ORIGINAL ARTICLE

Thermal Behavior, Microstructure, Polymorphism,and Crystallization Properties of Zero Trans Fatsfrom Soybean Oil and Fully Hydrogenated Soybean Oil

Ana Paula Badan Ribeiro & Renato Grimaldi & Luiz Antonio Gioielli &Adenilson Oliveira dos Santos & Lisandro Pavie Cardoso &

Lireny A. Guaraldo Gonçalves

Received: 27 December 2008 /Accepted: 5 March 2009 /Published online: 19 March 2009# Springer Science + Business Media, LLC 2009

Abstract Blends of soybean oil (SO) and fully hydroge-nated soybean oil (FHSBO), with 10, 20, 30, 40, and 50%(w/w) FHSBO content were interesterified under thefollowing conditions: 20 min reaction time, 0.4% sodiummethoxide catalyst, and 500 rpm stirring speed, at 100 °C.The original and interesterified blends were examined fortriacylglycerol composition, thermal behavior, microstruc-ture, crystallization kinetics, and polymorphism. Interester-ification produced substantial rearrangement of thetriacylglycerol species in all the blends, reduction oftrisaturated triacylglycerol content and increase in mono-unsaturated–disaturated and diunsaturated–monosaturatedtriacylglycerols. Evaluation of thermal behavior parame-ters showed linear relations with FHSBO content in theoriginal blends. Blend melting and crystallization thermo-grams were significantly modified by the randomization.

Interesterification caused significant reductions in maximumcrystal diameter in all blends, in addition to modifyingcrystal morphology. Characterization of crystallizationkinetics revealed that crystal formation induction period(τSFC) and maximum solid fat content (SFCmáx) were alteredaccording to FHSBO content in the original blends and as aresult of the random rearrangement. Changes in Avramiconstant (k) and exponent (n) indicated, respectively, that—as compared with the original blends—interesterificationdecreased crystallization velocities and modified crystalli-zation processes, altering crystalline morphology andnucleation mechanism. X-ray diffraction analyses revealedthat interesterification altered crystalline polymorphism.The interesterified blends showed a predominance of theβ′ polymorph, which is of more interest for foodapplications.

Keywords Chemical interesterification . Low trans fats .

Thermal behavior . Microstructure . Crystallization kinetics .

Polymorphism

Introduction

Chemical interesterification of blends of fully hydrogenatedvegetable oils with liquid oils is currently the most versatileoption for producing low trans fats for a diversity ofindustrial purposes1,2. Given its economic importance andwide availability, soybean oil (SO) is a suitable rawmaterial for manufacturing trans fatty acid-free fat frac-tions. In order to raise the melting point of these fractions,fully hydrogenated soybean oil (FHSBO), which is alsozero trans, has proven highly useful3,4. As hardstockFHSBO offers the advantage, in addition to its economicfeasibility, of its high (approximately 85%) content of

Food Biophysics (2009) 4:106–118DOI 10.1007/s11483-009-9106-y

A. P. B. Ribeiro (*) : R. Grimaldi : L. A. G. GonçalvesFood Technology Department, Faculty of Food Engineering,State University of Campinas-UNICAMP,Bertrand Russel street, C.P. 6091,13083-970 Campinas, SP, Brazile-mail: badanribeiro@yahoo.com.br

L. A. GioielliDepartment of Biochemical and Pharmaceutical Technology,Faculty of Pharmaceutical Sciences, University of São Paulo-USP,05508-900 São Paulo, SP, Brazil

A. O. dos SantosSocial Sciences, Health and Technology Center,University of Maranhão-UFMA,65900-410 Imperatriz, MA, Brazil

L. P. CardosoInstitute of Physics Gleb Wataghin,State University of Campinas-UNICAMP,C.P. 6165,13083-970 Campinas, SP, Brazil

stearic acid, which is not atherogenic and therefore has noadverse effect on risk of cardiovascular disease5–7.

The main factors that influence a fat’s crystallizationprocess are its chemical composition and the conditions ofcrystallization. Natural lipids are composed of a variety oftriacylglycerol groups with specific requirements as toactivation energy for molecular diffusion and formation ofstable crystalline nuclei8. When these groups are modifiedby interesterification, the related energy requirements arealtered, and changes occur in crystal growth velocity andsize9. Interesterification thus produces significant modifi-cations in the crystallization properties of oils and fats byincreasing the number of triacylglycerol species presentand altering intersolubility among the triacylglycerolmolecules10,11.

Lipid crystallization behavior has very important impli-cations, mainly for industrial processing of products whosephysical properties depend largely on fat crystals, such aschocolates, margarines, and shortenings. The crystallizationprocess is divided into nucleation and crystal growthphases. Nucleation involves the formation of moleculeclusters that exceed a critical size and are therefore stable.Once a crystalline nucleus has formed, it begins to grow byincorporating other molecules. Crystal growth velocity isproportional to degree of supercooling and inverselyproportional to solution viscosity. Crystallization kineticshas a profound influence on fats’ final structures and isintrinsically related to their rheological and plasticityproperties12–15.

The hierarchical organization of structural levels of thefat’s crystalline network is understood better by examiningthe structural levels formed in the network when the fatcrystallizes out of the liquid state. The microstructural levelor mesoscale of a fat crystal network may be defined asstructures measuring 0.5 to 200 μm. It is quantified mainlyby visualizing its geometry. As nanostructural materials(0.4–250 nm), triacylglycerols crystallize into particularpolymorphic forms. Most triacylglycerols crystallize asspherulites, meaning that crystal growth occurs radially.The crystals grow to a size of 1 to 4 μm and then formclusters (larger than 100 μm) in a mass and heat transfer-limited process. The clustering process continues until acontinuous three-dimensional network is formed by thecoupling of these microstructures, trapping the fat’s liquidphase16–19.

Fats’ tendency to crystallize is of fundamental concern toprocessing techniques. Triacylglycerols generally crystal-lize initially into the α and β′ polymorphic forms, althoughthe β form is more stable. This phenomenon relates to thefact that the β form has greater activation free energy ofnucleation. The polymorphic transformation is an irrevers-ible 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–solidmechanisms20,21. Fats with crystals in the β′ form offergreater functionality, because they are softer, supportaeration better, and offer creaming properties. The β′ formis thus generally the polymorph of greatest interest forproducing high-fat foods, such as margarines and confec-tionary and bakery products. However, suitable productscan be obtained even using fats with a high propensitytowards the β form, such as FHSBO22.

