Food Chemistry 145 (2014) 725–735

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Effects of emulsion droplet sizes on the crystallisation of milk fat

0308-8146/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.foodchem.2013.08.072

⇑ Corresponding author. Tel.: +61 7 33469192; fax: +61 7 336511177.E-mail addresses: t.truong1@uq.edu.au (T. Truong), n.bansal@uq.edu.au

(N. Bansal), rsharma@dairyinnovation.com.au (R. Sharma), mpalmer@dairyinnovation.com.au (M. Palmer), b.bhandari@uq.edu.au (B. Bhandari).

Tuyen Truong a, Nidhi Bansal a, Ranjan Sharma b, Martin Palmer b, Bhesh Bhandari a,⇑a School of Agriculture & Food Sciences, The University of Queensland, St Lucia, QLD 4072, Australiab Dairy Innovation Australia Limited, Werribee, VIC 3030, Australia

a r t i c l e i n f o

Article history:Received 5 April 2013Received in revised form 2 August 2013Accepted 16 August 2013Available online 27 August 2013

Keywords:Cryo-TEMCrystallisationEmulsifierMilk fatNanoemulsionsOleinParticle sizeStearin

a b s t r a c t

The crystallisation properties of milk fat emulsions containing dairy-based ingredients as functions ofemulsion droplet size, cooling rate, and emulsifier type were investigated using a differential scanningcalorimeter (DSC). Anhydrous milk fat and its fractions (stearin and olein) were emulsified with wheyprotein concentrate, sodium caseinate, and Tween80 by homogenisation to produce emulsions in varioussize ranges (0.13–3.10 lm). Particle size, cooling rate, and types of emulsifier all had an influence on thecrystallisation properties of fat in the emulsions. In general, the crystallisation temperature of emulsifiedfats decreased with decreasing average droplet size and was of an exponent function of size, indicatingthat the influence of particle size on crystallisation temperature is more pronounced in the sub-micronrange. This particle size effect was also verified by electron microscopy.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Crystallisation of lipids in emulsions is of interest not only infoods but also pharmaceuticals, nutraceuticals, cosmetics, petro-leum, and agrochemicals. Many studies on the crystallisation oflipids in food emulsions and fat-containing food products haveshown that crystallisation behaviour, expressed as crystallisationtemperature, crystal polymorphism, solid fat content and/or lipiddestabilisation, is influenced by numerous factors such as the nat-ure of oil and aqueous phases, processing conditions, thermaltreatment (cooling/heating rate and temperature), and tempera-ture history (Lopez et al., 2002a; Lopez, Bourgaux, Lesieur, & Olli-von, 2002b; Martini & Tippetts, 2008; Palanuwech & Coupland,2003). The emulsion droplet size is also known to impact crystalli-sation properties of dispersed fat phase (Montenegro, Antonietti,Mastai, & Landfester, 2003; Walstra, van Vliet, & Kloek, 1995).Owing to the emulsified state in which each droplet needs a nu-cleus or impurity to initiate nucleation, extensive supercooling isrequired for lipid droplets in a dispersed system to induce crystal-lisation compared to its bulk counterpart (Walstra et al., 1995).Variations in emulsion droplet size may alter the crystallisationand structural behaviour of emulsified fats, particularly with

‘‘nanoemulsions’’ (droplet diameter below 0.2 lm) (Bugeat et al.,2011; Bunjes, Koch, & Westesen, 2000; Gulseren & Coupland,2007; Lopez et al., 2001; McClements, Dungan, German, Simoneau,& Kinsella, 1993; Montenegro et al., 2003). It was reported thatsubmicron-size droplets exhibited at lower crystallisation temper-ature than micron-size ones in n-hexadecane oil-in-water emul-sions (Dickinson, Mcclements, & Povey, 1991) whereascrystallisation temperature of tripalmitin, tristearin, and trilauroyl-glycerol nanoparticles just slightly changed with decreasing aver-age size in nanometre-size range (Bunjes et al., 2000; Higami,Ueno, Segawa, Iwanami, & Sato, 2003). Using edible emulsions, Lo-pez, Bourgaux, Lesieur, Bernadou, et al. (2002), Lopez, Bourgaux,Lesieur, & Ollivon (2002) observed that crystallisation temperatureof b-lactoglobulin-stabilised milk fat emulsions decreased withdecreasing droplet size (d32 1.25–0.38 lm) at a cooling rate of1 �C min�1. A recent study of Bugeat et al. (2011) showed thatthere was a slight decrease in melting enthalpy in small dropletsize (0.18 lm) versus larger droplet size (1.67 lm) of unsaturatedfatty acids enriched milk TAGs in emulsion state.

Milk lipids, which comprise 98% triacylglycerols (TAGs), areregarded as the most complex of edible fats owing to the wide var-iation in both TAG structure and fatty acid composition (Christie,1995). The melting range of milk lipids spans from around �40to 40 �C, typically giving three overlapping melting curve endo-therms (Christie, 1995; Lopez, Bourgaux, Lesieur, Riaublanc, &Ollivon, 2006). These correspond to the three incompletely misci-ble, low-, middle-, and high-melting point fractions of milk fat.

726 T. Truong et al. / Food Chemistry 145 (2014) 725–735

Single step fractionation of milk fat produces two primary fractions– stearin (hard) and olein (soft). As in initial milk fat, TAGs accountfor more than 98% of both stearin and olein fractions; however, theTAG components are different, with the stearin fraction comprisingmainly long-chain fatty acids while the olein fraction is enriched inshort-chain and unsaturated fatty acids (Lopez et al., 2006). Thereis a renewed interest in employing olein-rich dairy fractions asfood ingredients (Bazmi & Relkin, 2009) and despite the higherproportion of saturated fatty acids, it has been proposed that thestearin fraction might be particularly suitable for encapsulatingthe bioactive components in some functional food ingredients(Relkin, Yung, Kalnin, & Ollivon, 2008). The crystallisation proper-ties of milk fat unsaturated fatty acid-enriched emulsions uponstorage at 4 �C have been reported recently (Bugeat et al., 2011)but apart from this, there appear to be little information on thecrystallisation behaviour of milk fat nanoemulsions or even moregenerally on the effects of droplet size on lipid crystallisation inmilk fat emulsions. A better understanding of droplet size-depen-dent crystallisation properties of milk fat emulsions, particularlynanoemulsions, is needed in order to evaluate the potential indus-trial application of these materials, for example as food texturemodifying agents. New knowledge in this area may also find appli-cation in encapsulation technology, where crystallised fat nanopar-ticles in the form of solid lipid nanoparticles or nanostructuredlipid carriers could be used as carriers to encapsulate and deliverlipid-soluble nutrients (Relkin et al., 2008).

This study aimed to examine the effect of emulsion droplet size,from micron- to nanometric-scale, on the crystallisation behaviourof dairy-based oil-in-water emulsions. The dairy ingredients usedwere anhydrous milk fat, stearin or olein (oil phase) in combina-tion with whey protein or sodium caseinate (aqueous phase).Tween80 was also used to investigate the effect of a small molec-ular weight surfactant in comparison with the macromolecularones (whey protein concentrate and sodium caseinate). In thisstudy, Differential Scanning Calorimetry (DSC) was employed tomonitor changes in thermal properties and phase transitions inthe emulsion systems at various size ranges during cooling at dif-ferent rates. In addition to determining crystallisation tempera-tures, solid fat contents of the dairy-based emulsions were alsoderived from the DSC thermograms.

