
List et al.38 evaluated the interesterification of an 80:20 SO:FHSBO blend, with a view to producing margarines and shortenings. S3, S2U, SU2, and U3 contents before the reaction were 23.3%, 2.5%, 27.4%, and 45.3%, respectively. After randomization, the percentages of these triacylglycerol classes were, respectively, 3.3%, 20.4%, 46.9%, and 28.9%, very close to those found in this study. Interesterification of a 50:50 SO:FHSBO blend was studied by Zeitoun et al.39. The original blend contained 50.4%, 1.5%, 17.1%, and 31.0%, respectively, of S3, S2U, SU2, and U3. In the interesterified blend, the corresponding values were 12.3%, 46.4%, 32.1%, and 9.2%.
Crystallization onset temperature (Toc), crystallization peak temperature (Tpc), crystallization enthalpy (ΔHc) and crystallization end temperature (Tfinal crist) of the original and interesterified blends
ΔHc 1 (J/g)
ΔHc 2 (J/g)
Tfinal crist (°C)
Melting onset temperature (Tof), melting peak temperatures (Tpf), melting enthalpies (ΔHf) and melting end temperature (Tfinal fus) of the original and interesterified blends
Tpf 1 (°C)
Tpf 2 (°C)
ΔHf 1 (J/g)
ΔHf 2 (J/g)
After randomization, the first peak of the thermograms for all blends was observed to shift to the right, with respective increases in the values of Tof and Tpf1, relating to the significant increase in the content of intermediate melting point triacylglycerols (S2U and SU2). Also, ΔHf 1 values increased after the reaction, indicating increased participation by these triacylglycerol species in the interesterified blends. In parallel, interesterification caused a decline in the values of Tpf 2, Tfinal fus and ΔHf 2, resulting from the lower percentage of S3 triacylglycerols in the blends. Considering only the interesterified blends, comparison of the melting thermogram parameters revealed that the tendency for Tof, Tpf1, Tpf2, Tfinal fus, ΔHf 1, and ΔHf 2 to increase is in complete agreement with the triacylglycerol composition results (Figure 1), with S2U + SU2 triacylglycerol content corresponding to 60.07%, 67.57%, 72.77%, 74.20%, and 64.24%, respectively, for the 90:10-I, 80:20-I, 70:30-I, 60:40-I, and 50:50-I SO:FHSBO blends.
Maximum crystal diameter of the SO:FHSBO blends before and after interesterification
Maximum crystal diameter (μm)
47.52 ± 5.45
18.37 ± 7.88
52.71 ± 8.13
5.97 ± 2.74
55.09 ± 7.45
6.63 ± 3.90
60.79 ± 5.22
9.08 ± 6.44
116.45 ± 16.20
25.13 ± 15.18
The original blends produced spherulite-shaped crystals with maximum diameters from 47.52 to 116.45 μm. Particularly in the 50:50 SO:FHSBO blend, the spherulites were observed to form predominantly large clusters. According to Shi et al.44, crystalline morphology is dominated by the triacylglycerol species with the highest melting point in a blend. The presence of large spherulites or spherulite clusters gives the fat an undesirable grainy texture and is a function mainly of tristearin (StStSt) in the blends48,49. In line with these observations, crystal diameter diminished with increasing SO concentration in the blends, but crystalline morphology was not altered as a result of dilution. Rousseau et al.49 also observed this behavior when the proportion of liquid oil was increased in palm oil/soybean oil and lard/canola oil blends.
After randomization, the crystal morphology of the 90:10, 80:20, 70:30, and 60:40 SO:FHSBO blends was completely modified: disk-shaped crystals, with similar granular crystalline structure, were observed50. However, the 50:50 SO:FHSBO blend maintained the tendency to crystallize predominantly into spherulites after randomization, although a smaller proportion of disk-shaped crystals was also observed. This differential behavior probably relates to the presence of considerable tristearin content in the randomized blend, given the 24.50% S3 triacylglycerol content in this sample.
Interesterification produced significant reductions in crystal diameter in all blends. According to Herrera et al.51, fats for food product applications should have crystal diameters of less than 30 μm in order to prevent grainy mouthfeel. In all the blends evaluated, interesterification proved effective in yielding fat bases compatible with use in foods. Increased SO concentration in the randomized blends was also associated with decreased maximum crystal diameter, with the exception of the 90:10-I SO:FHSBO blend, which showed dimensions greater than the 80:20-I, 70:30-I, and 60:40-I SO:FHSBO blends. This effect may be related to the longer time required for this sample to crystallize, due to its low melting point. Also, interesterification considerably increased the number of crystals in the 80:20-I, 70:30-I, 60:40-I, and 50:50-I SO:FHSBO blends at 25 °C, as can be seen from the images in Figure 4d, f, h, j. According to Kloek et al.52, a dispersion of fat crystals with a large number of small crystals may represent desirable properties, such as good spreadability. A large number of small crystals make harder fat than a smaller number of large crystals, which is also associated with undesirable characteristics, such as grainy mouthfeel53.