Also, when the triacylglycerol composition of an oil orfat is subjected to change by interesterification, a series ofalterations in thermal profiles are observed. Melting andcrystallization thermograms are extremely useful tools forascertaining the changes caused by randomization, whilethe various thermal phenomena are examined by monitor-ing changes in enthalpy and phase transitions of the varioustriacylglycerol blends23,24.

This study evaluates the effect of chemical interester-ification on the thermal behavior and crystallizationproperties of binary SO:FHSBO blends, with a view tostudying interesterified bases for food product applications.The modifications resulting from randomization were eval-uated by way of triacylglycerol composition, differentialscanning calorimetry, crystallization isotherms, polarizedlight microscopy, and X-ray diffraction.

Materials and Methods

Raw Materials The materials used were refined soybean oil(SO), purchased commercially, and fully hydrogenatedsoybean oil (FHSBO), kindly provided by a local industry.The catalyst was 99%-pure sodium methoxide powder(Sigma-Aldrich).

Blends The blends were prepared in the proportions 90:10,80:20, 70:30, 60:40, and 50:50 SO:FHSBO (w/w), melted at100 °C and homogenized for 10 min at that temperature tomelt the crystals completely, prior to each interesterificationreaction.

Chemical Interesterification For the reactions, a 500-mLjacketed borosilicate glass reactor with bottom drain portand conical ground glass joints was coupled to a thermo-stated circulator bath (LAUDA RE 212, −30 to +200 °C,±0.02 °C), stirring system (universal motor with electronicspeed control up to 4,000 rpm—Marconi, BR) with axialflow stirrer, vacuum pump (Vacuubrand model 30 dia-phragm pump), and probe-type digital thermometer (−50 to+300 °C, ±1 °C—Incoterm). The samples (200 g) weredried for 20 min at 100 °C in the reactor itself, undervacuum with stirring at 500 rpm. Catalyst content was 0.4%

Food Biophysics (2009) 4:106–118 107

and the reaction was conducted under vacuum at 100 °C,with stirring at 500 rpm, for 20 min, according to theoptimization by Grimaldi et al.25. The reaction wasterminated by adding distilled water and 5% citric acidsolution. The interesterified samples were washed carefullywith distilled water (80 °C) to remove any soaps that hadformed and were then dried under vacuum at 110 °C for30 min.

Triacylglycerol Composition Triacylglycerol compositionwas analyzed in a CGC Agilent 6850 Series GC Systemcapillary 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). Theconditions were: split injection, ratio 1:100; column temper-ature 250 °C, programmed up to 350 °C at 5 °C/min; carriergas helium at 1.0 mL/min flow rate; injector temperature360 °C; detector temperature 375 °C; injection volume1.0 μL; and sample concentration of 100 mg/5 mL oftetrahydrofurane. Triacylglycerol groups were identified bycomparing retention times, following the procedures ofAntoniosi Filho et al.26.

Thermal Analysis Thermal analysis of the samples wasperformed by differential scanning calorimetry according toAOCS method Cj 1–9427. The equipment used was aPerkin Elmer DSC 7 thermal analyzer coupled to a TAC 7/DX Thermal Analysis Controller Cooler. The data process-ing software used was Pyris Series Thermal AnalysisSystem. 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 meltingcurves, −40 °C to 80 °C (5 °C/min). The followingparameters were used in evaluating the results: crystalli-zation and melting onset temperatures (Toc and Tof),crystallization and melting peak temperatures (Tpc andTpf), crystallization and melting enthalpies (ΔHc and ΔHf)and crystallization and melting end temperatures (Tfinal cristand Tfinal fus)

28.

Polarized Light Microscopy The samples were melted at70 °C in an oven and, with the aid of a capillary tube, onedrop of the blend was placed on a glass slide, preheated to acontrolled temperature (70 °C), and covered with a slip.Duplicate slides were prepared for each sample. The slideswere kept in the oven at the analysis temperature (25 °C)for 24 h. Crystal morphology was examined by polarizedlight microscope (Olympus, model BX 50) coupled to adigital video camera (Media Cybernetics). The slides wereplaced on the heating plate support (Mettler Toledo, FP82Microscope Hot Stage), which was kept at the crystalliza-tion temperature. Images were captured under polarizedlight and ×40 magnification using Image Pro-Plus software

version 4.5.1.22 (Media Cybernetics). On each slide, threefields were examined, of which only one was selected torepresent the crystals observed. The evaluation parameterchosen for quantitative analysis of the images wasmaximum crystal diameter14.

Crystallization Isotherm The samples were melted (100 °C/15 min) and kept in a high precision dry bath (TCON 2000—Duratech, USA) at 70 °C for 1 h to completely erase their priorcrystalline history. Increase in the solid fat content withcrystallization time was monitored by nuclear magneticresonance spectrometer (BRUKER pc120 Minispec), with thereading compartment stabilized at 25 °C. Data acquisition wasautomatic, with measurements taken every minute, for100 min. Crystallization kinetics was characterized by induc-tion period (τSFC)—to onset of crystal formation—andmaximum solid fat content (SFCmáx). Induction time wasobtained graphically and reflects the time necessary for thestable, critical-size nucleus to form in the liquid phase29,30.The original Avrami equation, the kinetic model most used todescribe isothermal phase transformation, was used to studythe crystallization9,31:

CGS tð ÞCGS 1ð Þ ¼ 1� e�ktn

where CGS(t) describes the solid fat content (%) as a functionof time, CGS(∞) is the solid fat content limit as time tends toinfinity, k is the Avrami constant (min−n), which considersboth nucleation and crystal growth rate and n is the Avramiexponent, which indicates the crystal growth mechanism32,33.The equation was linearized and applied to the resultsobtained in the first 15 min of crystallization to determinethe values of k and n.

X-ray diffraction The crystal polymorphic form of the fatsample was determined according to AOCS method Cj2–9527. The analyses were performed on a Philipsdiffractometer (PW 1710), using Bragg-Bretano geometry(θ:2θ) with Cu-Kα radiation (λ=1.54056 Å, at 40 KV and30 mA). Measurements were taken at 0.02° step size at 2θand 2-s acquisition time with 5 to 40° scans (2θ scale). Thesamples were melted in a microwave oven at approximately80 °C and stabilized at 25 °C for 24 h in an oven. Analyseswere performed in duplicate at 25 °C. Polymorphic formswere 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 linesat 3.8 and 4.2 Å, while the β form is associated with aseries of diffraction lines, one prominent at 4.6 Å and linesof lesser intensity at 3.7 and 3.8 Å27,34,35. β and β′ typecrystal contents in the samples were estimated by relativeintensity of the short spacings at 4.2 and 4.6 Å23,27,35,36.