2. Materials and methods

2.1. Materials

2.1.1. Anhydrous milk fat (AMF) and its primary fractions (stearin andolein)

Dry fractionation of melted milk fat: In this work, three kinds ofdairy-based fat/oil, i.e. anhydrous milk fat (AMF) and its two frac-tions (St: stearin and O: olein), were employed as the oil phase.Stearin and olein fractions were obtained by dry fractionation;i.e. crystallisation from melted milk fat, in a single step procedure.Anhydrous milk fat (AMF) was supplied by Tatura Milk IndustriesLimited (Tatura, Victoria, Australia). According to the manufac-turer, the AMF contained 99.92% butter fat, 0.08% moisture, 0.13%free fatty acid (expressed as oleic acid), with a peroxide value of0.1 meq 02/kg fat. The AMF was placed in a stainless steel containerand melted at 60 �C for 2 h in recirculated water bath. A completelymelted milk fat at 60 �C was filtered through a Whatman #4 filterpaper under vacuum in a Buchner funnel to remove proteinaceousmaterial (adapted from Kaylegian & Lindsay, 1995). The meltedmilk fat was then allowed to cool to 21 �C in an incubator (coolingrate recorded �0.2 �C min�1) and held at crystallisation tempera-ture (21 �C) for the total residence time of 24 h (at 21 �C) withoutagitation. The liquid olein fraction was separated from the solid

stearin fraction by vacuum filtration, using a Buchner funnel. Theoriginal milk fat and fractionated samples were kept at 4 �C beforeuse.

Determination of fatty acid composition: To quantify fatty acidcomponents, stearin, AMF, and olein were transmethylated foranalysis as fatty acid methyl esters (FAMEs) by gas chromatogra-phy (GC). For each sample, heptanoic acid (1 mg) was used as aninternal standard that was added to a test tube containing approx-imately 15–20 mg of the lipid sample. Addition of 0.5 mL methano-lic NaOH to each sample was undertaken, following by flushingsamples with nitrogen gas and covering tubes with Sub-Seals (Sig-ma–Aldrich). The covered tubes were placed on a steam bath at95 �C for 3 min to undergo saponification of the lipid sample. Aftercooling step, 2.5 mL of boron trifluoride (BF3) in methanol (14%,v/v) were added to the cooled tubes. Esterification of the fatty acidwas done by heating tubes again on the steam bath for 5 min andthen allowed to cool to room temperature. Subsequent treatmentsincluded addition of 2.0 mL of heptane and 5.0 mL of saturated so-dium chloride solution, and agitation using a vortex mixer. Theheptane added test tubes were then allowed to settle to obtaintwo layers. Clear heptane solution (1.5 mL) was recovered intothe auto sampler vial.

Derivatisation of fatty acids was carried out by GC on a DB-2360 m � 0.25 mm � 0.25 lm capillary column using a flame ionisa-tion detector (GC-17A, Shimadzu Co., Japan). The injection port wasmaintained at 250 �C, detector at 285 �C, and the carrier gas heliumat 110 kPa, linear velocity 2.2 mL min�1. The column temperaturegradient started at 100 �C, rising to 180 �C at 10 �C min�1, and then240 �C at 5 �C min�1. Identification of methyl esters of specific fattyacids was done by referencing their retention times to those of anauthentic standard of methylated fatty acids (Nu-Chek Prep Inc.,Minnesota, USA). Quantification of fatty acid composition wasdetermined according to the area of the internal standard peak.Average values were calculated out of determinations of fatty acidcompositions of two independent batches. Among the three sam-ples, the stearin fraction had highest level of palmitic (C16:0) andstearic (C18:0) acids (42.4% C16:0, 14.5% C18:0), followed by theAMF (37.6% C16:0, 11.9% C18:0). The olein fraction was enriched inoleic acid (30.3% C18:1) while a lower proportion of unsaturatedfatty acids was found in the AMF and stearin samples, 28.0% and21.8% of C18:1, respectively. Full details of fatty acid compositionof the stearin, AMF, and olein can be found in Supplementary data.

2.1.2. EmulsifiersDairy-based emulsifiers including sodium caseinate (NaCN:

92.6% protein, 0.25% lactose–casein, 0.7% fat, 1.2% sodium) andwhey protein concentrate (WPC80: 80% protein, 6% lactose and7% fat) were obtained from Murray Goulburn Co-op (Melbourne,Victoria, Australia). Tw80 (Tween80 LR CAS # 9008-65-5; polyoxy-ethelene sorbitan monooleate C64H124O26; Labtek Pty Ltd., Bren-dale, Queensland, Australia) was also chosen to investigate theeffect of a small molecular weight surfactant on crystallisationbehaviour of dairy-based emulsions.

2.2. Preparation of dairy-based emulsions

The aqueous phase (90% wt.) of each sample comprised of 2%emulsifiers (NaCN, WPC or Tw80) dispersed in distilled water withthe addition of 0.02% sodium azide as an antimicrobial agent. Theaqueous phase was stirred using an overhead mixer for 2 h andstored overnight to allow complete protein hydration. The pH ofthe aqueous phase was then adjusted to 6.7 by adding aliquots of1 M NaOH solution. Before mixing with the aqueous phase,10% wt. of oil phase (AMF, stearin, or olein) was heated well abovetheir highest melting points (as measured by DSC) for 30 min at55–60 �C to completely melt the fat. The aqueous phase was also

T. Truong et al. / Food Chemistry 145 (2014) 725–735 727

brought to the same temperature as the heated milk fat and itsfractions. It should be noted that these temperatures were belowthe temperature of protein denaturation at pH 6.5, i.e. 75.9 �C (b-lactoglobulin), 61 �C (a-lactalbumin) and 71.9 �C (bovine serumalbumin) (Bernal & Jelen, 1985).

Coarse emulsions were prepared using a rotor–stator system(Model L2R, Silverson Machines Ltd., UK) at 5000 rpm for 2 minat 55–60 �C to pre-homogenise the oil phase into the aqueousphase. The coarse emulsions were then homogenised using a labo-ratory two-stage valve homogeniser (Twin Panda, GEA Niro Soavi,Parma, Italy) at different homogenising pressures (3.0, 5.0, 10.0,15.0, 25.0, and 35.0 MPa) and number of cycles (3–5) to obtain var-ious average droplet sizes. The emulsion samples were then cooledto room temperature and subjected to size measurement and ther-mal analysis within 24 h.

In total, there were 9 dairy-based emulsions (3 fats/oils � 3emulsifiers) examined in this study. Each kind of dairy-basedemulsion was abbreviated by the name of emulsifier-oil phaseused. They are WPC-St, NaCN-St, Tw80-St; WPC-AMF, NaCN-AMF, Tw80-AMF; and WPC-O, NaCN-O, Tw80-O. Two independentreplicates for all of the emulsions were prepared.

2.3. Analyses

2.3.1. Measurement of oil droplet sizeEmulsion droplet size and size distribution analyses were under-

taken using a laser light scattering analyser (Malvern Mastersizer2000, Malvern Instruments Ltd., Worcestershire, UK). Aliquots ofemulsions were added to distilled water under stirring(2000 rpm) until an obscuration rate of 10% was gained. The calcu-lation of size distribution was performed by Malvern software usingthe Mie theory which requires refractive indices (milk fat: 1.462,water: 1.330) and assumes sphericity of particles. In this study,the mean particle size was reported as modal diameter. The modeis defined as a maximum size frequency of the size distribution.

