[PDF] Polymorphism, microstructure and rheology - Nano-Science Center

39509_7ronholt_2012.pdf Polymorphism,microstructureand rheologyofbutter. Effectsofcream heattreatment

StineRønholt

a,! ,JacobJudas KainKirkensgaard b ,ThomasBaek Pedersen a ,KellMortensen b ,

JesChristianKnudsen

a a

DepartmentofFood Science,Facultyof Science,Universityof Copenhagen,Rolighedsvej 30,DK-1958Frederiksberg C,Denmark

b NielsBohrInstitute, UniversityofCopenhagen, DK-1871FrederiksbergC, Denmark articleinfo

Articlehistory:

Received5March 2012

Receivedinrevised form25 April2012

Accepted23May 2012

Availableonline30 May2012

Keywords:

Butter

Milkfat

X-raydiffraction

Fatcrystallization

Rheology

abstract Theeffectof creamheattreatment priortobutter manufacturing,fluctuatingtemperatures during storageandpresence offatglobules vs.nofat globuleswasexamined inlaboratory scaleproducedbutter. X-raydiffractionand differentialscanning calorimetrywasused tostudy crystallizationbehaviourand nuclearmagnetic resonancetomeasure solidfatcontent andwaterdroplet sizedistribution.Further- more,thecrystal structurewaslinked totherheological propertiesandmicrostructure ofthe butterusing confocallaserscanning microscopy.Butterproduced fromnon-maturedcream mainlyformed a-and b 0 -crystalswithminor tracesof b-crystals.Maturingof thecreamcaused atransitionfrom a-tob 0 - andb-form.The rheologicalbehaviourof slowcooledbutter deviatedfromthe maturedonesby having alowerelastic modulus,causedby aweakercrystal network.Presence offat globulesdidnot affectthe rheologicalpropertiessignificantly. !2012ElsevierLtd. Allrights reserved.

1.Introduction

Butterisproduced byamechanical phaseinversionof cream,an oil-in-wateremulsion,to reachawater-in-oil emulsion.Morepre- cisely,butterconsists ofa continuousfatphase inwhichwater droplets;fatglobules andanetwork offatcrystals aredispersed. Thefatcrystal networkisessential, sinceitdetermines thespread- ability,appearance andmouthfeelof thebutterand isstronglyre- latedtothe buttercomposition andoverallstructure .Theratio betweenthesolid andliquidfat isofoutmost importanceforthe rheologicalpropertiesof butterandspreads: withoutsolidfat, but- terisfully liquid.Withoutliquid fat,thebutter wouldappearhard andbrittle( Narine&Marangoni, 1999).Eventhough thesolid fat contentisthe same,fatcan havevery differentphysicalcharacter- istics(Haighton,1965;Shama &Sherman,1970 ).Sincea greater partofthe solidfat isinsidethe fatglobules,not allfatcrystals areableto formanetwork outsidetheglobule. Duetothe large volumefractionof fatglobulesin butter,theirpresence isthusbe- lievedtoinfluence thefirmnessof theproductalthough resultsare notconclusiveas towhatextent (Fedotova&Lencki, 2010;Mulder &W alstra,1974). Recently,therehas beenanincreasing awarenessonthe nutri- tionalaspectsof milkfat. However,beforechanging fatcontent (andpotentiallythe microstructure)of thebutter,it isessential togainmore informationonhow theindividualparameters, such aspresence ofmilkfat globules,creamheat treatmentandfluctu- atingtemperatureduring storage,all contributetothe textural propertiesofbutter.Therefore,we aimtosimulate theindustrially appliedcontinuousbutter makingprocess(Fritz-me thod)togain knowledgeofthestructural andrheologicalproperties arisingfrom suchconditions. IntheFritz-methodthecream isseparatedinto buttergrainsandbuttermilkina churningcylinder,followed by processingofthebuttergrains andfinallyevacuation (Frede& Buchheim,1994).Ouraim wastoinvestigate theeffectof cream heattreatmenton thefinal buttertexture.Hence, westudyhow thethermalhistory ofthecream affectsrheologicalproperties, so- lidfatcontent, microstructureandthe crystallizationcharacteris- ticsofthe butter,i.e.the structuralorganization ofthesolid fat. Further,wecompare sampleswithand withoutfatglobules. Theexactcrystallization characteristicsof thefatis influenced bymanyfactors suchasthe wayinwhich thesampleis cooled fromthebulk fat(Herrera&Hartel, 2000a,2000b,2000c; ten Grotenhuis,vanAken,vanMalassen, &Schenk,1999 )andthe mechanicaltreatment( Heertje,1993).Also, thebroadrange oftria- cylglycerolsfoundinmilk fatresultsin differentpolymorphic formsasa resultofvarying chainlengthand degreeofsaturation offattyacids. Thepolymorphism ofthevarious constituentscan beidentified byX-raydiffraction andhasbeen thetopicof numer- ousstudies (Fredricketal., 2011;Lopez,Lavigne, Lesieur,Keller,& Ollivon,2001;Lopez, Lesieur,Bourgaux, &Ollivon,2005; Lopez etal.,2002; tenGrotenhuiset al.,1999;Wiking, DeGraef,

0308-8146/$-see frontmatter!2012ElsevierLtd. Allrightsreserved.

http://dx.doi.org/10.1016/j.foodchem.2012.05.087 ! Correspondingauthor.Tel.: +4523983044; fax:+453533 3190.

E-mailaddress: roenholt@life.ku.dk(S.Rønholt).