108 Food Biophysics (2009) 4:106–118

Results and Discussion

Triacylglycerol Composition From the technologicalstandpoint, the triacylglycerol profile is key to under-standing the various physical properties of an oil or fat37.Figure 1 shows the triacylglycerol composition of theblends before and after interesterification, in terms of S3(trisaturated), S2U (disaturated–monounsaturated), SU2

(monosaturated–diunsaturated), and U3 (triunsaturated)triacylglycerol content.

Interesterification caused significant alteration in thetriacylglycerol composition of the blends studied. IncreasedFHSBO concentration in the blends is associated withhigher S3 triacylglycerol content and progressive decline inother triacylglycerol classes. In all blends, interesterificationproduced a significant decrease in S3 and, to a lesser degreeU3, triacylglycerol content with a corresponding increase inthe percentages of mainly SU2 and S2U triacylglycerols.These changes appear as a function of blend FHSBOcontent. With increasing FHSBO concentration in theblends, after-interesterification decreases in S3 and U3

triacylglycerol content ranged, respectively, from 95.64%to 54.72% and 23.50% to 60.38%. After-interesterificationincreases in S2U and SU2 triacylglycerols ranged, respec-tively, from 95.10% to 74.52% and 31.53% to 38.30%.Randomization thus caused the largest percentage varia-tions in triacylglycerols of classes S3 and S2U.

List et al.38 evaluated the interesterification of an 80:20SO:FHSBO blend, with a view to producing margarinesand shortenings. S3, S2U, SU2, and U3 contents before thereaction were 23.3%, 2.5%, 27.4%, and 45.3%, respectively.After randomization, the percentages of these triacylglycerolclasses were, respectively, 3.3%, 20.4%, 46.9%, and 28.9%,very close to those found in this study. Interesterification of a50:50 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 interesterifiedblend, the corresponding values were 12.3%, 46.4%,32.1%, and 9.2%.

Thermal Behavior Differential scanning calorimetry is thethermo-analytical technique most employed in studying oilsand fats. It is considered an important tool for characterizinginteresterified products. Evaluation by differential scanningcalorimetry yields direct measurements of the energyinvolved in the processes of melting and crystallization ofoils and fats. Crystallization of oils results in shrinkingvolume, associated with an exothermic effect. Conversely,when fats melt, their volume expands, characterizing anendothermic effect40. Figure 2 shows the crystallization andmelting thermograms for the blends before and afterchemical interesterification. The shape of the curves wasmodified by the reaction, reflecting the alteration intriacylglycerol composition associated with randomization41.Table 1 shows the crystallization thermogram parameters forthe blends before and after randomization.

The crystallization curve of an oil or fat can besubdivided into different exothermic regions reflectingdifferent types of triacylglycerols40. The original blendsshowed a prominent peak in the crystallization thermogram(Figure 2), which characterizes the trisaturated fraction ofFHSBO, containing high melting point triacylglycerols.Before the reaction, the values of Toc, Tpc1, and Tfinal crist

increased as a function of blend FHSBO concentration,demonstrating that the crystallization process was acceler-ated by increased S3 triacylglycerol content in the samples.The crystallization enthalpy values were between 9.47 and85.81 J/g. All parameters evaluated in relation to thecrystallization thermograms showed a positive linearrelation to increasing FHSBO concentration in the original

Fig. 1 Triacylglycerol classes(%) in SO:FHSBO blendsbefore and after inte-resterification. I interesterifiedblend. Triacylglycerols:S3 (trisaturated), S2U(disaturated-monounsaturated),SU2 (monosaturated-diunsaturated), and U3

(triunsaturated). Meansamong duplicate sampleinjections

Food Biophysics (2009) 4:106–118 109

blends, with coefficients of determination (R2) higher than0.90, as in Figure 3. A similar relation among theseparameters and hardfats concentration is observed byHumphrey et al.42 and by Humphrey and Narine43 instudying blends containing various liquid and fully hydro-genated oils.

Interesterification caused a second peak to appear in thecrystallization thermograms, characteristic of a substantial

increase in the content of intermediate melting pointtriacylglycerols (S2U and SU2) as a result of the random-ization process. Also, the intensity and breadth of the firstpeak diminished, which is typical of a mixed crystallizationprocess starting from sporadic, instantaneous nuclei42. Thiswas accompanied by a reduction in crystallization onset(Toc) and end (Tfinal crist) temperatures, enabling theinteresterified samples to crystallize at lower temperatures

1

2

1

2

1

2

1

2

1

2

1

2

1

2

50:50

50:50 - I

60:40

60:40 - I

70:30

70:30 - I

80:20

80:20 - I

90:10

90:10 - I

End

o

90:10 - I

90:10

80:20 - I

80:20

70:30 - I

70:30

60:40 - I

60:40

50:50 - I

50:50

-40 80

Temperature (ºC)-40 80

Temperature (ºC)

1

12

1

12

1

2 1

1

12

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

1

12

1

2

1

2

2

a bFig. 2 Crystallization (a) andmelting (b) thermograms of theblends before and afterinteresterification

110 Food Biophysics (2009) 4:106–118

than their original blends. Tpc2 rose as a function of the S2Uand SU2 triacylglycerol content in the interesterified blends,indicating that the intensity of crystallization of this groupof compounds is related to the proportions of these speciesin the samples. Values of ΔHc1 diminished followingrandomization, as a result of the lower concentration of S3triacylglycerols in the randomized blends. Enthalpy valuesfor the second peak observed in the thermograms (ΔHc 2)agreed with the triacylglycerol composition results in thatthey displayed a relation with the sum total of the S2U andSU2 triacylglycerol contents in the randomized blends.Table 2 shows the melting thermogram parameters for theblends before and after randomization.

The original blends produced two characteristic peaksrelating to the low melting point triacylglycerols of theunsaturated SO fraction (peak 1) and the trisaturatedfraction corresponding to the FHSBO (peak 2). Peak 1diminished, and peak 2 increased, in intensity withincreasing FHSBO content in the original blends, as inFigure 2. Melting onset temperature (Tof) varied slightly

with FHSBO content in the original blends, while thevalues of Tpf1, Tpf2, and Tfinal crist increased with blendFHSBO content, as a result of the decrease in thepercentage of triacylglycerol species from the SO and theaddition of high melting point triacylglycerols characteristicof FHSBO. This effect also shows an association with therespective reduction and increase in the values of ΔHf 1and ΔHf 2 when FHSBO is added to SO. The meltingthermogram parameters evaluated displayed a linear rela-tionship with FHSBO concentration in the original blends.However, coefficients of determination (R2) lower than 0.90were encountered for the parameters Tof and ΔHf 1, as inFigure 4.