2.3.2. Thermal analysisA differential scanning calorimetry (DSC1 STARe System, Met-

tler-Toledo, Schwerzenbach, Switzerland) was used to monitorcrystallisation behaviour of the dairy-based emulsions at varioussize ranges in non-isothermal mode. Calibration was undertakenusing an indium standard, melting point = 156.66 �C, DH melt-ing = 28.41 J g�1. Approximately 10–15 mg of the dairy-basedemulsions were introduced to a 40 lL aluminium pan (00026763,Mettler-Toledo, Schwerzenbach, Switzerland) and hermeticallysealed (an empty pan was used as a reference). Since the emulsionformulation consisted of 10% of oil component, the dry mass of to-tal triacylglycerols in sample was within a range of 1.0–1.5 mg.Heating–cooling-reheating protocol was applied. Samples wereheated from 25 to 50 �C (heating rate 20 �C min�1) and held for5 min to eliminate thermal history, then cooled to 0.5 �C (stearin-based emulsions), �4 �C (AMF-based emulsions), or �8 �C (olein-based emulsions) at three cooling rates (1, 5, and 10 �C min�1).Reheating to 50 �C was undertaken at a fixed heating rate of2 �C min�1 to determine final melting point (TM-endset). In order tocompare with bulk phase, DSC measurements of bulk fats (AMF,stearin, and olein) were also carried out at corresponding scanningprograms of their emulsion counterparts. STARe Excellence Soft-ware (Mettler-Toledo, Schwerzenbach, Switzerland) was used toobtain calorimetric parameters, including initial crystallisationtemperature (TC-onset, �C; signal of first heat flow release), finalcrystallisation temperature (TC-endset, �C; completion of crystallisa-tion process); and heat of crystallisation (DH, J g�1). Plotting ofTC-onset versus emulsion droplet size and non-linear regressions fit-ting to correlate their relationship were undertaken with Sigma-Plot 11 software (Systat Software Inc., San Jose, CA, USA).

To analyse the proportion of solid fat increase in emulsified fatdroplets during crystallisation, calculation of total area under theDSC cooling curves was performed using sequential peak integra-tion between exothermal heat flow curve and temperature rangecovering DSC signal from TC-onset to TC-endset. The percentage of solidfat content (SFC%) was determined as a ratio of DHpartial/DHtotal;where DHpartial is the apparent enthalpy due to crystallisation ofthe emulsified fat droplets and DHtotal is the total enthalpy changeof the bulk fat. The total enthalpies of the bulk stearin, AMF, andolein were 86.6, 79.8, and 73.4 J g�1, respectively, as determinedby cooling bulk phases from 70 to �70 �C at 5 �C min�1.

2.3.3. Cryogenic transmission electronic microscopyIn this work, the olein-based nanoemulsion was chosen to

examine the crystallisation of emulsified fats in vitrified frozen-hy-drated state using cryo-TEM (cryo-Transmission Electron Micros-copy). Freshly made nanoemulsion, as described in Section 2.2,was cooled from 20 to 4 �C at 10 �C min�1 using the Peltier platetemperature system of a discovery hybrid rheometer (HR-1 model,TA Instruments, New Castle, USA). The chilled emulsions were sub-sequently stored at 4 �C for 12–24 h before cryopreparation. Allsubsequent sample handling and operations such as pipetting,transport of dispersion, cryopreparation (cryo-box and specimengrid was chilled at 4 �C before introducing samples) were under-taken at 4 �C to maintain the solid state of crystallised emulsifiedfat. A small aliquot (3 lL) of dispersion was positioned on a dis-charged specimen copper grid with perforated carbon film. Auto-matic blotting was undertaken in 3–4 s to remove an excessliquid and generate a thin layer on the support specimen grid.The specimen grids were rapidly immersed into liquid ethane(temperature approximately �175 �C) in a cryo-box (Vitrobot,FEI, Oregon, US). The vitrified samples were transferred into aTEM cryo-holder under liquid nitrogen. Observations were madewith either Tecnai T12 120 keV TEM, with a 4 k FEI Eagle CCD Cam-era or Tecnai F30 300 keV Field Emission Gun (FEG) TEM with a 4 kDirect Electron LC1100 (lens-coupled) Camera, electron micro-scopes (FEI, Oregon, US) at magnification of 30,000 and 39,000times, respectively, at temperature �175 to �179 �C.

2.3.4. Statistical analysisAll measurements were carried out at least in duplicate and the

whole study was done in two replicates. Statistical analyses of thedata were conducted using the statistical package MINITAB�

Released 16 (Minitab Co., USA). Analysis of Variance of GLM(General Linear Model) and one-way analysis of variance (ANOVA)were used. Tukey’s multiple comparison test was employed todetermine significant differences of treatment means at P < 0.05.

3. Results and discussion

3.1. Crystallisation properties of AMF, stearin and olein fractions in thebulk state

The crystallisation properties of the milk fat, stearin and oleinfractions obtained from dry fractionation of melted milk fat andsubsequent vacuum filtration were investigated. Since crystallisa-tion of milk fat, stearin and olein fractions is a function of numerousvariables including temperature, cooling rate, agitation rate, chem-ical composition of initial milk fat (Kaylegian & Lindsay, 1995),characterisation of physicochemical properties of AMF, stearinand olein fractions in bulk state is essential before investigatingthem in the emulsified state. As can be seen in Fig. 1, the differencein fatty acid composition (see Section 2.1.1) between the stearinand olein fractions after fractionation step altered their DSC profilesas compared to that of the initial milk fat. Upon cooling, two main

Fig. 1. Cooling DSC curves of bulk anhydrous milk fat, stearin, and olein (21 subscripts denote the fractionation temperature) at three cooling rates 1, 5, and 10 �C min�1.

728 T. Truong et al. / Food Chemistry 145 (2014) 725–735

exothermic peaks were found in both the AMF and stearin fraction.The AMF started to crystallise at 20.9 �C (TC-onset) and the first crys-tallisation peak was within 20.9–14.4 �C at 1 �C min�1 cooling rate.The second peak with a small shoulder peak extended from 14.4 �Cto a final temperature of the cooling program investigated (�4 �C).In response to the same cooling rate (1 �C min�1), the higher con-centration of saturated fatty acids in the stearin fraction resultedin higher crystallisation temperature (TC-onset: 28.8 �C) and a sharpexothermic event joined by two overlapping peaks (TC-peak: 27.5and 23.2 �C) within a temperature range of 28.8–13.5 �C. Due tothe removal of high melting point TAGs from the milk fat duringfractionation, the crystallisation of the olein fraction was character-ised with a single exothermic peak set off at 10.6 �C. A small shoul-der peak initiated at 9.2 �C was also observed in a DSC cooling curveof the olein fraction at the rate of cooling 1 �C min�1. It is apparentfrom Fig. 1 that the crystallisation of the AMF, stearin and olein inbulk state were delayed with faster cooling rates (5 and10 �C min�1). The small shoulder peaks seen in the cooling profilesof the olein fraction and the AMF at slower cooling rate (1 �C min�1)disappeared at faster cooling rates, i.e. 5 or 10 �C min�1.

The crystallisation properties of the AMF, stearin and olein frac-tions observed in this study are in agreement with previous re-search in this field (Kaylegian & Lindsay, 1995; Lopez et al.,2006) that differences in chemical composition and cooling ratemodify the crystallisation of TAG molecules. According to theseauthors, during the X-ray diffraction measurement undertaken be-tween 60 and �7 �C at cooling rate 1 �C min�1, milk fat crystallisedin two double chain length and one triple chain length lamellarstructure in a form. Formation of crystals in the olein fraction cor-responded to a form with a triple chain length structure. Therewere two types of crystals, a and b0, presented in stearin fraction.The difference in crystallisation curves recorded at different cool-ing rates has been known to relate to formation of various poly-morphic forms. There were several lateral packing of TAGs incrystalline state detected in milk fat (a, sub-a, two b0, and one b),in which their presence as compound crystals depending on cool-ing rate employed (Lopez, Bourgaux, Lesieur, Bernadou et al.(2002), Lopez, Bourgaux, Lesieur, & Ollivon (2002); Lopez, Bourg-aux, Lesieur, & Ollivon, 2007).