FoodChemistry135 (2012)1730-1739

Contentslistsavailable atSciVerseScienceDirect

FoodChemistry

journalhomepage: www.elsevier.com/locate/foodchem Rasmussen,&Dewettnick, 2009).The mainstructuralcharacteri s- ticofthe moleculararrangem entsoftriacylglycerols isthatthey packintolamellar structures inthelongitudina ldirection;mostof- teninstacks eithertwo orthreefatty acidslong(termed 2Lor3L respectively).Thechainslie perpendicularto thelamellaeplanes (possiblytilted)and thein-plane packingofthe chainsgivesrise tovariouspolymorphic forms.Thethree mostimportantof these foundin fatsaredenoted a,b 0 andbinorder ofincreasingstability (Mazzanti,Marangoni, &Idziak,2009 ).Thedifferent polymorphic formsarecharacteri zedbythe shortandlongd-spacingsoftheir crystallatticewhich constitutesanidentifying structuralfinger- print(Larsson,1966;ten Grotenhuisetal., 1999;Vaeck,1960 ). Todeterminethis fingerprintanddescribe thecrystalstructure oneneedsknowledge ofboth thelateralpacking andthelongitu- dinalstacking. Inpreviousstudies, thepolymorphiccharacterist icsofmilk fat aretypically studiedinmodel systemswithfocus ontheeffect of temperaturetreatment andprocessingconditions onthecrystalli- zationkinetics.In thepresentstudy, westudypolymorphism , microstructureandrheologyof butter.Moreover,we studythe crystalpolymorphismin cream,subjectedto differenttemperature treatmentspriorto buttermaking. Wequantifythe propertiesof thebutterusing avarietyof characterizationtools. SmallandWide AngleX-rayScattering (SAXSandWAXS) istodescribe thecrystal- lizationstatebefore (i.e.ofthe cream)andafter buttermanufactur- ing.Further,we combinetheSAXS andWAXSwith Differential ScanningCalorimetry(DSC) tofollow thethermalevolution of thecrystallinityon subsequentheating.Light scatteringis used tostudy milkfatglobule sizeandzeta potential.Rheological mea- surementsare usedtoquantify themechanicalpropertie softhe fat crystalnetwork andconfocallaser scanningmicroscopyto visual- izethemicrostructur eofthe butter.Finally,usingLowResolution NuclearMagneticResonance (LR-NMR)wemeasure thewater dropletsizedistribution andsolidfat contentofthe butter.

2.Materialsand methods

2.1.Materials

Cream(38%fat) andskimmedmilk (0.1%fat)were collected fromthelocal supermarket. Theywereall fromARLAFoodsDairy inSlagelse,Denmark. Sodiumazide fromSigmaAldrich, St.Louis, USAwasadded toavoidmicrobial growth(0.2g/L cream).Anhy- drousmilkfat fromARLAFoods, Götene,Swedenwas usedfor thereferencesamples. Fluorescein-5-isothiocyana t(FITC)(Merck,

Damstadt,Germany),Nile redand1,1

0 -dioctadecyl-3,3,3 0 ,3 0 -tetra- methyl-indodicarbocyanineperchlorate(D307)(Molecular Probes, Taastrup,Denmark)was usedasfluoresce ntdyesfor theconfocal laserscanning microscopyimages.

2.2.Samplepreparation

Fivebuttersamples withdifferingcream heattreatmentwere preparedinlaboratory scaleintriplicate. Furthermore,threerefer- encesamples ofanhydrousmilk fatandskimmed milkwerepre- pared,alsoin triplicate,and subjectedtodifferent heat treatmentsaccordingto thebuttersamples. Thevarioustreat- mentswere:slow cooling(butter andreference), fastcooling(but- terandreference) ,slow coolingandmaturing(butteronly),fast coolingandmaturing (butteronly) andfinallyfast coolingand storageatfluctuating temperatures(butter andreference).In order tocontrolthe heattreatmentof thecream,a laboratory-scalebut- termakingmethod wasdevelopedand systematicallyapplied.To eraseallcrystal memorythecream washeatedto 65"Cfor

10minfollowed byeither fast(7.5"C/min)orslow cooling

(0.4"C/min)tochurning temperature(10 "C).Forthe matured samples,thecream wasstoredat 5"Cfor48 h.Non-matured sam- pleswereprepared fromthecream immediatelyafterreaching

10"C.Thesamples storedat fluctuatingwereproduced fromfast

cooledcream,and aftermanufacturing storedat5 "Cfor3 h,

20"Cfor 3h,followed byninecycles of1h at5"Cand1 hat

20"C.Thecream wassubjectedto phaseinversionin akitchenma-

chineandworked inafood grinder(Beem,Gigant ES-10/12)fol- lowedby vacuumtreatmentto removeair.The reference sampleswereprepared fromanhydrousmilk fatmeltedat 65"C for15min andskimmedmilk accordingtothe watercontentin thebuttersamples. Itshouldbe noted,thatbutter bydefinition containsmaximum 16%ofwater. Inthiswork thewatercontent variesfrom24.6% to27.3% (w/w).Eventhough oursamplesdo notmeetthe formalrequirements wewillstill refertothe samples asbutter.

2.3.Lightscattering

Initially,themilk fatglobulesize andzetapotential ofthe creamwasmeasure dusinglight scattering(MalvernMastersizer connectedtoa 50ml stirringunitand MalvernZetasizer,Malvern InstrumentsLtd.,Malvern,UK).For determinationofzeta potential therefractiveindex (RI)ofthe creammustbe known.Forall sam- plesitwas assumedthatthe RIwas1.39 (Calhoun,Maeta, Roy,Bali, &Bali,2010 ).Thezeta potentialgeneratedfrom thesamplesis a combinationofthesignal fromthefat globulesasserum phase inwhichthey aredispersed. Consequently,thesignal obtained fromtheserum phasemustbe subtractedtoget thezetapotential ofthefat globules.Aserum phasewastherefore preparedbycen- trifugationat16,100rpmfor 1hat 4"C(SL16Rcentrifuge from Holm&Halby,Brøndby,Denmark).Theserum phasewasremoved withasyringe. Anyproteins remaininginthe bottomofthe centri- fugetubewere removedand redispersedinthe plasmaphasedur- ingultrasonicationfor 30minfollowed bycentrifugationat

3400rpmfor 0.5h, asdescribedby WadeandBeattie (1997).All

measurementsweredonein triplicateonsamples diluted1:10 withwater.The distributionsof fatglobulesizes werederivedfrom measurementsoncreamdilutedin deionisedwater.The volume- surfacemean diameter(d 3,2 )was calculatedusingthe Malvern Mastersizersoftware. Themeasurements ofzetapotential and globulesizeswere conductedtheday thesampleswere prepared.

2.4.Drymatter

Watercontent(dry matter)wasmeasured induplicateon all samples.Thesamples wereplaced inanoven at100"Cfor2 hfol- lowedby30 mininan exicatoratroom temperature.Thewater contentwascalculated asthe% w/wdifferencebefore andafter heating.

2.5.Conductivity

Conductivitywasmeasured withahand-held conductmeterin thefinalbutter sample(Cond.330i/SET, WTWWissenscha ftlich- TechnischeWerstättenGmbH, Weilheim,Germany).The results aretheaverage ofthreemeasureme nts.

2.6.Lowresolution nuclearmagneticresonance

Thesolidfat contentandwater dropletsizedistribut ionwere determinedinall butterandreference samplesusinga Brukerwide lineLR-NMR system(BrukerMinispec mq20,Bruker OptikGmbH, Ettlingen,Germany)equipped witha pulsedgradientfield unit, operatedat5 "C.Thesamples wereobtainedby pressingthe NMRtubes(0.8 cmindiameter forwaterdroplet sizedistribution S.Rønholtet al./Food Chemistry135(2012) 1730-17391731 and0.5cm indiameterfor solidfatcontent) toaheight of2 cm intothesamples atrandomlocations byhand,as describedby Rousseau,Gosh,and Park(2009).Thesamples wereloadedcare- fullyto avoidanyair space.Theshown resultsarethe averageof threeruns. Thesizedistribution isgivenby thevolume-weighted geometricmeandiameter (d 3,3 ),as definedbyAlderliesten(1990).