After randomization, the first peak of the thermogramsfor all blends was observed to shift to the right, withrespective increases in the values of Tof and Tpf1, relating tothe significant increase in the content of intermediatemelting point triacylglycerols (S2U and SU2). Also, ΔHf 1values increased after the reaction, indicating increasedparticipation by these triacylglycerol species in the inter-esterified blends. In parallel, interesterification caused adecline in the values of Tpf 2, Tfinal fus and ΔHf 2, resultingfrom the lower percentage of S3 triacylglycerols in theblends. Considering only the interesterified blends, com-parison of the melting thermogram parameters revealed thatthe tendency for Tof, Tpf1, Tpf2, Tfinal fus, ΔHf 1, and ΔHf 2to increase is in complete agreement with the triacylgly-cerol 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.

Microstructure The concept of microstructure comprisesinformation on the state, quantity, shape, size, spatialrelations, and interaction among all the components of thecrystalline network. Microstructure influences fats’ macro-scopic properties enormously44. The microstructural level(or mesoscale) of a fat crystalline network can be defined asthe structures with dimensions between approximately 0.5

Fig. 3 Linear relations among the parameters Toc (°C), Tpc1 (°C),Tfinal crist (°C), ΔHc1 (J/g), and FHSBO content in the original blends

SO:FHSBO Toc (°C) Tpc1 (°C) Tpc2 (°C) ΔHc 1 (J/g) ΔHc 2 (J/g) Tfinal crist (°C)

90:10 33.82 29.47 – 9.47 – 12.52

90:10-I 30.63 22.59 4.76 7.18 13.12 −27.1580:20 39.31 34.63 – 30.13 – 14.82

80:20 I 38.97 34.33 9.97 24.09 26.13 −16.9070:30 39.48 37.33 – 57.83 – 16.38

70:30-I 38.62 34.80 12.46 33.15 32.67 −13.2860:40 43.97 39.33 – 75.86 – 20.02

60:40-I 38.10 34.47 13.80 38.44 53.28 −11.6350:50 45.34 41.30 – 85.81 – 21.03

50:50-I 42.59 34.63 15.04 38.45 40.75 −7.59

Table 1 Crystallization onsettemperature (Toc), crystallizationpeak temperature (Tpc), crystal-lization enthalpy (ΔHc) andcrystallization end temperature(Tfinal crist) of the original andinteresterified blends

I interesterified blend

Food Biophysics (2009) 4:106–118 111

and 200 μm17. Polarized light microscopy is the techniquemost used to visualize the microstructural network of fatsand has been applied with a view to explaining differencesin the texture of fat blends and to show crystalline types andmorphological alterations in crystal growth45. Figure 5shows the SO:FHSBO blends’ before- and after-interesterification crystalline structures obtained by slowcrystallization at 25 °C. The maximum crystal diametervalues of the SO:FHSBO blends, before and after random-ization, are shown in Table 3. The high standard deviationvalues for mean crystal diameter, i.e. the high coefficient ofvariation, are characteristic of crystallized fats whenobserved using polarized light microscopy46,47.

The original blends produced spherulite-shaped crystalswith 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 isdominated by the triacylglycerol species with the highestmelting point in a blend. The presence of large spherulitesor spherulite clusters gives the fat an undesirable grainytexture and is a function mainly of tristearin (StStSt) in theblends48,49. In line with these observations, crystal diameterdiminished with increasing SO concentration in the blends,but crystalline morphology was not altered as a result ofdilution. Rousseau et al.49 also observed this behavior whenthe proportion of liquid oil was increased in palm oil/soybean oil and lard/canola oil blends.

After randomization, the crystal morphology of the90:10, 80:20, 70:30, and 60:40 SO:FHSBO blends wascompletely modified: disk-shaped crystals, with similargranular crystalline structure, were observed50. However,the 50:50 SO:FHSBO blend maintained the tendency tocrystallize predominantly into spherulites after randomiza-tion, although a smaller proportion of disk-shaped crystalswas also observed. This differential behavior probablyrelates to the presence of considerable tristearin content inthe randomized blend, given the 24.50% S3 triacylglycerolcontent in this sample.

Interesterification produced significant reductions incrystal diameter in all blends. According to Herrera et al.51,fats for food product applications should have crystaldiameters of less than 30 μm in order to prevent grainymouthfeel. In all the blends evaluated, interesterificationproved effective in yielding fat bases compatible with usein foods. Increased SO concentration in the randomizedblends was also associated with decreased maximum crystaldiameter, with the exception of the 90:10-I SO:FHSBOblend, which showed dimensions greater than the 80:20-I,70:30-I, and 60:40-I SO:FHSBO blends. This effect may berelated to the longer time required for this sample tocrystallize, due to its low melting point. Also, interester-ification considerably increased the number of crystals inthe 80:20-I, 70:30-I, 60:40-I, and 50:50-I SO:FHSBOblends at 25 °C, as can be seen from the images in

Fig. 4 Linear relations among the parameters Tof (°C), Tpf1 (°C), Tpf2(°C), Tfinal fus (°C), ΔHf1 (J/g), ΔHf2 (J/g), and original blend FHSBOcontent

SO:FHSBO Tof (°C) Tpf 1 (°C) Tpf 2 (°C) ΔHf 1 (J/g) ΔHf 2 (J/g) Tfinal fus(°C)

90:10 −22.20 −17.90 56.77 7.26 20.48 60.11

90:10-I −21.00 −12.50 54.22 13.96 31.14 57.18

80:20 −23.31 −16.40 59.60 7.16 47.13 62.19

80:20-I −23.08 −11.92 57.83 14.80 42.57 61.56

70:30 −23.87 −13.07 63.43 6.11 101.98 65.94

70:30-I −21.04 14.60 58.68 31.58 68.54 61.81

60:40 −23.62 −12.73 64.43 5.90 105.68 66.97

60:40-I −20.57 18.85 58.18 59.87 84.23 61.84

50:50 −23.68 −11.96 65.35 4.09 144.92 67.83

50:50-I −14.06 20.18 52.60 43.75 107.02 62.67

Table 2 Melting onset temper-ature (Tof), melting peak tem-peratures (Tpf), meltingenthalpies (ΔHf) and meltingend temperature (Tfinal fus) of theoriginal and interesterifiedblends

I interesterified blend

112 Food Biophysics (2009) 4:106–118

Fig. 5 Images of crystallization of SO:FHSBO blends a 90:10, b 90:10-I, c 80:20, d 80:20-I, e 70:30, f 70:30-I, g 60:40, h 60:40-I, i 50:50, j50:50-I. I interesterified blend. The bar represents 200 μm

Food Biophysics (2009) 4:106–118 113

Figure 4d, f, h, j. According to Kloek et al.52, a dispersionof fat crystals with a large number of small crystals mayrepresent desirable properties, such as good spreadability. Alarge number of small crystals make harder fat than asmaller number of large crystals, which is also associatedwith undesirable characteristics, such as grainy mouth-feel53.