3.2. Crystallisation behaviour of AMF, stearin and olein fractions in theemulsified state

3.2.1. Preparation of dairy-based emulsionsThe modal particle size diameters of all freshly made

dairy-based emulsions at different homogenising conditions(2.0–35.0 MPa and 3–5 cycles) were in the range of 0.13–

3.10 lm. Particle size measurement and visual observation showedthat all emulsions were stable after 24 h storage at room tempera-ture (20 �C). All the emulsions maintained the same particle sizedistribution as that of freshly-made ones and there was no visibleaggregation or creaming of droplets. The modal particle size ofdairy-based emulsions produced using pressures 15.0–35.0 MPawas about 0.20–0.13 lm. The particle size distribution of emul-sions fabricated at 15.0–35.0 MPa exhibited a Gaussian-type parti-cle size distribution while those prepared at lower homogenisingpressures (2.0–10.0 MPa) had an asymmetrical size distributionshifted towards larger particles. As an illustration, Fig. 2 representsthe particle size distribution of the three different stearin-basedemulsions, which were emulsified with the WPC, NaCN, or Tw80.Emulsions with a droplet size range below 0.2 lm can be regardedas nanoemulsions (Tadros, Izquierdo, Esquena, & Solans, 2004).Hence, in this study the dairy-based emulsions were preparedacross the relatively wide spectrum of droplet size from micron-to submicron- or nano-scale. In general, the mean particle size ofdairy-based emulsions stabilised by a small molecular surfactant,Tw80, was lower than those of dairy-based emulsifiers (WPC,NaCN) at the same lipid concentration and manufacturing condi-tion. This is probably because the small molecular surfactant iscapable of higher diffusion rates and rapid adsorption at the newlycreated oil/water interface, creating a rapid reduction in interfacialtension. Dairy proteins, being macromolecular emulsifiers, are lesseffective at reducing droplet size compared to small molecular sur-factant due for this reason (McClements, 2011).

3.2.2. Validation of DSC conditionsPreliminary experiments were undertaken to evaluate water

crystallisation in the emulsions, which can interfere with the crys-tallisation of the fat component. Crystallisation of water wasobserved in the emulsions after supercooling below �8 �C (AMFand stearin-based emulsions), and �10 �C (olein-based emulsions).It was found that the crystallisation peak of the olein-based emul-sions was broadened across the sub-zero temperature range. Toclarify that the crystallisation temperature of the olein-basedemulsions is due to crystallisation of olein TAGs rather than watercrystallisation, 10% of sodium chloride (NaCl) was added into WPC-stabilised olein emulsion to depress the freezing point of the water.Based on the colligative properties of solutions, the amount offreezing point depression DTf calculated for 10% NaCl added towater was �6.3 �C as below:

DTf ¼ �i � Kf �mDTf ¼ �½2� ð1:86 �C Kg mol�1 � ½10 g=58:44 g mol�1

=

100� 10�3 kgÞ� ¼ �6:3 �C

ð1Þ

Fig. 2. Droplet size distribution of the stearin emulsions fabricated at varioushomogenising pressures (vertical reference lines from right to left were of 1.0, 0.5,and 0.2 lm).

T. Truong et al. / Food Chemistry 145 (2014) 725–735 729

where, i is the number of ions, Kf is the molal freezing point depres-sion constant (1.86 �C kg mol�1 for pure water), m is the molal con-centration of the solution. DSC measurement showed that thecrystallisation of the emulsion with 10% NaCl happened at a lowertemperature (�6.12 �C) than in the original emulsion (withoutNaCl). This difference in crystallisation temperature was very closeto the calculated value of freezing point depression. The crystallisa-tion peaks recorded over the range 0 to �4 �C for both WPC–oleinemulsions with and without 10% NaCl crystallisation were notchanged during cooling (data not shown). It should also be notedthat addition of salts at this concentration (10%) did not affect theemulsion droplet size or emulsifying property. From these resultsit is clear that the thermal events observed across the zero temper-ature range were caused by crystallisation of the oil phase (olein)independently and not by water crystallisation.

In another calibration study, the aqueous solution (without oiladdition) as described in Section 2.2 was homogenised and thensubjected to DSC measurement under the same conditions todetermine the cooling profile of the oil-free dispersions. Therewas no measurable thermal event detected in the aqueous solutionwithin the scanning temperature range, indicating that any DSCsignals subsequently obtained from emulsion analysis can be as-signed solely to phase transition effects of the lipid component inthe formulation.

3.2.3. Effect of emulsion droplet sizeThe onset crystallisation temperature, TC-onset, reported as an

average of two replicates, was plotted against the mean modal par-ticle size of the nine kinds of emulsions at three cooling rates (1, 5,and 10 �C min�1) in Fig. 3. Under the same cooling conditions as forthe bulk milk fat, stearin, and olein the crystallisation of dairy-based emulsions initiated at much lower temperatures (Figs. 1and 3). Over the whole series of size ranges and cooling rates,the crystallisation temperatures of emulsified stearin, AMF, andolein were depressed by about 15–20, 9–13, and 7–14 �C, respec-tively, compared to their bulk counterparts (see Section 3.1). Thistype of shift to lower crystallisation temperatures by emulsifiedfats, compared to their non-emulsified state, is well known andhas been described for many other emulsion systems (Lopez,Bourgaux, Lesieur, Bernadou et al. (2002), Lopez, Bourgaux, Lesieur,& Ollivon (2002); McClements et al., 1993). In bulk systems, cata-lytic impurities (e.g. foreign particles on the surface, minor lipids inmilk components, etc.) are needed to initiate heterogeneous nucle-ation. Whenever a crystal is formed in any location of the bulk oil,secondary nucleation would occur, resulting in crystal formationthroughout the whole volume of bulk oil. As fats are divided intodroplets in an emulsion system, the number of catalytic impuritiesrequired to initiate crystallisation in this way is much greater, asthey would need to be distributed among a very large number ofisolated droplets (�2–3 � 1013 droplets per litre for sub-micronparticle size range in this study). In such emulsions, the chanceof an impurity catalysing any single droplet is greatly reduced,the crystallisation process is then limited and this results in anincreasing propensity for supercooling (McClements et al., 1993).

A high supercooling effect before fat crystallisation was ob-served within the dairy-based emulsions in this study. Themeasured supercooling DT, defined as the difference between theinitial heat flow release upon cooling (TC-onset) and the completionof fat melting upon reheating (TM-endset), was 27–33, 23–33, and11–18 �C for stearin, AMF, and olein-based emulsions, respectively,at all the three cooling rates. Table 1 showed the supercoolingeffect of the WPC-stabilised emulsions with the AMF, stearin andolein fractions as the oil phase. Similar supercooling effect was ob-served in the NaCN- and Tw80-stabilisied emulsions. This superco-oling tendency tended to slightly increase with decreasing dropletsize, indicating that the time to obtain the first nucleus was inver-sely proportional to droplet volume. Extensive supercooling thus isrequired to crystallise smaller droplet size. In other words, a largeamount of droplets in a fine emulsion system requires a corre-spondingly high number of nucleation sites and confinement ofsuch small droplets at submicron- and nanometric-size depressesthe crystallisation temperature and limits the crystal growth dueto physical confinement by the droplet border (Higami et al.,2003; McClements et al., 1993; Sonoda, Takata, Ueno, & Sato,2006).