2.7.Rheology

AnARG2 Rheometer(TAInstrume nt,WestSussex, England) equippedwith atemperaturecontrolle dflutedcup (radius

30mm)and vane(radius28 mm,height42 mm)geometrywas

usedinall measurements.Time sweepswereconducted for21h at5 "C(frequency1 rad/s,strain of0.1%)and 3hat 5"C,3h at

20"Cfollowed byninecycles of1h at20"Cand1 hat5 "Cfor

thesamplestored atfluctuatingtemperatures. Thefrequency sweepswereperformed inan intervalof500-0.05 rad/sdivided into21steps. Thecriticalstrain wasconstantat 0.1%.Theampli- tudesweepswere performedinan intervalof0.001-10 %straindi- videdinto21 steps.Allmeasureme ntswereperformed withinthe linearviscoelastic region(tested,data notshown).Strain atfrac- turewasdefined asthe strainvaluewhere a10%and 50%decrease oftheelastic modulusrelative totheplateau inthelinear visco- elasticregion wasseen.Results aretheaverage ofthreeruns.

2.8.Confocallaser scanningmicroscopy

TheLeicaSP5 (LeicaMicrosystems,Wetzlar GmbH,Wetzlar, Germany)confocallaser scanningmicroscope withkrypton/argon andhelium/neonlaser wasusedto captureimages.A water immersionobjectivewas usedfor a63!magnification.FITC,Nile redandD307 wereusedas fluorescentdyesin a0.01%(v/v) solu- tion.Thedyes wereimmersedon acooledobject glassuntilthe solventevaporated;the samplewasplaced andequilibratedat

5"Cfor30 min.

2.9.X-rayscattering

X-rayscatteringwas performedatthe SAXSLabinstrument(JJ- Xray,Denmark)at theUniversity ofCopenhagen,and equipped witha100XL +micro-focussealed X-raytubefrom Rigakuanda

2D300K Pilatusdetectorfrom Dectris.Measurementswere per-

formedwitha pin-holecollimated beamwiththe detectorposi- tionedasymmetrically toyieldasinglemeasurement q-rangeof

0.05-2.8Å

"1 withthe magnitudeof thescatteringvectordefined byq=4p/ksinh,where k=1.54Å isthe X-raywavelengthand his halfofthe scatteringangle. Inthissetting SAXSandWAXS are measuredsimultaneously sothatallrelevantpeak information forbothshort andlongspacings canbeobtained inasingle mea- surement.Thed-spacingsarecalculated asd=2p/q ⁄ ,whereq ⁄ is theBraggpeak position.The sampleswereloaded at5"Cincooled sampleholdersand sealedbetween5 and7lmthickmica win- dows.Thebackground scatteringfromthe micawassubtracted fromthesample spectra.The cooledsampleholders wereloaded ontoatemperature controlledsamplestage fromLinkam.A tem- peratureramp of2"C/minwasused withsimultaneous 60slong scatteringmeasurements followingtheDSCtemperaturescans.

2.10.Differentialscanning calorimetry

Isothermalinformation ofthecrystallizationkineticswasob- tainedbymeasuring withDSC usingaMettler ToledoDSC(Mettler Toledo,Greifensee,Switzerland ).DSC measurementswereper- formedinthe 5-60"Ctemperaturerange duringheatingwith a scanrateof 2"C/min.Priorto heating,thesamples wereheldfor

10minat 5"Ctoensure asampletemperature of5"C.Sample

masseswere20-30 mg.Anempty sealedpanwas usedasa reference.

2.11.Statisticalanalysis

Statisticalanalysiswascarriedout usingGraphPadPrism (Ver- sion5.02, GraphPadSoftware,Inc., LaJolla,CA, USA).Aone-way analysisofvariance (ANOVA)followed byTurkey'smultiple com- parisontestwas usedinthe dataanalysis.ANOVA wasapplied ontherepeated measurements(at leasttriplicate)of alleightsam- plesusingaverage diameterofboth averagesizeof milkfatglob- ulesincream andwaterdroplets inbutter,zeta potential, conductivity,solidfatcontent, watercontent,strain at10%and

50%decreasein G

0 ,respectivelyand finallyG 0 asvariables. Inaddi- tion,thedata structurewasanalysed byaPrincipal Components Analysis(PCA) onthemean responsesofaverage waterdroplet size,watercontent, solidfatcontent, strainat10% fractureand G 0 at5rad/s). Unscrambler(Version9.2, CAMOA/S, Trondheim, Norway)wasused forthePCA analysis,whereall variableswere standardised(1/SD).Afullcross validationwasused asmodelval- idationcriterion.

3.Resultsand discussion

Ithasbeen foundthat mechanicaltreatmentof milkcandisrupt thefatglobules leadingtochanges ofthefat globulemembrane (Morin,Jiménez-Flores, &Pouliot,2007).Sincethe surfaceofthe fatglobules canactas acatalyticimpurity, changesinthe mem- branecanlead todifferences innucleationproperties andthereby affectthefat crystalnetwork.Also, suchchangeswould enablethe milkproteinsto adsorbonthe damagedspot.This willchangethe surfacecomposition andherebyaffectthezetapotential (Wade& Beattie,1997 ).Toensure thatthefat globulesremainedintact after heattreatment,zeta potentialand fatglobulesize distributions weremeasured( Table1).No significantchangeswere observed inzetapotential orsize distribution(Fig.1)afterheating ormatur- ingofthe cream.The averagefatglobule sizeincommerc ialcream wasfound to2.48lmcorresponding tothesizesfoundinour test cream.Asa result,weconclude thattheheat treatmentofthe creamdidnot causeanychanges tothefat globules.Previously, WadeandBeattie (1997)measuredthezetapotential to"22mV incommercialcream, whichagreewith ourfindings. Toensurethat aphaseinversion fromoil-in-waterto water-in- oilhadoccurred ,conductivitywas measuredinallbuttersamples. Inall casestheconductivi tyappearedvery low(seeTable2).This confirmsformationof anon-conducting water-in-oilemulsion, withacontinuous fatphase( Bordietal., 1996). Alsothewater contentwasmeasured inall samples(Table2), andavariation lessthan3% wasobserved.Even thoughthewater contentinthe referenceslowcooled andfluctuatingsamples was statisticallydifferentfromthe othersamples( Table2),thisvaria- tiondidnot affecttherheological propertiesofthe butter(data notshown).