According to Rousseau et al.49, the alterations in fatmicrostructure caused by interesterification result frommodifications to morphology and the density of thecrystalline network and affect the texture and functionalityof the interesterified bases. The alteration in triacylglycerolcomposition caused by randomization modifies the relativestrengths of the intraparticle bonds among the crystallineelements in a cluster and interparticle bonds amongclusters, leading to the formation of differing structures44.In addition, the change in solid fat content intrinsic to theprocess of interesterification influences how the crystallinenetwork is structured. When solid fat content decreases, asobserved in this study, changes occur towards greatermolecular area and mobility for crystal formation29.Substantial changes in crystalline network density, andcrystal size and morphology, due to randomization have

also been reported by Rodríguez et al.53 and Norizzahet al.54, respectively, for lard/sunflower oil and palmstearin/palm olein blends.

Crystallization Kinetics Study of the crystallization ofinteresterified fat bases is extremely important in order tosuit their use to industrial process constraints and toimprove control of processing stages that involve recrys-tallization of the fat fraction so as to ensure final productquality15.

Figure 6 shows the crystallization isotherms obtained forthe original and interesterified blends at 25 °C. The 90:10-ISO:FHSBO sample did not crystallize at 25 °C. Interester-ification changed curve format for all the blends evaluated.Before the reaction, the curves were hyperbolic. After thereaction, however, the curves showed the sigmoid formcharacteristic of the Avrami model, in which crystallizationtakes place more slowly55. Silva et al.46 also observed asimilar effect on isotherms in the interesterification of lard/soybean oil blends.

Table 4 shows the values of τSFC, SFCmax, and theAvrami parameters (and respective coefficients of determi-nation, R2), indicating velocity (k) and mechanism (n) ofcrystallization, before and after interesterification. Thevalues of SFCmáx and τSFC, particularly, are of significantimportance for outlining applications for interesterified fats.In some specific processes, fats must be completelycrystallized by the end of the production line, in order toensure that crystallization equilibrium is reached. Other-wise, standard processing times have to be adjusted to suitthe characteristics of the fat used55,56.

Addition of FHSBO to soybean oil raised SFCmáx

values. Interesterification led to a decrease in SFCmáx inall blends, an effect associated with reduced percentages ofS3 triacylglycerols and accompanying increases in S2U andSU2 triacylglycerols

35,57. Randomization produced 78.19%,

Fig. 6 Crystallization isothermsat 25 °C for the original andinteresterified blends (I)

Table 3 Maximum crystal diameter of the SO:FHSBO blends beforeand after interesterification

Maximum crystal diameter (μm)

SO: FHSBO Before interesterification After interesterification

90:10 47.52±5.45 18.37±7.88

80:20 52.71±8.13 5.97±2.74

70:30 55.09±7.45 6.63±3.90

60:40 60.79±5.22 9.08±6.44

50:50 116.45±16.20 25.13±15.18

114 Food Biophysics (2009) 4:106–118

69.37%, 45.49%, and 24.01% reductions in SFCmáx valuesfor the 80:20, 70:30, 60:40, and 50:50 SO:FHSBO blends,respectively, in line with the percentage reductions in S3and U3 triacylglycerols. Blend τSFC was significantlymodified by chemical interesterification. Once again, thegreatest alterations were observed in the 80:20 and 70:30SO:FHSBO blends. The decline in crystallization processinduction or nucleation period results primarily fromchanges in solid fat content inherent to the interesterifica-tion process, which influences the formation and structuringof the crystalline network29. In addition, the formation of asmall quantity of partial acylglycerols, such as mono- anddiacylglycerols, as a result of interesterification, can retardonset of nucleation58. At 25 °C, τSFC was 1 to 4 min for theoriginal blends and 4 to 17 min for the interesterifiedblends.

The greater the difference between the sample’s crystal-lization temperature and melting point, the greater thedegree of super-cooling associated with crystallization and,consequently, the smaller the values of τSFC. When thedegree of super-cooling is small, the triacylglycerolmolecules are incorporated into the crystalline structure inthe most appropriate configuration and the most stablepolymorph is formed, because there is time sufficient forthem to be properly oriented. With high degrees of super-cooling, however, triacylglycerol molecules are incorporat-ed into the crystalline surface very quickly, and thereforeimperfectly, resulting in the formation of mixed crystals andless stable polymorphs, which may persist indefinitely30,33.Therefore, although the percentage increase in lowermelting point, mixed triacylglycerols (represented by theS2U and SU2 species) caused by randomization contributesto higher τSFC values, the interesterified blends areassociated with greater stability and polymorphic homoge-neity at 25 °C, which is of significant importance for foodapplications59,60.

The Avrami constant, k, which is primarily a function ofcrystallization temperature, decreased with increased pro-portion of SO in the original and interesterified blends,

indicating that the velocity of crystallization was lower as aresult of dilution of the FHSBO with liquid oil, increasedU3 triacylglycerol content and reduced S3 triacylglycerolcontent. This effect was also observed by Cerdeira et al.61

in studying the crystallization of milkfat/sunflower oilblends. Interesterification produced a drop in the crystalli-zation velocity of the samples, related mainly to thedecrease in percentage S3 triacylglycerol content

62.The Avrami exponent, n, is sensitive to the crystalliza-

tion mechanism, with respect to nucleation process anddimensions of growth. Nucleation may be sporadic orinstantaneous, and crystal growth may occur in one, two, orthree dimensions, characterizing the formation of needle-,disk-, or spherulite-shaped crystals, respectively33. Valuesof n=3 correspond to spherulitic growth from instantaneousnuclei or disk-shaped growth from sporadic nuclei, whilevalues of n=2 denote needle-shaped growth from sporadicnuclei or disk-shaped growth from instantaneous nuclei33,62.Although n should be a whole number, fractional values areusually obtained. This anomaly relates primarily to theformation of crystals of similar morphology from differentnucleation types (sporadic and instantaneous). Fractionalvalues of the exponent, n, may also be explained by thesimultaneous development of crystals with differentmorphologies62.