A significant difference in terms of the onset crystallisationtemperature among particle size range for each emulsion systemwas found (P < 0.05). For instance, onset crystallisation tempera-ture and crystallisation enthalpies of the WPC-stabilised emulsionstended to decrease with decreasing droplet size (Table 1 andFig. 3). The same trends were observed in the NaCN- and Tw80-sta-bilised emulsions (Fig. 3). In general, the magnitude of decrease incrystallisation temperature with decreasing droplet size was morepronounced in WPC-stabilised emulsions (by 3.9 �C) than NaCN- orTw80-stabilised emulsions (by about 1.0 �C) between the largestand smallest average droplet sizes investigated (Fig. 3). This differ-ence in crystallisation behaviour may be attributable to an effect ofemulsifier type that will be mentioned later in Section 3.2.4. A sim-ilar trend of decreasing crystallisation temperature as a function ofdroplet size has also been observed in some non-edible modelemulsion systems (Bunjes et al., 2000; Dickinson et al., 1991;

Fig. 3. The onset crystallisation temperatures (TC-onset, �C) against mean emulsion droplet sizes of the dairy-based emulsions at cooling rates of 1, 5, and 10 �C min�1. Sub-plotof (F) illustrates mean droplet size versus TC-onset of NaCN-olein emulsions in the size range of 1.5–3.1 lm. Capped vertical bars are STD of the measured means. Regressivefitting curves at the three cooling rates are shown in solid line (1 �C min�1), medium dash line ( �C min�1), and dotted line ( �C min�1) with the regressive coefficient R2 andsignificance at the 1% level by F-test (⁄⁄). The onset crystallisation temperatures of corresponding bulk fats were of 28.8, 26.9, and 26.1 �C (stearin); 20.9, 18.2, and 16.4 �C(AMF); 10.6, 9.9, and 9.5 �C (olein) at the three cooling rates, respectively.

730 T. Truong et al. / Food Chemistry 145 (2014) 725–735

McClements et al., 1993). In reference to a similar dairy-basedemulsion system at the same cooling rate (1 �C min�1), in our studyWPC-stabilised AMF emulsions having droplet size d32 0.59 lmand d32 0.38 lm crystallised at 11.97 and 10.2 �C, respectively,which is comparable to the observations of Lopez, Bourgaux,Lesieur, Bernadou et al. (2002), Lopez, Bourgaux, Lesieur, & Ollivon(2002) (11.6 �C/0.67 lm; 10.9 �C/0.38 lm).

Further reduction of the droplet size range to 0.13 lm in thisstudy, showed a continuing decreasing trend of crystallisationtemperature with reduced droplet size in all nine kinds of thedairy-based emulsions examined. This indicates that there was aparticle size effect on the crystallisation temperature of dairy-based emulsions within the size range 0.13–3.10 lm. It may beof industrial interest if this discrepancy (1.0–3.9 �C) in crystallisa-tion temperature between conventional and nanoemulsions couldbe used to generate significant differences in physical and func-tional properties of fat-containing food products or fat-basedencapsulation and delivery systems, made from different types ofemulsion.

Fig. 3 shows the best fitting curves that describe the associationbetween the mean emulsion droplet size (modal diameter) and themeasured onset points of crystallisation temperature of each dairy-based emulsion. The model was of a simple exponential growthy = yo + a(1 � bx), in which y is a predicted onset crystallisationtemperature (�C) at given cooling rate, yo is a constant, a is theslope, b is the base of the exponential function, and x is range ofmean particle size investigated, 0.13 < x < 3.10 lm. The leadingcoefficient a reflects the slope of the exponential curve while the

base b exhibits the shape or rate of growth or decay. In the contextof this study, at a constant b, higher a value indicates that the smal-ler the particle size, the lower the onset crystallisation tempera-ture. At a constant a, 0 < b < 1, the smaller b value denotes onsetcrystallisation temperature is getting similar when droplet size isgetting bigger than 0.5 lm. In case b P 1, crystallisation tempera-ture increases rapidly with greater particle size.

It was found that the a values were highest in WPC-based emul-sions (a = 3.67–8.92) whereas similar values of a were observed inNaCN- and Tw80-based emulsions (0.7–3.13). This means thatWPC had more pronounced effect on lowering the onset crystalli-sation temperature with smaller droplet size compared to NaCNand Tw80. NaCN-based emulsions had the smallest range of b val-ues (0.32–0.91) compared to that of WPC-based (0.52–1.41) andTw80-based (1.14–3.20) emulsions. Those b ranges showed thatamong the three emulsifiers used, NaCN had a slight impact onthe magnitude of increase in crystallisation temperature in greaterdroplet sizes while rapid increasing the onset crystallisation tem-perature (b > 1) was obvious in Tw80.

Regarding oil phase, there was no remarkable difference inranges of a and b values, e.g. stearin (a: 0.70–5.31; b: 0.32–2.75),AMF (a: 1.33–8.92; b: 0.36–2.84), and olein (a: 1.32–5.78; b:0.61–1.41). This demonstrates a similar trend in relationship be-tween onset crystallisation temperature and droplet sizes in oil-based emulsions. However, differences in parameter yo were founddecreasing in the order of stearin (7.80–11.73) > AMF (2.62–8.91) > olein (1.71–5.10). This might be associated with chemicalcomposition of oil phase as stearin crystallises in higher

Table 1Mean droplet mode diameter (d), supercooling degree (DT), and crystallisation enthalpies (DHC) of the WPC-stabilised stearin-, AMF-, and olein emulsions.

d lm Cooling rate 1 �C min�1 Cooling rate 5 �C min�1 Cooling rate 10 �C min�1

TC-onset DT DHC TC-onset DT DHC TC-onset DT DHC

�C �C J g�1 �C �C J g�1 �C �C J g�1

WPC-stabilised stearin emulsions1.49 14.0 28.0 39.3 11.9 27.7 44.2 11.1 30.7 48.91.31 14.3 27.7 36.7 12.1 29.7 48.7 11.1 30.7 43.50.90 13.4 28.6 37.3 11.4 30.5 44.7 10.9 30.9 44.40.53 12.3 29.8 36.1 10.8 31.1 40.6 10.2 31.6 42.10.22 11.9 29.9 37.6 10.7 31.1 36.9 10.0 31.6 41.50.21 11.5 30.0 30.6 10.6 30.9 34.3 9.8 31.7 35.20.17 11.1 30.0 32.6 10.1 31.0 35.7 9.7 31.4 36.2

WPC-stabilised AMF emulsions1.66 12.2 23.3 13.2 9.5 25.9 20.6 8.2 27.3 28.11.27 12.0 23.8 12.1 9.5 26.4 18.4 8.2 27.1 27.50.79 10.2 25.5 11.6 8.4 27.4 19.3 7.4 27.7 26.60.54 9.1 25.7 10.2 7.4 27.3 17.4 6.6 28.3 19.90.52 9.0 25.8 9.9 7.3 27.5 16.7 6.6 28.3 20.40.19 8.0 24.8 6.7 6.7 26.9 11.9 6.0 27.6 13.40.18 7.8 25.3 6.3 6.6 26.6 10.6 5.9 27.8 14.0

WPC-stabilised olein emulsions1.60 2.8 17.7 22.7 7.9 11.5 30.2 8.6 11.3 31.70.94 2.4 18.0 29.6 7.7 11.7 31.4 8.5 11.3 30.80.67 2.3 18.1 27.5 6.3 13.1 30.3 6.2 11.2 29.20.54 1.7 16.8 23.1 5.5 11.7 26.4 5.0 12.2 28.80.21 1.5 17.0 20.9 5.3 11.9 27.5 4.8 12.3 27.60.19 1.3 17.5 19.0 4.9 12.2 25.9 4.5 12.7 23.30.17 0.9 16.5 16.9 4.7 12.4 26.3 4.3 12.8 23.7

All values are means of two measurements. Percentage of coefficient of variation (CV %) of all values are less than 1%.