According to Rousseau et al.49, the alterations in fat microstructure caused by interesterification result from modifications to morphology and the density of the crystalline network and affect the texture and functionality of the interesterified bases. The alteration in triacylglycerol composition caused by randomization modifies the relative strengths of the intraparticle bonds among the crystalline elements in a cluster and interparticle bonds among clusters, leading to the formation of differing structures44. In addition, the change in solid fat content intrinsic to the process of interesterification influences how the crystalline network is structured. When solid fat content decreases, as observed in this study, changes occur towards greater molecular area and mobility for crystal formation29. Substantial changes in crystalline network density, and crystal size and morphology, due to randomization have also been reported by Rodríguez et al.53 and Norizzah et al.54, respectively, for lard/sunflower oil and palm stearin/palm olein blends.
Study of the crystallization of interesterified fat bases is extremely important in order to suit their use to industrial process constraints and to improve control of processing stages that involve recrystallization of the fat fraction so as to ensure final product quality15.
Induction period (τSFC), maximum solid fat content (SFCmáx), Avrami constant (k), Avrami exponent (n) and respective coefficients of determination (R2) for SO:FHSBO blends before and after interesterification
Addition of FHSBO to soybean oil raised SFCmáx values. Interesterification led to a decrease in SFCmáx in all blends, an effect associated with reduced percentages of S3 triacylglycerols and accompanying increases in S2U and SU2 triacylglycerols35,57. Randomization produced 78.19%, 69.37%, 45.49%, and 24.01% reductions in SFCmáx values for the 80:20, 70:30, 60:40, and 50:50 SO:FHSBO blends, respectively, in line with the percentage reductions in S3 and U3 triacylglycerols. Blend τSFC was significantly modified by chemical interesterification. Once again, the greatest alterations were observed in the 80:20 and 70:30 SO:FHSBO blends. The decline in crystallization process induction or nucleation period results primarily from changes in solid fat content inherent to the interesterification process, which influences the formation and structuring of the crystalline network29. In addition, the formation of a small quantity of partial acylglycerols, such as mono- and diacylglycerols, as a result of interesterification, can retard onset of nucleation58. At 25 °C, τSFC was 1 to 4 min for the original blends and 4 to 17 min for the interesterified blends.
The greater the difference between the sample’s crystallization temperature and melting point, the greater the degree of super-cooling associated with crystallization and, consequently, the smaller the values of τSFC. When the degree of super-cooling is small, the triacylglycerol molecules are incorporated into the crystalline structure in the most appropriate configuration and the most stable polymorph is formed, because there is time sufficient for them to be properly oriented. With high degrees of super-cooling, however, triacylglycerol molecules are incorporated into the crystalline surface very quickly, and therefore imperfectly, resulting in the formation of mixed crystals and less stable polymorphs, which may persist indefinitely30,33. Therefore, although the percentage increase in lower melting point, mixed triacylglycerols (represented by the S2U and SU2 species) caused by randomization contributes to higher τSFC values, the interesterified blends are associated with greater stability and polymorphic homogeneity at 25 °C, which is of significant importance for food applications59,60.
The Avrami constant, k, which is primarily a function of crystallization temperature, decreased with increased proportion of SO in the original and interesterified blends, indicating that the velocity of crystallization was lower as a result of dilution of the FHSBO with liquid oil, increased U3 triacylglycerol content and reduced S3 triacylglycerol content. This effect was also observed by Cerdeira et al.61 in studying the crystallization of milkfat/sunflower oil blends. Interesterification produced a drop in the crystallization velocity of the samples, related mainly to the decrease in percentage S3 triacylglycerol content62.
The Avrami exponent, n, is sensitive to the crystallization mechanism, with respect to nucleation process and dimensions of growth. Nucleation may be sporadic or instantaneous, and crystal growth may occur in one, two, or three dimensions, characterizing the formation of needle-, disk-, or spherulite-shaped crystals, respectively33. Values of n = 3 correspond to spherulitic growth from instantaneous nuclei or disk-shaped growth from sporadic nuclei, while values of n = 2 denote needle-shaped growth from sporadic nuclei or disk-shaped growth from instantaneous nuclei33,62. Although n should be a whole number, fractional values are usually obtained. This anomaly relates primarily to the formation of crystals of similar morphology from different nucleation types (sporadic and instantaneous). Fractional values of the exponent, n, may also be explained by the simultaneous development of crystals with different morphologies62.