The value of n for the original blends was 3.59 to 2.88.According to Metin and Hartel63, values of n increase withdecreasing melting point, i.e. with decreasing degree ofsuper-cooling of the sample, which agrees with the resultsof this study. Observation of the original samples’crystalline structure (Figure 5), which is made up ofspherulites, corroborated the n values characteristic ofspherulitic growth from instantaneous nuclei. The inter-esterified blends gave n values of 2.40 to 2.60. Values of nfrom 2 to 3 reflect disk-shaped growth, as observed bypolarized light microscopy for the 80:20-I, 70:30-I, and60:40-I SO:FHSBO blends, denoting mixed nucleation,where crystals of the same morphology arise fromsporadic and instantaneous nuclei. However, the value of

SO:FHSBO τSFC (min) SFCmáx (%) k (min−n) n R2

90:10 4 10.83 0.0011 3.59 0.9328

90:10-Ia – – – – –

80:20 3 20.04 0.0014 3.38 0.8825

80:20-I 17 4.37 0.0001 2.56 0.7314

70:30 3 30.00 0.0025 2.79 0.8943

70:30-I 8 9.19 0.0005 2.60 0.6435

60:40 3 36.36 0.0035 2.86 0.9027

60:40-I 5 19.82 0.0010 2.60 0.9069

50:50 1 49.77 0.0039 2.88 0.9773

50:50-I 4 37.82 0.0020 2.40 0.8493

Table 4 Induction period(τSFC), maximum solid fat con-tent (SFCmáx), Avrami constant(k), Avrami exponent (n) andrespective coefficients of deter-mination (R2) for SO:FHSBOblends before and afterinteresterification

I interesterified blenda The 90:10-I SO:FHSBO sampledid not crystallize at 25 °C

Food Biophysics (2009) 4:106–118 115

n=2.4 for the 50:50 SO:FHSBO-I blend, where spheruliteswere the predominant crystal type (Figure 5), is possiblyrelated to the concomitant presence of disk-shapedcrystals, with both crystal types originating from instanta-neous nucleation33,62. The results thus show that chemicalinteresterification modified the mechanism of crystalliza-tion of the original blends, altering both crystallinemorphology and nucleation mechanism of the blends withfrom 20% to 40% FHSBO content; and with simultaneousdevelopment of crystals with different morphologies, butthe same nucleation mechanism, in the 50% FHSBOblend.

Polymorphism X-ray diffraction is often used as a tech-nique to evaluate chemical interesterification, helpingoutline applications for the fat bases produced64. Diffrac-tion patterns of natural or interesterified fats show broaderpeaks than pure compounds, due to the presence of multipletriacylglycerols in the cell units accompanied by liquidoils49.

Diffractograms of the blends before and after chemicalinteresterification are shown in Figure 7. The 90:10 SO:

FHSBO and 90:10-I SO:FHSBO samples were liquids at25 °C, and thus impossible to immobilize on the analysissupport. The original blends produced peaks of strongintensity at 4.6 Å and lesser intensity at 3.7 and 3.8 Å,which is characteristic of the β form. However, the 60:40and 50:50 SO:FHSBO blends also produced a low-intensitydiffraction line at 4.2 Å, which is associated with thepresence of the β′ polymorph. The interesterified samplesshowed two peaks of varying intensity at 4.2 and 3.8 Å,which correspond to the occurrence of the β′ polymorph, aswell as a low-intensity peak at 4.6 Å. Due to the lower solidfat content of the 80:20-I SO:FHSBO sample at 25 °C, thediffractogram at this temperature of analysis showed low-intensity peaks.

Table 5 details the polymorphic changes produced byblend and by interesterification. The original blendsdisplayed β-type polymorphic behavior. The 80:20 and70:30 SO:FHSBO samples showed exclusively β-type.In the 60:40 and 50:50 SO:FHSBO samples, however,the presence of a small quantity of β′ crystals wasobserved, although SO and FHSBO display a well-known tendency to crystallize in the β form, which

Fig. 7 X-ray diffraction patterns for SO:FHSBO samples crystallized at 25 °C for 24 h before (a) and after (b) interesterification

Short spacings (Å)

SO:FHSBO 4.8 4.6 4.2 3.8 3.7 Polymorphic form

80:20 4.83 (vw) 4.60 (s) 3.86 (m) 3.70 (m) β

70:30 4.82 (vw) 4.59 (s) 3.85 (m) 3.69 (m) β

60:40 4.78 (vw) 4.56 (s) 4.19 (w) 3.77 (m) 3.67 (m) β> >β′

50:50 4.84 (vw) 4.60 (s) 4.22 (w) 3.79 (m) 3.70 (m) β> >β′

80:20-I 4.57 (w) 4.24 (m) 3.84 (m) β′> >β

70:30-I 4.63 (w) 4.23 (m) 3.84 (m) β′> >β

60:40-I 4.58 (w) 4.23 (s) 3.83 (m) β′> >β

50:50-I 4.62 (w) 4.21 (s) 3.80 (m) β′> >β

Table 5 Polymorphic formsand short spacings of the originaland interesterified SO:FHSBOblends

v very, w weak, m medium,s strong

116 Food Biophysics (2009) 4:106–118

is associated with the low diversity of fatty acidcomposition and relatively homogeneous triacylglycerolcomposition59,65.

Zeitoun et al.39, studying the polymorphic behavior of a50:50 SO:FHSBO blend, observed the simultaneouspresence of β′ crystals (24.7%) and β crystals (75.3%).Narine and Humphrey48 also found the β′ polymorph in a75:25 SO:FHSBO blend. DeMan et al.66 report the β′polymorph in fully hydrogenated canola oil, whichcharacteristically tends to β stabilize. Humphrey et al.42,evaluating polymorphic behavior in soybean oil/fullyhydrogenated canola oil blends, found that the poly-morphic forms varied with the concentration of fullyhydrogenated canola oil in the samples. According toMetin and Hartel30 and Kloek et al.52, the persistence of β′crystals in cases where stabilization is preferentiallyβ-type probably results from the formation of mixedcrystals, which may be favored by the triacylglycerolcomposition of the blend and by the tempering protocolused, particularly when a high degree of super-coolingoccurs. According to McGauley and Marangoni33, twofactors relate to this phenomenon: the unfavorable config-uration of triacylglycerol molecules resulting from rapidnucleation and/or rapid increase in viscosity, which limitsheat and mass transfer and hinders molecular arrangement.Therefore, the formation of mixed crystals may beassociated with the presence of the β′ polymorph in the60:40 and 50:50 SO:FHSBO blends.

Interesterification modified the crystalline habit of thefats evaluated towards significant predominance of the β′polymorph. The β′ crystals were small and their morphol-ogy appropriate to the plasticity characteristics desirable inproducts such as margarines, shortenings, and fats forbakery and confectionary products. Conversely, the βpolymorphic form tends to produce broad granular crystals,resulting in grainy products with low aeration potential,which may impair the macroscopic properties of somefoods13. This result agrees with the microstructure of theblends observed by polarized light microscopy before andafter randomization (Figure 5).