Table1

Averagediameter( d

3,2 )andzeta potentials( f)forfat globulesfromcommercial cream andcreamwith acontrolledtemperature history.

Creamf(mV)d

3,2 (m)

Commercialcream "27±6 2.48± 0.1

Fastcooledcream "28±5 2.46±0.02

Slowcooledcream "28±4 2.53± 0.03

Maturedfastcooled cream"30±2 2.48±0.05

Maturedslow cooledcream"27±4 2.54±0.1

Fluctuatingtemp.cream "33±4 2.39± 0.08

1732S.Rønholtet al./Food Chemistry135(2012) 1730-1739

Thewaterdroplets aredispersedwithin thecontinuousfat phaseofbutter, andform anetworktogether withthefat globules. Sofar,it isnot fullyestablishedhow thewatercontent affectsbut- termicrostructure. Itislikelythatincreasingthe watercontentin thesamplesreduces thenumberof contactpointsbetween thefat crystalshence weakeningthefat crystalnetwork.Furthermore, sincetheviscosity ofwateris lowerthanthat ofthecrystallized fat,onecould speculatethatincreasing thewatercontent might decreasethe hardnessofthe butter,whilea decreasingwaterdrop- letsizewould increasebutterhardness. Bychanging theheattreat- mentofthe cream,wecan affectthefat crystalsandthereby the networkwithinthe butterincludingthe waterdropletsize. Also, thesize ofthewater dropletsinbutter influencesthemicrobial stability,aswell asthe sensorialpropertiesof buttersuchas spreadabilityandmouthfeel (Vanlent,Vanlerberghe, VanOos- tveldt,Thas, &Vander Meeren,2008).Inthis workLR-NMRwas usedto determinethewater dropletsizedistribut ioninbutter (Ta- ble2 ).Themeasured d 3,3 wasin therangefrom 8.9to14.40 lmin buttersamples, and28.65to 37.00lminthe referencesamples. Nostatisticallysignifican tdifference wasobservedbetweenthe varyingheattreatments. Wemeasuredd 3,3 incommercial butter to1.90lm,whichis inagreementwith theliterature,where d 3,3 ismeasured incommercialnon-salted butter,witha fatcontent of82%,to 2.3-6.4lm(Vanlentet al.,2008)and 2.6-10.6lmin commercialspreadswith 40-80%fat( vanDalen,2002 ).Thewater dropletsizesin thisworkare slightlyhigherthan thevaluesre- portedinthe literature,whichis likelydueto thesmallerscale manufacturingcomparedtotheindustrial andalsoan increased watercontentcompared tothe samplesmeasured inprevious studies(Vanlentet al.,2008andvanDalen,2002 ). Thesolidfat contentdeterminationshows nosignificant changesasa resultofthe creamcoolingrate (Table2).However, withrespectto thereferencesamples asignificantdifference was observedbetween thefastcooled sampleandthe samplestored atfluctuatingtemperatures. Furthermore,the solidfatcontent in commercialbutterwassignificantly highercompared tooursam- ples.Oursamples weremeasured 24hpost manufacturingwhile thecommercialbutter wasmeasured3 weekspostmanufacturing andconsequently allowingmoreliquidfatto solidify.Inour sam- ples,theobserved rheologicaldifferences canfullybe explainedby thedifferentphysical characteristicsin thevarioussamples, with anexception ofthedifference betweenthefast cooledreference sampleandthe referencesamplestored atfluctuatingtempera- tures.Thisconfirms thefindings ofWikinget al.(2009),where milkfatsubjected todifferentcooling rateshada similarsolid fatcontent,but differedinrheologic alpropertiesand microstruc- ture.In addition,variablessuch assizedistribution sofboth water dropletsandfat globulestogetherwith chemicalcompositionof thesamplesare alsoexpected toaffectthe rheologicalproperties oftheproduct (Afoakwa,Paterson, &Fowler,2007 ).

3.1.Effectof creamcoolingrate andmaturingtime

Inthepresent study,the creamwascooled ateitherfast orslow rateprior tobuttermanufactur ing.Inaddition, theeffectsof 48h maturingof thecreamat 5"Cwerestudied afterfastand slow cooling,respectively. Smallandlargeamplitudeoscillatory shear rheologywas appliedtodetermine theelasticmodulus ofthebut- ter(Fig.2). Ithasbeen shown,thatthe elasticmodulusof afatcrystal net- workisdirectly relatedtothe hardnessindex(as determinedby conepenetrometry).This makestheelastic modulusareliable indicatorofthe macroscopic consistencyofbutter (Narine& Marangoni,1999).Toevaluate thestrainbrittleness ofthesamples, thestrainapplied toobtaina 10%and50% decreaseinthe elastic modulusrelativeto theelasticmodulus inthelinear viscoelastic regionwasrecorded (Table3). Inthenon-matured buttersamples, nosignificantdifferenc es wereobservedin elasticmodulus( Table4)orbrittleness (Table3 ). Interestingly,maturingofthe slowcooledcream significantlyin- creasedtheelastic modulusfrom0.30 to0.38MPa comparedto butterproducedfrom non-maturedcream. Maturingofthe cream enhancescrystal growth,asshown inFig3.Here,the microstruc- ture21h aftermanufac turingisrevealed byconfocallaser scan- ningmicroscopy. 110
2 4 6 8 10 12

Differential volume [%]

Droplet diameter [µm]

Cream fast Cream slow Matured cream fast Matured cream slow Reference cream Fig.1.Sizedistributionof milkfatglobules obtainedfromlight scattering measurementsof cream.Thecream sampleswerecooled from65to 10"Ceither fast,7.5"C/minorslow 0.4"C/min.Two sampleswerestored at5"C/minfor48 h afterfastand slowcooling,respectively, whileonewas storedatfluctuating temperatures(3h at5"C,3h at20"C,8cycles of1h at20"Cand1 hat5 "C).

Furthermore,a referencecreamwas analysed.

Table2

Conductivity,solidfat content,averagewater dropletdiameter( d 3,3 )andwater contentinthe butter(B)and referencesamples (R).