T. Truong et al. / Food Chemistry 145 (2014) 725–735 731

temperature due to enrichment in saturated fatty acids (refer toSection 3.1).

In general, the relationship between the droplet size and the on-set crystallisation temperature of dairy-based emulsion exhibitedan exponential function, which implies that the smaller the dropletsize in ranges below 0.5 lm, the greater the magnitude in decreaseof onset crystallisation temperature.

For greater droplet size, the crystallisation temperatureincreased gradually and reached unchanged. It is suggested thateither the droplets are large enough to collide during the coolingprocess and/or the larger droplets are unstable during coolingand form aggregates, so that any crystal formation is readilyextended to other droplets via heterogeneous nucleation. In addi-tion, emulsifier also plays a role in governing the crystallisationbehaviour of dairy-based emulsions, in which its impact will beaddressed in Section 3.2.4.

The influence of particle size on the crystallisation behaviour ofdairy-based emulsions was confirmed visually by cryogenic trans-mission electronic microscopy (Fig. 4). Freshly made NaCN-stabi-lised olein-rich nanoemulsion (modal diameter of 0.16 lm) wasallowed to cool at 10 �C min�1 from room temperature (20 �C) torefrigerator temperature (4 �C) using a cold stage and kept at4 �C for 12 h before vitrification at the same temperature (4 �C).The onset crystallisation temperature of this nanoemulsion(d = 0.16 lm) determined by DSC at the same cooling rate(10 �C min�1) was 3.97 ± 0.11 �C. Interestingly, as can be seen inFig. 4A, some larger droplets were already in a solid state at 4 �C,exhibiting lamellar layer structures of fat crystals (closer view inFig. 4B) whereas surrounding smaller droplets were of liquidphase. An average arithmetic diameter (DFeret) of crystallised drop-lets measured in these micrographs was 0.50 ± 0.07 lm (n = 4),whereas droplets that had not yet crystallised had an averagearithmetic diameter of 0.11 ± 0.03 lm (n = 108). Food emulsionsare normally polydisperse, containing various droplet size classes.It is obvious from Fig. 4 that at the same cooling rate and temper-ature, big droplets tended to crystallise first within the same

specimen, indicating the particle size effect on the crystallisationbehaviour of the emulsion. This provides another indication ofthe effect of particle confinement on the crystallisation of ediblefat nanoemulsions.

3.2.4. Effect of emulsifier typesEmulsifiers are known to play a role in crystallisation behaviour

of fats in dispersed systems. Hydrophobic emulsifier tail groups actas a template for initiating the nucleation in the emulsified drop-lets. For instance, hydrophobic parts of Tween 40 and Tween 60,having the same chain length to n-hexadecane molecules, initiatecrystallisation of n-hexadecane o/w emulsions (McClementset al., 1993). Conversely, food proteins lack the long hydrophobicchains of surfactant and are considered as having little effect onfat crystallisation in edible emulsions (Dickinson et al., 1991;Palanuwech & Coupland, 2003). Food proteins are amphiphilicmolecules, in which non polar parts are in contact with oil phasewhereas polar parts are toward aqueous phase. As proteins areadsorbed onto oil/water interface, a stabilising interfacial layer isformed because of proteins unfolding and rearrangement (Dalgle-ish, 1996). The protein hydrophobic part protruding into the oildroplet, however, does not cause surface heterogeneous nucleationdue to incompatibility of chemical structure between the food pro-tein and triglyceride mix. In fact, the occurrence of surface or inter-facial heterogeneous nucleation is only achievable if theemulsifiers used are of similar molecular structure to that of thedispersed lipids (such as when n-hexadecane, C16H34, is used withTween40 having hydrophobic chains of –C16H33) (Dickinson et al.,1991; McClements et al., 1993; Palanuwech & Coupland, 2003).

It appears that crystallisation temperature also depended on thetypes of emulsifier used to fabricate the dairy-based emulsions.Within each oil-based emulsions (Fig. 3A, D, G: stearin; Fig. 3B, E,H: AMF, and Fig. 3C, F, I: olein), Tw80-stabilised emulsions tendedto crystallise at lower temperature than those of stabilised by dairyproteins (WPC and NaCN) (P < 0.05). This result agreed with previ-ous study performed on n-alkanes emulsions (0.15–3.45 lm) that

Fig. 4. (A) Micrographs of NaCN–olein nanoemulsions (modal diameter of 0.16 lm) at 4 �C; (B) close-up view of lamellar layer structure of crystallised fat droplet as shown in(A), and (C) smaller droplets were in liquid state similar to those observed at 20 �C.

732 T. Truong et al. / Food Chemistry 145 (2014) 725–735

Tw20-stabilised emulsions crystallised about 1 �C lower than case-inate-stabilised ones (Gulseren & Coupland, 2007). Depression ofTC-onset in Tw80-stabilised emulsions in reference to dairy pro-teins-stabilised emulsions may be in association with role ofhydrophobic parts of Tw80, which acts as a template initiatingnucleation process (Gulseren & Coupland, 2007; McClementset al., 1993).

It was also shown that onset crystallisation temperatures ofWPC- and NaCN-stabilised emulsions for each fat phase were notstatistically different (P > 0.05). Though similar ranges of onsetcrystallisation temperatures were found between WPC- andNaCN-stabilised emulsions, the magnitude of decrease in Tc-onset

with smaller droplet size was greater in WPC-stabilised ones asshown in Fig. 3. As mentioned in Section 3.2.3, below the sub-mi-cron range of droplet size, crystallisation is more retarded in WPC-stabilised emulsions. This could be attributed to the different per-formance of WPC and NaCN at the oil/water interface. In emulsionsystems, a flexible structure of sodium caseinate unfolds rapidly atthe o/w interface and coats the oil droplet with a layer of �10 nmthickness while whey proteins are partially unfolded at the inter-face and form a relatively thin layer (�2 nm) surrounding the oildroplet (Dalgleish, 1996). Sodium caseinate is resistant to heatand pressure treatment, while whey proteins are known to be sen-sitive to heating and high pressure processing (Huppertz, Fox, deKruif, & Kelly, 2006). Denaturation of whey proteins caused bythe combined heat and pressure effects of homogenisation mayhave modified the conformation of whey protein at the oil/waterinterface. Although the WPC solution was not warmed above itsdenaturation temperature, i.e. 76.9 �C (Bernal & Jelen, 1985), aninstantaneous increase in temperature may have occurred duringthe droplet break-up process. A temperature rise during high pres-sure emulsification and its effect on whey protein denaturation has

been reported previously (Desrumaux & Marcand, 2002). Upondenaturation, hydrophobic groups buried inside are exposed lead-ing to an increase in surface hydrophobicity and protein aggrega-tion. Lee, Lefevre, Subirade, and Paquin (2009) found thatinteractions between adsorbed proteins are promoted under highpressure forming a more rigid interfacial layer. Therefore, we pro-pose that either more sites for hydrophobic interaction or strongprotein–protein interaction providing a rigid shield at the interfaceretarded the nucleation in the WPC-stabilised emulsions. As theenergy input level (pressure and operating duration) was increasedto form smaller droplet sizes, the magnitude of this effect may beincreased with decreasing droplet size, i.e. 0.5 lm and below. Incontrast, no influence was found on the surface activity of bovineb-casein up to 800 MPa (Dickinson, Murray, & Pawlowsky, 1997).This observation further supports our assumption that the confor-mation change of whey proteins during high pressure processingmay play a role in the subsequent crystallisation behaviour ofdairy-based nanoemulsions.