According to Rousseau and Marangoni35, stabilizationof the β′ polymorph caused by chemical interesterificationis associated with the formation of triacylglycerols withgreater variation in chain length. This results in moredisorderly packing of the terminal methyl groups, conse-quently forming less dense crystalline structures. Also, inall the blends evaluated, randomization caused a reductionin S3 triacylglycerol content for the FHSBO, representedmainly by tristearin, triacylglycerol characteristic of βpolymorph stabilisation67. List et al.38,68 evaluated inter-esterification of various oils with hardstocks: the reactionalso favored formation of the β′ structure, enabling thebases produced to be applied in margarines.

Conclusion

A comprehensive understanding of the functions andproperties of fats or oil bases produced by interesterificationis essential to outlining applications for them and obtainingfood products with the desired final attributes. This studymade it possible to determine that the thermal behavior,microstructure, crystallization kinetics, and polymorphismcharacteristics of SO:FHSBO blends were significantlyaltered by increased FHSBO concentration and the chemicalinteresterification process. In addition, these propertiesproved to be directly related to the blends’ triacylglycerolcomposition before and after randomization and, takentogether, can determine their applicability.

Acknowledgments The authors would like to thank the BrazilianAgencies FAPESP and CNPq for the financial support.

References

1. D.R. Erickson, Practical Handbook of Soybean Processing andUtilization. (American Oil Chemists’Society, Champaign, 1995),p. 227

2. A.P.B. Ribeiro, J.M.L.N. Moura, R. Grimaldi, L.A.G. Gonçalves,Quim Nova 30, 1295 (2007)

3. R.D. O’Brien, Fats and Oils—Formulating and Processing forApplications. (CRC, New York, 2004)

4. A.G. Marangoni, D. Rousseau, Trends. Food Sci. Technol. 6, 329(1995). doi:10.1016/S0924-2244(00)89167-0

5. R. Rao, B.R. Lokesh, Lipids 38, 913 (2003). doi:10.1007/s11745-003-1144-3

6. J.E. Hunter, Nutr. Res. 25, 499 (2005). doi:10.1016/j.nutres.2005.04.002

7. M.A. Dirienzo, Lipids 43, 451 (2008). doi:10.1007/s11745-008-3173-6

8. I. Foubert, K. Dewettinck, G. Janssen, P.A. Vanrolleghem, J. FoodEng. 75, 551 (2006). doi:10.1016/j.jfoodeng.2005.04.038

9. S.S. Narine, K.L. Humphrey, L. Bouzidi, J. Am. Oil Chem. Soc.83, 913 (2006). doi:10.1007/s11746-006-5046-6

10. I. Piska, M. Zárubová, T. Louzecký, H. Karami, V. Filip, J. FoodEng. 77, 347 (2006). doi:10.1016/j.jfoodeng.2005.07.010

11. A.M.D.N. Lida, K. Sundram, W.L. Siew, A. Amina, S. Mamot,J. Am. Oil Chem. Soc. 79, 1138 (2002)

12. D. Rousseau, S. Sonwai, Food Biophys. 3, 273 (2008).doi:10.1007/s11483-008-9061-z

13. K. Sato, Chem. Eng. Sci. 56, 2255 (2001). doi:10.1016/S0009-2509(00)00458-9

14. O.W.D. Gamboa, L.A. Gioielli, Quim Nova 29, 646 (2006).doi:10.1590/S0100-40422006000400004

15. P. Wassel, N.W.G. Young, Int. J. Food Sci. Technol. 42, 503(2007). doi:10.1111/j.1365-2621.2007.01571.x

16. S.S. Narine, A.G. Marangoni, Food Biophys. 3, 113 (2008).doi:10.1007/s11483-007-9049-0

17. S.S. Narine, A.G. Marangoni, in Fat Crystal Networks, ed. byA.G. Marangoni (Marcel Dekker, New York, 2005), p. 179

18. F.F.A. Hollander, D. Kaminski, D. Duret, W.J.P. van Enckevort,H. Meekes, P. Bennema, Food Res. Int. 35, 909 (2002).doi:10.1016/S0963-9969(02)00153-9

19. A.G. Marangoni, S.S. Narine, Food Res. Int. 35, 957 (2002).doi:10.1016/S0963-9969(02)00158-8

Food Biophysics (2009) 4:106–118 117

20. S. Martini, T. Awad, A.G.Marangoni, inModifying Lipids for Use inFood, ed. by F.D. Gunstone (Woodhead Publishing, Cambridge,2006), p. 142

21. M.L. Herrera, F.J. Marquez Rocha, J. Am. Oil Chem. Soc. 73, 321(1996). doi:10.1007/BF02523425

22. J. Oh, A.R. McCurdy, S. Clark, B.G. Swanson, J. Am. Oil Chem.Soc. 82, 13 (2005). doi:10.1007/s11746-005-1036-y

23. Y.B. Che Man, K. Shamsi, M.S.A. Yusoff, S. Jinap, J. Am. OilChem. Soc. 80, 409 (2003). doi:10.1007/s11746-003-0713-1

24. J.M.N. Marikkar, H.M. Ghazali, Y.B. Che Man, O.M. Lai, J. FoodLipids. 10, 63 (2003). doi:10.1111/j.1745-4522.2003.tb00006.x

25. R. Grimaldi, L.A.G. Gonçalves, M.Y. Ando, Quim Nova 28, 633(2005). doi:10.1590/S0100-40422005000400015

26. N.R. Antoniossi Fillho, O.L. Mendes, F.M. Lanças, Chromatographia40, 557 (1995). doi:10.1007/BF02290268

27. AOCS, Official methods and recommended practices of theAmerican Oil Chemists’ Society. (American Oil Chemists’Society, Champaign, 2004)

28. C.G. Biliaderis, Food Chem. 10, 239 (1983). doi:10.1016/0308-8146(83)90081-X

29. C. Himavan, V.M. Starov, A.G.F. Stapley, Adv. Colloid. InterfaceSci. 122, 3 (2006). doi:10.1016/j.cis.2006.06.016

30. S. Metin, R.W. Hartel, in Bailey’s Industrial Oil and Fat Products,ed. by F. Shahidi (Wiley Interscience, New York, 2005)

31. J.F. Toro-Vazquez, E. Dibildox-Alvarado, M. Charó-Alonso, V.Charó-Alonso, C.A. Gómez-Aldapa, J. Am. Oil Chem. Soc. 79,855 (2002). doi:10.1007/s11746-002-0570-y