SampleConductivity( lS/cm)Solidfat content(%)d

3,3 (lm)Watercontent (%w/w)

Commercialbutter0.03 ±0.0060.09 ±0.153

a

1.90±0.00

a

16.00±0.02

a

BFastcooled 1.41±0.23 51.74±0.866

b,c

11.46±1.33

b,c

27.15±0.32

b

BSlowcooled 1.62±0.27 50.19±2.18

b,c

11.63±1.56

b,c

27.33±0.89

b

BMaturedfast cooled1.48± 0.1749.81± 2.42

b,c

11.19±3.47

b,c

27.15±0.32

b

BMaturedslow cooled1.64± 0.2147.75± 3.17

b,c

8.920±2.04

b,c

26.39±0.26

b

BFluctuation0.28 ±0.1847.47 ±0.916

b,c

14.40±1.47

b,c

25.99±0.21

b,c

RFastcooled 0.59±0.10 46.04±2.05

b

28.65±12.88

b,c,d

26.10±0.78

b,c

RSlowcooled 0.44±0.14 48.20±2.19

b,c

37.00±3.91

d

26.87±0.91

b

RFluctuation0.56 ±0.2053.92 ±3.00

c

35.00±1.38

d

24.64±0.48

c

Thesuperscriptletters withineachcolumn indicatessignificantdifferences amongthevalues (P<0.05)according theone-wayANOVA analysis.Samplesthat areNOT

significantlydifferent fromeachother aredenotedwith theSAMEsuperscript letters. S.Rønholtet al./Food Chemistry135(2012) 1730-17391733 Sincethedyes areonlysoluble inliquidphases, thefatcrystals appearasgrey-black zonesinthe images.Inall non-maturedslow cooledsamples,the crystalsshoweda widecrystalsize distribu- tion,whilethe fastcooled sampleshada narrowdistribution. Thedifferencein thenumberof crystalsandcontact pointsbe- tweenthemis alsonoticeable.The fastcooledsamples hadam ore densecrystalnetwork withmorecontact points,comparedto the slowcooledsamples. Interestingly,no differenceinmicrostructur e wasobservedbetween thematuredsamples, whichallrevealed a densecrystalline network.Thisexplains theobservedrheologic al properties.Maturingoftheslow cooledsampleresults inamore densecrystalnetwork withanincreased numberofcontact points givingsignificant increaseinelasticmoduluscompared tononma- turedbutter( Table4).Theobserved structuraldifferencesin the non-maturedsamplesaremostlikely aconsequenceof the changesin coolingrate.One couldspeculatethat afastercooling rateacceleratenucleation rateand therebycrystalgrowth. Conse- quently,thecrystals willbesmaller intheearly stagesandhave a somewhatfasteraggregation rate(Walstra,Kloek,& vanVliet,

2001).Smaller crystalsgivea firmernetworkof fatproducts,con-

verselylargercrystals givea sandymouthfeel( Mulder&W alstra,

1974).Aggregationof fatcrystalsoccurs duetoVan derWaals

attractionassoon astheyhave obtainedagiven size.Thisexplains whynodifference isobservedbetween thesamplesafter 48hof creammaturing; allcrystalsmayhavebeenreached thecritical sizeforaggregation atthis point. Withrespectto brittleness, maturingofthe fastcooledsample decreasedthebrittlenesscomparedto thenon-matured.In theref- erencesamples,the fastcooledhad asignificantlylower elastic moduluscomparedto theslow cooled,according totheANOVA. Previously,HerreraandHartelstudied theeffectof fast(5.5"C/ min)andslow (0.2"C/min)coolingon therheologicalpropertie s (2000c)andmicrostru cture(2000aand2000b)offractionated milk fatduringagitation. Slowcooling causedformationcrystals witha widesizedistribution andmorecontact pointscompared tofast cooling,withsmaller crystalsseparatedby aliquidphase. Inagree- mentwith ourfindings,slow coolingcauseda highermodulus

Fig.2.Frequencysweepof butterproducedfrom creamwithdifferent temperaturehistory(left column).The elasticmodulus( G

0 )is showninthe topandthe viscous modulus(G 00

)atthe bottom.Two coolingratesare used:slowcooled (0.4"C/min)andfast cooled(7.5"C/min).Thereference samplesareprepared fromanhydrousmilk fat

andwater (rightcolumn).The butterandreference samplestoredat fluctuatingtemperatureswas preparedfrom fastcooledcream andafterproduction subjectedto3 hat

5"Cfollowedby 3hat 20"Candnine cyclesof1 hat 5"Cand1 hat20 "C.

Table3

Strainappliedto obtaina 10%and50% decreaseinthe elasticmodulusrelative tothe elasticmodulisin thelinearviscoe lasticregime.Both butter(B)and reference samples(R)were studied.

SampleStrainat 10%

decreaseG 0 (%)

Strainat50%

decreaseG 0 (%)

BFastcooled 0.04

a 1.0 a

BSlowcooled 0.04

a 1.0 a

BMaturedfast cooled0.04

a 1.3 b

BMaturedslow cooled0.04

a 1.0 a

BFluctuation0.01

b 0.19 c

RFastcooled 0.02

c 0.19 c

RSlowcooled 0.02

c 0.13 d

RFluctuation0.02

c 0.18 e Thesuperscriptletters withineachcolumn indicatessignificantdifferences among thevalues( P<0.05)according theone-wayANOVA analysis.Samplesthat areNOT significantlydifferent fromeachother aredenotedwith theSAMEsuperscript letters.

1734S.Rønholtet al./Food Chemistry135(2012) 1730-1739

comparedtofast cooledsamples.Later, Wikingetal. (2009)stud- iedtherelation betweencrystallization mechanismand micro- structureinmilk fat.Themilk fatwascooled atwithout agitationeither0.1 or10"C/minto20 "Cfollowedby isothermal crystallization.Theyfound afirmercrystal networkinthe fast cooledsampleresulting inahigher complexmoduluscompared totheslower cooledmilk.Similar toourfindings andthoseby HerreraandHartel (2000c)andWikingetal. (2009)are,thatan in- creasednumberof contactpointsbetween thecrystalsincreases theharnessof thesamples. Besidestherheological profile,otherfactors contributetothe consistencyandappearanceof afatcrystal network.Accordingto theliterature,the elasticmodulus isrelatedto solidfatcontent viathefractal dimensionof thefatcrystal network(Narine& Marangoni,1999).Thefractal dimensiondescribeshow themass andsizeof thecrystalaggregates grows.Sincethe solidfatcontent didnotchange asaconsequence ofvaryingthe thermalhistoryof thesamples,the observedchangesin theelasticmodulus mustbe largelyrelatedto changesinthe fractaldimension. Therefore,we usedSAXSand WAXStostudy thecrystalpolymorphi sm.

Milkfatcrystallizes inthreepolymorphic forms:a,b

0 andb, whereaisleaststable andbmoststable.We identifiedthepoly- morphismofthe crystalsusingSAXS andWAXS.From theSAXS andWAXSspectra (Fig.4),wecan identifythepolymorphic forms ofthefour creamprecursorsamples. TheWAXSdata showthatthe non-maturedcreamsamples mainlyformaandb 0 withminor tracesofb,asalso foundbyLopezetal. (2005).Maturingof the creamleadsto atransitionfrom atob 0 andb.Thesame conclusion canbedrawn fromtheSAXS datawherethe non-maturedsamples showana-related3L structurearound67 Åcombinedwith the typicalb 0

2Llamellaearound 41Å (Lopezetal., 2005).Uponmatur-

ing,the3L arrangementchanges toa57 Åstackingindicatingcrys- talrearrangement fromatob 0 ,whichagree withour findings.