3.2.5. Effect of cooling rateFig. 3 illustrates influence of cooling rate on crystallisation tem-

perature (TC-onset) of the dairy-based emulsions with various drop-let sizes. A similar pattern was seen for all the dairy-basedemulsions subjected to different cooling rates (1, 5, and10 �C min�1), with slower cooling rates resulting in faster crystalformation. The cooling rate had a significant impact on onset crys-tallisation temperature of all emulsions investigated (P < 0.05). Atthe same particle size, highest onset crystallisation temperaturewas obtained at 1 �C min�1, followed by at 5 and 10 �C min�1. Thiscooling rate effect has been reported in previous studies on bothbulk systems and conventional emulsions (Lopez, Bourgaux,Lesieur, Bernadou, et al. (2002), Lopez, Bourgaux, Lesieur, & Ollivon

Fig. 5. A representative graph of increase in solid fat fraction during crystallisation of olein-based emulsions (modal diameter 0.13–3.10 lm) as a function of droplet size,emulsifier type, and cooling rate; (A, B, C): WPC–olein emulsions; (D, E, F): NaCN–olein emulsions; and (G, H, I): Tw80–olein emulsions.

T. Truong et al. / Food Chemistry 145 (2014) 725–735 733

(2002); Martini, Herrera, & Hartel, 2001). When fat droplets expe-rience slow cooling, the molecules are able to arrange themselvesin the most favourable thermodynamic conformation, from whichcrystals form very readily whenever reaching the crystallisationtemperature. Conversely, under fast cooling rate conditions, mole-cules do not have sufficient time to orientate themselves in theideal conformation and additional time is required, leading toretarded crystallisation (Martini et al., 2001).

3.2.6. Solid fat contentThe increase in solid fat fractions on cooling of the olein-based

emulsions, as measured by DSC, is shown graphically in Fig. 5. Thesame tendencies were noted for stearin- and AMF-basedemulsions. It was seen that fat crystallisation occurred in thoseemulsions not only to a lesser extent than in their bulk fats(100% solid) but also in a different behaviour depending on themean droplet size and type of emulsifiers. While there was a strongtendency towards decreasing proportion of fat crystallinity withsmaller droplet size in the emulsions stabilised by dairy proteins,such particle size effects were less pronounced in the Tw80-stabi-lised emulsions. It is also shown in Fig. 5 that evolution of solidswas higher in NaCN-stabilised olein-rich emulsions than those of

WPC-stabilised counterparts. The results indicated that the rateof increase in solid fat fraction during cooling of these dairy-basedemulsions is related to the molecular adsorption and droplet sizeeffects that were discussed earlier in Sections 3.2.4 and 3.2.3,respectively. The impact of emulsifier type on fat crystallinity ofemulsions was in agreement with previous studies; thus it hasbeen reported previously that solid fat content is a function of ad-sorbed proteins (type, conformation, surface coverage, stabilityagainst droplet aggregation (Relkin, 2004). Relkin, Sourdet, andFosseux (2003) also found that fat droplets in whipped emulsionsstabilised by whey proteins had lesser solid fat content comparedto mixture of whey proteins and caseins. The percentage of solidsin n-hexadecane oil-in-water emulsions stabilised by casein ap-peared less than Tw80-stabilised emulsions (n-hexadecane mole-cules has similar structure to hydrophobic part of Tw80)(McClements et al., 1993). However, the relationship of droplet sizeto rate of increase in solid fat content during cooling seen in thisstudy contrasts to the findings of Relkin (2004), who reported thatdifference in the increase of fat solids was not due to differences infat droplet size distribution (0.81 ± 0.02 to 0.87 ± 0.07 lm). A pos-sible explanation for this might be that wider range of droplet sizeswas examined in our study.

734 T. Truong et al. / Food Chemistry 145 (2014) 725–735

4. Conclusion

This study provides an insight into the different crystallisa-tion behaviours of milk fat, stearin and olein fractions in dairy-based emulsions from micron- to nanometric-size range (0.13–3.10 lm). Crystallisation behaviour of the emulsions differeddepending on the components of the oil phase (milk fat, stearin,or olein) and the emulsifiers used (dairy proteins or Tween80),and observed differences were attributed to droplet size effect,emulsifier functionality, and cooling rates. It was demonstratedthat emulsified milk fats needed extensive supercooling to initi-ate crystallisation compared to the bulk fats of the same lipidcomposition. Emulsion droplet size appears to represent a phys-ical barrier, constraining and retarding the crystallisation pro-cess. Such confinement depresses crystallisation temperatureand growth of fat crystals as indicated by a decrease in crystal-lisation temperature and percent of solid fat with a reduction inemulsion droplet size. The trend of increasing crystallisationtemperature with increasing droplet size followed a power-lawmodel, implying that the particle size effect is less pronouncedonce droplet size becomes larger than 0.5 lm. The effect of par-ticle size on fat crystallisation in dairy-based nanoemulsions isclearly supported by micrographs taken with cryogenic transmis-sion electronic microscopy. To our knowledge, this is the firstpublished attempt to visualise not only the crystallisation ofcrystals derived from dairy fat nanoemulsions but also influenceof droplet size on crystallisation. It was seen that at the onset ofcrystallisation, only big droplets (�0.50 lm) tended to crystallisewhile nanometric-size range droplets (0.11 lm) were still in li-quid state.

Cooling rate had a strong influence on crystallisation tempera-ture of these dairy-based emulsions. At the same average particlesize, increasing cooling rate (1, 5, and 10 �C min�1) delayed crystal-lisation. The presence of emulsifiers also governed the crystallisa-tion behaviour of dairy-based micron- and nano-emulsions suchas crystallisation temperature and proportion of solid fat contentsin various ways, presumably imparted by effective thickness of theemulsifier layer surrounding the oil droplets, hydrophobic interac-tions, protein denaturation, and similarity in molecular structurebetween oil and emulsifier.

Taken together, the results of this study point to a range of dif-ferent effects of emulsion droplet size on the crystallisation behav-iour of dairy-based nanoemulsions, which appears associated withthe nature of components in emulsion systems, as well as with thephysical limits and properties of particle size itself. From the foodapplication point of view, the outcome of this study underlines thesignificance of particle size, choice of emulsifiers and process (cool-ing rate and temperature) on crystallisation of food-ingredientnanoemulsions. Some of these physico-chemical parameters aredifferent from those applying to conventional emulsions. Thesedistinct properties of dairy nanoemulsions may provide the basisfor a new means of modifying the physical, functional or rheolog-ical properties of fat-containing food products or encapsulationand delivery systems. Further work needs to be done to establishrole of nano-droplet template crystals in determining propertiesof dairy or food products. Investigation of crystal polymorphs usingX-ray diffraction technique was not presented here but will be cov-ered in continuing research.

Acknowledgements

The authors greatly appreciate the financial support of DIAL(Dairy Innovation Australia Limited). The authors acknowledgethe facilities, and the scientific and technical assistance, of the Aus-tralian Microscopy & Microanalysis Research Facility at the Centrefor Microscopy and Microanalysis, The University of Queensland.