32. A.J. Wright, R.W. Hartel, S.S. Narine, A.G. Marangoni, J. Am.Oil Chem. Soc. 77, 463 (2000). doi:10.1007/s11746-000-0075-8

33. S.E. McGauley, A.G. Marangoni, in Physical Properties ofLipids, ed. by A.G. Marangoni, S.S. Narine (CRC, New York,2002)

34. P.H. Yap, J.M. deMan, L. deMan, J. Am. Oil Chem. Soc. 66, 693(1989). doi:10.1007/BF02669954

35. D. Rousseau, A.G. Marangoni, in Food Lipids: Chemistry,Nutrition, and Biotechnology, ed. by C.C. Akoh (CRC, NewYork, 2002), p. 301

36. A. Szydlouska-Czerniak, G. Karlovits, M. Lach, E. Szlyk, FoodChem. 92, 133 (2005). doi:10.1016/j.foodchem.2004.07.010

37. M. Buchgraber, F. Ulberth, H. Emons, E. Anklan, Eur. J. LipidSci. Technol. 106, 621 (2004). doi:10.1002/ejlt.200400986

38. G.R. List, T.L. Mounts, F. Orthoefer, W.E. Neff, J. Am. Oil Chem.Soc. 72, 379 (1995). doi:10.1007/BF02541100

39. M.A.M. Zeitoun, W.E. Neff, T.L. Mounts, J. Am. Oil Chem. Soc.70, 467 (1993). doi:10.1007/BF02542577

40. C.P. Tan, Y.B. Che Man, Food Chem. 76, 89 (2002). doi:10.1016/S0308-8146(01)00241-2

41. K.A. Al-Rashood, R.R.A. Abou-Shaaban, E.M. Abdel-Moety, A.Raulf, J. Am. Oil Chem. Soc. 73, 303 (1996). doi:10.1007/BF02523423

42. K.L. Humphrey, P.H.L. Moquin, S.S. Narine, J. Am. Oil Chem.Soc. 80, 1175 (2003). doi:10.1007/s11746-003-0838-2

43. K.L. Humphrey, S.S. Narine, Food Res. Int. 37, 11 (2004).doi:10.1016/j.foodres.2003.09.012

44. Y. Shi, B. Liang, R.W. Hartel, J. Am. Oil Chem. Soc. 82, 399(2005). doi:10.1007/s11746-005-1084-3

45. L.A. Gioielli, I.S. Simões, J.N. Rodrigues, J. Food Eng. 57, 347(2003). doi:10.1016/S0260-8774(02)00355-2

46. R.C. Silva, J.P. Escobedo, L.A. Gioielli, Quim Nova 31, 330 (2008)47. D. Rousseau, A.R. Hill, A.G. Marangoni, J. Am. Oil Chem. Soc.

83, 377 (2006). doi:10.1007/s11746-006-1216-948. S.S. Narine, K.L. Humphrey, Food Res. Int. 37, 28 (2004).

doi:10.1016/j.foodres.2003.09.01349. D. Rousseau, A.G. Marangoni, K.R. Jeffrey, J. Am. Oil Chem.

Soc. 75, 1833 (1998). doi:10.1007/s11746-998-0339-650. R. Campos, in Fat Crystal Networks, ed. by A.G. Marangoni

(Marcel Dekker, New York, 2005), p. 26751. M.L. Herrera, C. Falabella, M. Melgarejo, M.C. Anón, J. Am. Oil

Chem. Soc. 75, 1273 (1998). doi:10.1007/s11746-998-0172-y52. W. Kloek, P. Walstra, T. van Vliet, J. Am. Oil Chem. Soc. 77, 48

(2000)53. A. Rodríguez, E. Castro, M.C. Salinas, R. López, M. Miranda,

J. Am. Oil Chem. Soc. 78, 432 (2001)54. A.R. Norizzah, C.L. Chong, C.S. Cheow, O. Zaliha, Food Chem.

86, 229 (2004). doi:10.1016/j.foodchem.2003.09.03055. M.L. Herrera, C. Falabella, M. Melgarejo, M.C. Anón, J. Am. Oil

Chem. Soc. 75, 1 (1999). doi:10.1007/s11746-999-0039-x56. A. Bell, M.H. Gordon, W. Jirasubkunakorn, K.W. Smith, Food

Chem. 101, 799 (2007). doi:10.1016/j.foodchem.2006.02.03357. D. Rousseau, A.R. Hill, A.G. Marangoni, J. Am. Oil Chem. Soc.

72, 983 (1996). doi:10.1007/BF0252340558. M.L. Herrera, M.L. Gatti, R.W. Hartel, Food Res. Int. 32, 289

(1999). doi:10.1016/S0963-9969(99)00083-659. L.H. Wiedermann, J. Am. Oil Chem. Soc. 55, 823 (1978).

doi:10.1007/BF0268265560. S. Martini, M.L. Herrera, R.W. Hartel, J. Am. Oil Chem. Soc. 79,

1055 (2002). doi:10.1007/s11746-002-0603-661. M. Cerdeira, S. Martini, R.J. Candal, M.L. Herrera, J. Am. Oil

Chem. Soc. 83, 489 (2006). doi:10.1007/s11746-006-1231-x62. A.G. Marangoni, in Fat Crystal Networks, ed. by A.G. Marangoni

(Marcel Dekker, New York, 2005), p. 163. S. Metin, R.W. Hartel, J. Am. Oil Chem. Soc. 75, 1617 (1998).

doi:10.1007/s11746-998-0102-z64. Z. Omar, C.C. Let, C.C. Seng, N.A. Rashid, Eur. J. Lipid Sci.

Technol. 107, 634 (2005). doi:10.1002/ejlt.20050118065. I. Foubert, K. Dewettinck, D. Van de Walk, A.J. Dijkstra,

P.J. Quinn, in The Lipid Handbook, ed. by F.D. Gunstone, J.L.Harwood, A.J. Dijkstra (CRC, Boca Raton, 2007), p. 471

66. L. deMan, J.M. deMan, B. Blackman, J. Am. Oil Chem. Soc. 66,1777 (1989). doi:10.1007/BF02660746

67. B.S. Ghotra, S.D. Dyal, S.S. Narine, Food Res. Int. 35, 1015(2002). doi:10.1016/S0963-9969(02)00163-1

68. G.R. List, E.A. Emken, W.F. Kwolek, H.J. Dutton, J. Am. OilChem. Soc. 74, 408 (1977). doi:10.1007/BF02671021

118 Food Biophysics (2009) 4:106–118