Table4

ANOVAconductedon theelasticmodulus (G

0 )of allsamples.The samplesaredenoted RF:fastcooled reference,RS:slow cooledreference,BF: fastcooledbutter, BS:slow cooledbutter,BMF: fastcooled andmaturedbutt er,BMS:slow cooledandmatured butter,RFL:Reference storedat fluctuatingtemperaturesand BFL:fastcooled butter storedatfluctuating temperatures.

ANOVAanalysisof G

0

BFBSBMF BMSBFLRF RS

BF BSNS BMFNS * BMSNS *** NS BFL ********* *** RF ******** ****** RS *** NS ***** ***** RFL ********* *** NS ******

NS=no significantdifference.

*

P<0.05.

**

P<0.01.

***

P<0.001.

Fig.3.Confocalimagesof buttersamples.The greencolourillustrates thewaterphase (FITC),the redphaseis fat(Nilered) andbluethe phospholipids(DiD oil).Fat crystals

arecolournegative (blackshadows).(For interpretationofthe referencestocolour inthis figurelegend,the readerisreferred totheweb versionof thisarticle.).

S.Rønholtet al./Food Chemistry135(2012) 1730-17391735 Aftersufficientl ylongcrystallizationtime,m ilkfattendsto formb 0 -crystalsandsometimesalsob-crystalsarereported(de- Man,1961; Lopezetal.,2005;te nGrotenhuis etal .,1999), whichconfirms ourfindingsinthematur edcream .Thebutter sampleswereanalysedwith SAXSandWAXS21 haftermanu- facturing(Fig.5).Despi tethefact,thatonewould expectthe crystallizationtoevolvefurtherafterthecreami ssubjecte d topha se-inversionduringbuttermanufacturingand subse- quentlystoredat5"Cfor 21h,the SAXSandWA XSpattern s wereidentica lincreamandbuttersubjected tothesame ther- malhistor y.Thisisinaccordancewit hFredricketal.(2011), whofound thatmilkfatinbu lkandcre amhadidenticalcr ys- tallizationmechanismswhenrapidl ycooled.Theyobserved firsta-crystalsimmediatelyafterfa stcoolingandsecondlyb 0 - crystalsduringfurthers torageat5"C,orwhen slow cooled(de- Man,1961;F redricketal .,2011).Aposs ibl eexplanationforthis behaviouristhatthenucleati onofthef atcrystalspr imarily oc- cursinthefat globules( Fredricketal.,2011).Duri ngphase inversion,alargerfractionofthe fatgl obulesareruptured hencefullcry stallization cannotbeachieved.Thosefindings highlighttheimportanceof theheattr eatmentofthecream priortobuttermanu factur ing.

Fig.4.SAXS(Top)and WAXS(Bottom)spectra fromcreamat 5"Csubjectto differenttemperaturetreatments. CS:slow cooledcream,CF: fastcooledcream, CMS:slow

cooledandmatured creamandCMF: fastcooledand maturedcream.Vertical linesindicate thepositionsof thecharacteristicpeaks fromthepolymorph sforms:a:4.15Å, b

0 :

3.81Å,4.2 Å,b:4.6Å (Lopezetal., 2005).

1736S.Rønholtet al./Food Chemistry135(2012) 1730-1739

InFig.6,theintensity ofeachdiffraction peak(WAXSmiddle andSAXSbottom) isplottedas afunction oftemperatureduring heatingtogetherwith DSCrecording.The DSCrecordingsreveal thetransitionof atob 0 -crystals.Comparingthe creamheattreat- ments,theDSC measurementsshowed nochangesin themelting ofthea-crystalsbetweenbutter producedfromfast andmatured fastcream,whereas theb 0 -crystalsmeltedat highertemperatures forthe maturedsamples.This indicatesahigher stabilityofthe a- crystalsinthe maturedsamplescompared tonon-matured, since moredrivingforce isneeded totransformation intob 0 -crystals (tenGrotenhuiset al.,1999 ). Summingup,heat treatmentofcream priortobutter manufac- turingdoesnot significantlyaffect therheological propertiesof the butterorsolid fatcontent. However,themicrostructure isaffected. Maturingofthe slowcooledcream significantlyincreasesthe hardnessof thebutterwhereas maturingoffast cooledcreamsig- nificantlydecreases thefragilityofthebutter.

3.2.Effectof globularstructure

Buttercontainsa smalleror largerfractionof milkfatglobules, dependingonthe processingconditions. Itisa widelyaccepted thatincreasingthe numberoffat globulesinbutter decreasesthe hardnessof theproduct:when thefatglobules aredestroyedmore crystallineinterglobular phaseisformed,herebyincreasing the hardnessofthe butter(Juriaanse&Heertje, 1988;Mulder& Wal- stra,1974).Nevertheless,FedotovaandLencki (2010)foundbutter containing60%globular fattobe morebrittleand lessspreadabl e comparedtobutter madefromanhydrous milkfat,when tested withconepenetrometr y.Accordingl y,wecomparedbutterpro- ducedfromfast andslowcooled creamwithreference samples ofanhydrous milkfatsubjectedtothe sameheattreatment. Skimmedmilkwas addedtothe referencesamples,to ensureequal watercontentin allsamples.The imagescapturedwith confocal laserscanningmicroscopy (Fig.3)clearlyshow adifferentmicro- structureinthe referencevs.the buttersamples.All buttersamples containedsimilaramounts offatglobules, whilenonewere ob- servedinthe referencesamples. Themeandiameter offatglobules wasabout2-3 lm(seeTable1(numberweighteddiameter) and Fig.1(volumeweighteddiameter)), whichison theorderof the sizefoundby MulderandWalstra (1974).The averagediameter ofthefat globulesobservedby confocallaserscanning microscopy iswellwithin thissize range. Thefrequencysweep (Fig.2andTable4)showsno significant differencebetweenthe slowcooled butterandreference. However, thefastcooled referencehasa significantlylowerelastic modulus comparedto thefastcooled butter.Furthermore, asignificantdif- ferenceis observedinbrittleness (Tables3and 4),thereference samplesfracturesat afivefoldlower straincomparedto thebutter samples.Withrespect tomicrostructure, itisnot possibletodistin- guishbetweenthe crystalsizebetween thereferencesamples (Fig.3),likewise theobservedWAXS patternwereidentical for notonlythe fastandslow cooledreferencesamples, butalsosim- ilartothe fastandslow cooledbutter. Inthemicrostru cture,morecontact pointsbetweenthecrystals areobserved inthesamples containingfatglobules (Fig.3).Thefat globulestakeup somespacein themicrostructure,moving thefat crystalsclosertogether facilitatingcrystalaggregation andthereby developmentofastrong, andlessbrittle crystalnetwork.In addi- tion,thecompositi onoftriacylglyce rolsisexpectedtobesimilar intheanhydrou smilkfat andthecreamusedforbutter manufac- turing.Hence,the nucleationrateis likelythesame inthe refer- enceandbutter samplesubjectedto thesameheat treatment. Sincefatcrystallization primarilyoccurs withinthefat globules, ruptureleadsto imperfectcrystallization resultingina softer Fig.5.SAXS(Top)and WAXS(Bottom)spectra frombutterand referencesamples at5"Csubjectto differenttemperature treatments.RF:fast cooledreference,RS: slowcooledreference, BF:fast cooledbutter,BS: slowcooledbutter, BMF:fast cooledandmatured butter,BMS:slow cooledandmatured butter,RFL:Reference storedatfluctuating temperaturesandBFL: fastcooledbutter storedat fluctuating temperatures.Verticallines indicatethepositions ofthecharacteristic peaksfrom thepolymorphs:a:4.15Å, b 0 :3.81Å, 4.2Å, b:4.6Å (Lopezetal., 2005). S.Rønholtet al./Food Chemistry135(2012) 1730-17391737 product(Mulder&W alstra,1974).This, however,isnot reflectedin theWAXS data,sinceno differenceswere recorded.