The authors are grateful to Dr. Jennifer Waanders for providinganalytical service of fatty acid profiles.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2013.08.072.

References

Bazmi, A., & Relkin, P. (2009). Effects of processing conditions on structural andfunctional parameters of whipped dairy emulsions containing various fatty acidcompositions. Journal of Dairy Science, 92(8), 3566–3574.

Bernal, V., & Jelen, P. (1985). Thermal-stability of whey proteins – A calorimetricstudy. Journal of Dairy Science, 68(11), 2847–2852.

Bugeat, S., Briard-Bion, V., Perez, J., Pradel, P., Martin, B., Lesieur, S., et al. (2011).Enrichment in unsaturated fatty acids and emulsion droplet size affect thecrystallization behaviour of milk triacylglycerols upon storage at 4� C. FoodResearch International, 44(5), 1314–1330.

Bunjes, H., Koch, M. H. J., & Westesen, K. (2000). Effect of particle size on colloidalsolid triglycerides. Langmuir, 16(12), 5234–5241.

Christie, W. W. (1995). Composition and structure of milk lipids. In P. F. Fox (Ed.).Advanced Dairy Chemistry (Vol. 2, pp. 1–36). London: Chapman & Hall.

Dalgleish, D. G. (1996). Conformations and structures of milk proteins adsorbed tooil-water interfaces. Food Research International, 29(5–6), 541–547.

Desrumaux, A., & Marcand, J. (2002). Formation of sunflower oil emulsionsstabilized by whey proteins with high-pressure homogenization (up to350 MPa): Effect of pressure on emulsion characteristics. International Journalof Food Science and Technology, 37(3), 263–269.

Dickinson, E., Mcclements, D. J., & Povey, M. J. W. (1991). Ultrasonic investigation ofthe particle-size dependence of crystallization in n-hexadecane-in-wateremulsions. Journal of Colloid and Interface Science, 142(1), 103–110.

Dickinson, E., Murray, B. S., & Pawlowsky, K. (1997). On the effect of high-pressuretreatment on the surface activity of beta-casein. Food Hydrocolloids, 11(4),507–509.

Gulseren, I., & Coupland, J. N. (2007). The effect of emulsifier type and droplet sizeon phase transitions in emulsified even-numbered n-alkanes. Journal of theAmerican Oil Chemists Society, 84(7), 621–629.

Higami, M., Ueno, S., Segawa, T., Iwanami, K., & Sato, K. (2003). Simultaneoussynchrotron radiation X-ray diffraction – DSC analysis of melting andcrystallization behavior of trilauroylglycerol in nanoparticles of oil-in-wateremulsion. Journal of the American Oil Chemists Society, 80(8), 731–739.

Huppertz, T., Fox, P. F., de Kruif, K. G., & Kelly, A. L. (2006). High pressure-inducedchanges in bovine milk proteins: A review. Biochimica Et Biophysica Acta-Proteins and Proteomics, 1764(3), 593–598.

Kaylegian, K. E., & Lindsay, R. C. (1995). Handbook of milk fat fractionation technologyand applications. Champaign, Illinois: AOCS Press.

Lee, S. H., Lefevre, T., Subirade, M., & Paquin, P. (2009). Effects of ultra-high pressurehomogenization on the properties and structure of interfacial protein layer inwhey protein-stabilized emulsion. Food Chemistry, 113(1), 191–195.

Lopez, C., Bourgaux, C., Lesieur, P., Bernadou, S., Keller, G., & Ollivon, M. (2002a).Thermal and structural behavior of milk fat 3. influence of cooling rate anddroplet size on cream crystallization. Journal of Colloid and Interface Science, 254,64–78.

Lopez, C., Bourgaux, C., Lesieur, P., & Ollivon, M. (2002b). Crystalline structuresformed in cream and anhydrous milk fat at 4 �C. Lait, 82, 317–335.

Lopez, C., Bourgaux, C., Lesieur, P., & Ollivon, M. (2007). Coupling of time-resolvedsynchrotron X-ray diffraction and DSC to elucidate the crystallisation propertiesand polymorphism of triglycerides in milk fat globules. Lait, 87(4–5), 459–480.

Lopez, C., Bourgaux, C., Lesieur, P., Riaublanc, A., & Ollivon, M. (2006). Milk fat andprimary fractions obtained by dry fractionation – 1. Chemical composition andcrystallisation properties. Chemistry and Physics of Lipids, 144(1), 17–33.

Lopez, C., Riaublanc, A., Lesieur, P., Bourgaux, C., Keller, G., & Ollivon, M. (2001).Definition of a model fat for crystallization-in-emulsion studies. Journal of theAmerican Oil Chemists Society, 78(12), 1233–1244.

Martini, S., Herrera, M. L., & Hartel, R. W. (2001). Effect of cooling rate on nucleationbehavior of milk fat-sunflower oil blends. Journal of Agricultural and FoodChemistry, 49(7), 3223–3229.

Martini, S., & Tippetts, M. (2008). Effect of processing conditions on thecrystallization behavior and destabilization kinetics of oil-in-water emulsions.Journal of the American Oil Chemists Society, 85(2), 119–128.

McClements, D. J. (2011). Edible nanoemulsions: Fabrication, properties, andfunctional performance. Soft Matter, 7(6), 2297–2316.

McClements, D. J., Dungan, S. R., German, J. B., Simoneau, C., & Kinsella, J. E. (1993).Droplet size and emulsifier type affect crystallization and melting ofhydrocarbon-in-water emulsions. Journal of Food Science, 58(5), 1148.

Montenegro, R., Antonietti, M., Mastai, Y., & Landfester, K. (2003). Crystallization inminiemulsion droplets. Journal of Physical Chemistry B, 107(21), 5088–5094.

Palanuwech, J., & Coupland, J. N. (2003). Effect of surfactant type on the stability ofoil-in-water emulsions to dispersed phase crystallization. Colloids and Surfacesa-Physicochemical and Engineering Aspects, 223(1–3), 251–262.

T. Truong et al. / Food Chemistry 145 (2014) 725–735 735

Relkin, P. (2004). Using DSC for monitoring protein conformation stability andeffects of fat droplets crystallinity in complex food emulsions. In D. Lörinczy(Ed.), The Nature of Biological Systems as Revealed by Thermal Methods(pp. 99–126). Dordrecht, Netherlands: Kluwer Academic Publishers.

Relkin, P., Sourdet, S., & Fosseux, P. Y. (2003). Fat crystallization in complex foodemulsions – Effects of adsorbed milk proteins and of a whipping process. Journalof Thermal Analysis and Calorimetry, 71(1), 187–195.

Relkin, P., Yung, J. M., Kalnin, D., & Ollivon, M. (2008). Structural behaviour of lipiddroplets in protein-stabilized nano-emulsions and stability of alpha-tocopherol.Food Biophysics, 3(2), 163–168.

Sonoda, T., Takata, Y., Ueno, S., & Sato, K. (2006). Effects of emulsifiers oncrystallization behavior of lipid crystals in nanometer-size oil-in-wateremulsion droplets. Crystal Growth & Design, 6(1), 306–312.

Tadros, T., Izquierdo, R., Esquena, J., & Solans, C. (2004). Formation and stability ofnano-emulsions. Advances in Colloid and Interface Science, 108, 303–318.

Walstra, P., van Vliet, T., & Kloek, W. (1995). Crystallisation and rheologicalproperties of milk fat. In P. F. Fox (Ed.). Advanced in dairy chemistry (Vol. 2,pp. 179–212). London: Chapman & Hall.