The value of n for the original blends was 3.59 to 2.88. According to Metin and Hartel63, values of n increase with decreasing melting point, i.e. with decreasing degree of super-cooling of the sample, which agrees with the results of this study. Observation of the original samples’ crystalline structure (Figure 5), which is made up of spherulites, corroborated the n values characteristic of spherulitic growth from instantaneous nuclei. The interesterified blends gave n values of 2.40 to 2.60. Values of n from 2 to 3 reflect disk-shaped growth, as observed by polarized light microscopy for the 80:20-I, 70:30-I, and 60:40-I SO:FHSBO blends, denoting mixed nucleation, where crystals of the same morphology arise from sporadic and instantaneous nuclei. However, the value of n = 2.4 for the 50:50 SO:FHSBO-I blend, where spherulites were the predominant crystal type (Figure 5), is possibly related to the concomitant presence of disk-shaped crystals, with both crystal types originating from instantaneous nucleation33,62. The results thus show that chemical interesterification modified the mechanism of crystallization of the original blends, altering both crystalline morphology and nucleation mechanism of the blends with from 20% to 40% FHSBO content; and with simultaneous development of crystals with different morphologies, but the same nucleation mechanism, in the 50% FHSBO blend.
X-ray diffraction is often used as a technique to evaluate chemical interesterification, helping outline applications for the fat bases produced64. Diffraction patterns of natural or interesterified fats show broader peaks than pure compounds, due to the presence of multiple triacylglycerols in the cell units accompanied by liquid oils49.
Polymorphic forms and short spacings of the original and interesterified SO:FHSBO blends
Short spacings (Å)
Zeitoun et al.39, studying the polymorphic behavior of a 50:50 SO:FHSBO blend, observed the simultaneous presence of β′ crystals (24.7%) and β crystals (75.3%). Narine and Humphrey48 also found the β′ polymorph in a 75:25 SO:FHSBO blend. DeMan et al.66 report the β′ polymorph in fully hydrogenated canola oil, which characteristically tends to β stabilize. Humphrey et al.42, evaluating polymorphic behavior in soybean oil/fully hydrogenated canola oil blends, found that the polymorphic forms varied with the concentration of fully hydrogenated canola oil in the samples. According to Metin and Hartel30 and Kloek et al.52, the persistence of β′ crystals in cases where stabilization is preferentially β-type probably results from the formation of mixed crystals, which may be favored by the triacylglycerol composition of the blend and by the tempering protocol used, particularly when a high degree of super-cooling occurs. According to McGauley and Marangoni33, two factors relate to this phenomenon: the unfavorable configuration of triacylglycerol molecules resulting from rapid nucleation and/or rapid increase in viscosity, which limits heat and mass transfer and hinders molecular arrangement. Therefore, the formation of mixed crystals may be associated with the presence of the β′ polymorph in the 60:40 and 50:50 SO:FHSBO blends.
Interesterification modified the crystalline habit of the fats evaluated towards significant predominance of the β′ polymorph. The β′ crystals were small and their morphology appropriate to the plasticity characteristics desirable in products such as margarines, shortenings, and fats for bakery and confectionary products. Conversely, the β polymorphic form tends to produce broad granular crystals, resulting in grainy products with low aeration potential, which may impair the macroscopic properties of some foods13. This result agrees with the microstructure of the blends observed by polarized light microscopy before and after randomization (Figure 5).
According to Rousseau and Marangoni35, stabilization of the β′ polymorph caused by chemical interesterification is associated with the formation of triacylglycerols with greater variation in chain length. This results in more disorderly packing of the terminal methyl groups, consequently forming less dense crystalline structures. Also, in all the blends evaluated, randomization caused a reduction in S3 triacylglycerol content for the FHSBO, represented mainly by tristearin, triacylglycerol characteristic of β polymorph stabilisation67. List et al.38,68 evaluated interesterification of various oils with hardstocks: the reaction also favored formation of the β′ structure, enabling the bases produced to be applied in margarines.
A comprehensive understanding of the functions and properties of fats or oil bases produced by interesterification is essential to outlining applications for them and obtaining food products with the desired final attributes. This study made it possible to determine that the thermal behavior, microstructure, crystallization kinetics, and polymorphism characteristics of SO:FHSBO blends were significantly altered by increased FHSBO concentration and the chemical interesterification process. In addition, these properties proved to be directly related to the blends’ triacylglycerol composition before and after randomization and, taken together, can determine their applicability.
The authors would like to thank the Brazilian Agencies FAPESP and CNPq for the financial support.