3.3.Effectof fluctuatingtemperature duringstorage

Weaimedto testhowtaking thebutterin andoutof therefrid- geratoraffectsthe rheology,micro-and nano-structureofthe but- ter.Heatingof thebutterwill affectandmight evenmelta smaller orlargerfraction ofthefat crystals.Thatimplies disappearanceof somefatglobules andeventuallycoalescence ofwaterdroplets, hencechangesin themicrostructur ethatare notnecessarily reversibleuponcoolingofthe butter.Tofurther investigatethis, wetestedthe effectoffluctuating temperatureduring storage.But- terandreference producedfromfast cooledcreamwere subjected to3h at5 "Cfollowedby ninecyclesof 1hat 20"Cand1 hat5 "C. Fig.2showsthatthe samplesstoredat fluctuatingtemperatures hadasignificantly decreasedelastic moduluscomparedto allthe othersamples( Table4). Fromconfocal laserscanningmicrosco pyimagesitisd iffi- culttoquantifyd iffe rencesbetweenthes amplesstoredatfluc- tuatingtemperaturescom paredtotheonesstoredat5"C. Nevertheless,therheologicaldataindicat esthatthe microstruc- tureisposs iblychange dasaresultofthefluctuatingtem pera- ture.Previousstud ieshavefocusedontheeffec tofcold/warm/ coldtreatme ntofcreampriortobutterma nufacturi ng(Frede& Buchheim,1994;Ulberth,1989).The yconcludedtha ttempera- tureandlength ofthewarm periodsignificant lya ffectsthe spreadabilityofthebutter,whileincreasi ngtem peraturea nd lengthofheatingp eriodde creasesthehardnes softhebutter. Inconc lusion,fluctuatingtemperaturesdu ringbothcreamheat treatmentaswellasduring stor ageseemstod ecrea sethe hardnessofthebutter. Thevariables discussedabove,can bestructuredin aprincipal componentanalysis( Fig.7).Thefirst principalcomponentex- plained58% ofthevariation instrainat fracture,G 0 ,water content andaveragewater dropletsize.A highstrainat fracture,G 0 and watercontentis relatedtothe buttersampleswhile highaverage waterdropletsize ischaracteri sticforthe referencesamplesto- getherwithbutter storedat fluctuatingtemperature.The second principalcomponentprovides informationonvariation insolid fatcontent,where thereferencesample storedatfluctuating tem- peratureischaracterized byahigher number. Fig.6.Comparingpeakevolution uponheatingbetween fastcooled(BF) and matured(BMF) butter.(Top)Typical SAXSandWAXS peakprogressioncurve upon heatingfrom5 to65"C.Datafrom BF.(Bottom)Combined plotofDSC andevolution ofselected WAXSandSAXS peaks.Thepeak d-spacingsareindicated inthelegends inÅ. TheDSCspectra representatypical evolutionplotupon heatingofmilk fat. Fig.7.Principalcomponentanalysis (bi-plotofscores andcorrelationloadings) of variablesrelated totherheological propertiesofthe samples.Noticethe difference inscaleof principalcomponent1 and2,since itleadsto visuallyoveremphasising theinfluenceof principalcomponent2.

1738S.Rønholt etal./ FoodChemistry135 (2012)1730-1739

4.Conclusion

Theheattreatment ofcreamprior tobuttermanufac turinglar- gelydeterminesthe finaltexturalcharacterist ics,suchas spread- abilityandmouthfeel, ofthe butter.Slowand fastcoolingof the creammayresult insimilar rheologicalpropertiesand polymor- phicforms,a-andb 0 -crystals,butdiffer inmicrostructure. Butter producedfrom slowcooledcream hadfewercrystals withawider sizedistributionwhereas thebutter producedfromfast cooled creamconsistedof moreuniformcrystals. Maturingofthe cream maylead toatransition fromb 0 tob.Maturingof thefast cooled creamdoesnot causeanychanges intherheological profilenor microstructure.However,maturingofthe slowcooledcream sig- nificantlyincreases thehardnessof theproducedbutter, asaresult ofamore densecrystalnetwork. Fatglobulesseem toaffectnot onlytherheological propertiesof butter,but alsohowbutter behaveswhensubjected tofluctuating temperature.Absenceoffat globulesresultsin asignificantlymore brittlebutterproduct comparedtowhen fatglobulesare present. Furtherstudies areneededto exploretheimpact offatglobules inafat crystalnetwork, whichisthe caseinbutter. Suchstudies couldexpandthe previousknowledge obtainedfromsimpler mod- elsystemswith nofatglobules present.

Acknowledgments

Thankstothe DanishDairyResearch FoundationandThe Dan- ishFood IndustryAgencyfor financialsupport.Thanks toSPX Gladsaxe,Denmark,for lettingususe oftheirBruker Minispeck LR-NMRandDepartment ofPharmaceut icsandAnalytical Chemis- try,FARMA,University ofCopenhagen,for theuseof theirZeta-si- zer.Thanksto ARLAFoodsfor thefreesupply ofanhydrousmilk fat.Thanksto theDanishAgency forScience,Technology andInno- vation,Carlsberg andLundbeckfor thefundingof ourSAXSLAB- instrument.

References

Afoakwa,E.O., Paterson,A.,& Fowler,M.(2007). Factorsinfluencingrheological and texturalqualitiesin chocolate-A review.TrendsinFood Science&Technology,

18,290-298.

Alderliesten,M.(1990). Meanparticlediameters. PartI:Evaluation ofdefinition systems.Particle&Particle SystemsCharacterization,7 ,233-241. Bordi,F.,Cametti, C.,Chen,S. H.,Rouch,J., Sciortino,F.,& Tartaglia,P.(1996). The staticelectricalconductivity ofwater-in-oil microemulsionsbelowpercolation threshold.PhysicaA,231 ,161-167. Calhoun,W.R., Maeta,H.,Roy, S.,Bali,L. M.,&Bali, S.(2010). Sensitivereal-time measurementof therefractiveindex andattenuationcoefficient ofmilkand milk-creammixtures. JournalofDairy Science,93,3497-3504. DeMan,J.M. (1961).Physical propertiesofmilk fat.II.Some factorsinfluencing crystallization.JournalofDairy Research,28.149-122. Fedotova,Y., &Lencki,R. (2010).Theeffect ofphospholipidson butterphysicaland sensoryproperties.JournalofAmerican OilChemist'sSociety, 87,75-82. Frede,E.,& Buchheim,W.(1994). Buttermakingandthe churningofblended fat emulsions.Journalofthe SocietyofDairy Technology,47,17-27. Fredrick,E.,Van deWalle,D., Walstra,P.,Zijtveld, J.H.,Fischer, S.,Van derMeeren, P.,et al.(2011).Isothermal crystallizationbehaviourof milkfatin bulkand emulsifiedstate.InternationalDairyJournal, 21,685-695. Haighton,A. J.(1965).Worksoftening ofmargarineand shortening.Journalof

AmericanOilChemist's Society,42,27-30.

Heertje,I.(1993). Microstructuralstudiesin fatresearch.FoodStructure,12 ,77-94. Herrera,M.L., &Hartel,R. W.(2000a).Effect ofprocessingconditions onphysical propertiesofa milkfat modelsystem:Microstructure. JournalofAmerican Oil

Chemist'sSociety,77 ,1197-11205.

Herrera,M.L., &Hartel,R. W.(2000b).Effect ofprocessingconditions on crystallizationkineticsof amilkfat modelsystem.JournalofAmerican Oil

Chemist'sSociety,77 ,1177-1187.

Herrera,M.L., &Hartel,R. W.(2000c).Effect ofprocessingconditions onphysical propertiesofa milkfatmodel system:Rheology.JournalofAmerican Oil

Chemist'sSociety,77 ,1189-111195.

Juriaanse,A.C., &Heertje,I. (1988).Microstructureof shortenings,margarineand butter-A review.FoodMicrostructure,7 ,181-188. Larsson,K.(1966). Classificationofglyceride crystalforms.ActaChemica

Scandinavica,20,2255-2260.

Lopez,C.,Bourgaux, C.,Lesieur,P., Bernadou,S.,Keller, G.,&Ollivon, M.(2002). Thermalandstructural behaviorof milkfat.3. Influenceofcooling rateand dropletsize oncreamcrystallization. JournalofColloid andInterface Science,254,

64-78.

Lopez,C.,Lavigne, F.,Lesieur,P., Keller,G.,& Ollivon,M.(2001). Thermaland structuralbehaviourof anhydrousmilkfat. 2.Crystallineforms obtainedby slowcooling.JournalofDairy Science,84,2402-2412. Lopez,C., Lesieur,P.,Bourgaux, C.,&Ollivon, M.(2005).Thermal andstructural behaviorofanhydrous milkfat.3. Influenceofcooling rate.JournalofDairy

Science,88,511-526.

Mazzanti,G.,Marangoni, A.G.,& Idziak,S.H. J.(2009). Synchrotronstudyon crystallizationkineticsof milkfat undershearflow. FoodResearchInternational,

42.682-69.

Morin,P.,Jiménez-Flores, R.,&Pouliot, Y.(2007).Effect ofprocessingon the compositionandmicrostructure ofbuttermilkand itsmilkfat globule membranes.InternationalDairyJournal, 17,1179-1187. Mulder,H.,& Walstra,P. (1974).Structureand textureofbutter. InThemilkfat globule(pp.246-287).Wagening: Centrefor AgriculturalPublishingand

Documentation.

Narine,S.S., &Marangoni,A. G.(1999).Fractal natureoffat crystalsnetworks. PhysicalReviewE: Statistical,Nonlinear,and SoftMatterPhysics, 59,1908-1920. Rousseau,D.,Gosh, S.,&Park, H.(2009).Comparison ofthedispersed phase coalescencemechanismsin differenttablespreads.JournalofFood Science,74,

E1-E7.

Shama,F.,& Sherman,P.(1970). Theinfluenceof worksofteningon theviscoelastic propertiesofbutter andmargarine.Journalof TextureStudies,1 ,196-205. tenGrotenhuis,E., vanAken, G.A.,van Malassen,K.F., &Schenk,H. (1999). Polymorphismofmilk fatstudiedby differentialscanningcalorimetry andreal- timeX-raypowder diffraction.JournalofAmerican OilChemists Society,76,

1031-1039.

Ulberth,Von.F. (1989).Beeinflugssung derbutterstreichfähigkeitdurch ausgewähltephysikalischerahmreifungsverfahren. Milchwissenschaft,44 ,

415-417.

Vaeck,S.V. (1960).Cocoa butterandfat bloom.TheManufacturingConfectioner,

40(35-46),71-74.

vanDalen,G. (2002).Determinationof thewaterdroplet sizedistributionof fat spreadsusingconfocal scanninglasermicroscopy. JournalofMicroscopy, 28,

116-133.

Vanlent,K., Vanlerberghe,B.,Van Oostveldt,P.,Thas, O.,&Van derMeeren, P. (2008).Determinationof waterdropletsize distributioninbutter: pulsedfield gradientNMRin comparisonwithconfocal scanninglasermicroscopy.

InternationalDairy Journal,18,12-22.

Wade,T.,& Beattie,J.K. (1997).Electroacousticdetermination ofsizeand zeta potentialoffat globulesinmilk andcreamemulsions. ColloidsandSurfaces, B:

Biointerfaces,10,73-85.

Walstra,P.,Kloek, W.,&van Vliet,T.(2001). Fatcrystalnetworks. InCrystallization ProcessesinFats andLipidSystems (pp.289-328).New York:MarcelDekker. Wiking,L.,De Graef,V.,Rasmussen, M.,&Dewettnick, K.(2009).Relations between crystallisationmechanismsand microstructureofmilk fat.InternationalDairy

Journal,19,424-430.

S.Rønholtet al./Food Chemistry135(2012) 1730-17391739