Correspondingauthorsandralerougeunivparisdiderotfr 2h1andwhenh1Theoreticallyattemptstorationalizethe3Dcharacteroftheshearbanding owhavebeenmadethroughlinearstabilityanalysiswithrespectto ID: 219607
Download Pdf The PPT/PDF document "Shear-bandinginsurfactantwormlikemicelle..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Shear-bandinginsurfactantwormlikemicelles:Elasticinstabilitiesandwallslip Correspondingauthor;sandra.lerouge@univ-paris-diderot.fr 2h1andwhenh=1.Theoretically,attemptstorationalizethe3Dcharacteroftheshear-banding owhavebeenmadethroughlinearstabilityanalysiswithre-specttoaxisymmetric[30,31]andnon-axisymmetric[32]disturbancesintheframeworkofthediusiveJohnson-Segalman(dJS)model[33{36].Theelasticinstabilityhypothesishasbeenconrmed,withapossibleinterplaybetweenbulkandinterfacemodes,dependinginpar-ticularonthecurvatureofthestreamlinesofthebase ow[31,32].Themostrecentresultsindeedsuggestthatshear-banding owscanbeperturbedbyelasticinstabilities.Butmostoftheexperimentsthathelpedtobuildthispicturewereperformedonasinglesystemconsistingofasemi-diluteaqueousmixtureofcetyltrimethylammo-niumbromideandsodiumnitrate(CTAB/NaNO3).Inthepresentpaper,wefocusonawormlikemicellarsolu-tionmadeof10%cetylpyridiniumchloride(CPCl)andsodiumsalicylate(NaSal)inNaClbrine.Thissystem,originallystudiedbyBerretetal.[37,38]hasbeenin-tensivelyinvestigatedtheselastyears,especiallybytheWellingtongroupandisnowwell-knowntoexhibit uc-tuations[17].Usingnuclearmagneticresonance(NMR)velocimetry,large uctuationsofthe1Dvelocityproleswereobservedasthesystemwasquenchedintheplateauregion,thesizeofthehighshearratebandbeingdrivenbythedegreeofslipatthemovingwall[18,39,40].De-pendingonthebatch,the uctuationscouldadopteitheraquasi-randomoraperiodiccharacterandwerecorre-latedwiththe uctuationsintheshearstresstimeseries.Moreover,theproportionofthehighshearratebandin-creasedlinearlywiththeappliedshearrate,followingthesimpleleverrule.However,inarecentstudy[20]performedbythesamegroup,theauthorsobservedadierentpicture,withstrongdeparturefromthestan-dardleverrule.InTaylor-Couette owgeometrieswithsmoothorroughboundaryconditions,theyfoundthatthelocalshearratesineachbandincreasedwiththeap-pliedshearrate(withnonethelessarapidsaturationinthelowshearband),whiletherelativeproportionsofthebandsremainedessentiallyconstant.Thisanomalousbe-haviorhasbeenascribedto uctuatingslipdynamicsto-getherwithsubtlechangesinthesampleinherenttothebatch[41].Also,thisanomalousbehaviorwasascribedtotheuseofmoderateshearrateincrementsratherthanquenches,whichwerepreferredinearlierstudies.Thelocalshearrateinthehighshearratebandwasalsofoundtoexhibittime-dependentvariations,suggestinginstabilityofthisband.Besides,a2DextensionoftheNMRvelocimetrytechniqueprovidedevidenceof uctu-ationsoftheazimuthalvelocityalongthevorticitydi-rectionwithacharacteristiclengthscaleontheorderofacentimeter,i.e.anorderofmagnitudelargerthanthegap[20].The10%CPCl/NaSalsystemhasalsobeenusedrecentlytoexploretheeectsofthebound-aryconditions(BC)ontheshear-banding owbymeansoftime-resolvedultrasonicvelocimetry(USV)[19].Slipwasobservedwhatevertheboundaryconditions(smoothorrough).However,withrough(or`stick')BC,shear-bandingwithlarge uctuationsinthehighshearratebanddevelopedwhereas,withsmooth(or`slip')BC,wallslipcompetedwithshear-bandingleadingtoatwo-bandsstructurethatwasonlyintermittentlyobserved.Tosummarize,alltherecentstudiesonthe10%CPCl/NaSalsystemevidenced uctuations.Inalmostallcases,the uctuationswererationalizedbytheauthorsofthestudiesbyinvokingacorrelationbetweentheshear-bandingstructureandthedynamicsofwallslip.Butthepossibilitythatsecondary owstriggeredbyelasticinstabilitywereattheoriginof uctuationswasneverconsideredthoroughly.Inthepresentpaper,weinvestigatetheTaylor-Couette owpropertiesof10%CPCl/NaSal/brinesamplesusingglobalrheology,1Dultrasonicvelocimetryand2Dopti-calvisualisations.Ourmaingoalistounderstandmoreclearlytheoriginofthe uctuationsreportedinthissys-tem,andthediscrepanciesbetweendierentbatches.Wewishtobeabletodistinguishtheimpactofwallslipfromtheimpactofsecondary owstriggeredbyelasticinsta-bilities.Oneoftheuniquefeaturesistheuseofasin-gle owgeometrythatenablesaprecisecorrelationforalltypesofexperiments.Weobservethattwosamplespreparedfromthesamebatchandshowingquasi-similarglobalrheologicalproperties,canexhibitdiscrepanciesintheirshear-banding owdynamics.Forthersttime,webuildasolidrationaletoexplainapossibleoriginforthediscrepancies.OurworkinghypothesisreliesonthesensitivityoftheCPCl/NaSalsolutionstoambientlightexposurethatmayinducesubtlechangesatthemolecu-larscale.Toallowforasystematicstudy,wearticiallytriggermodicationsinthesamplesusingUVirradiationprotocols.1DUSVmeasurementsrevealthatthemag-nitudeofwallslipislargerforirradiatedsamplessug-gestingthattheinteractionwiththewallsismodiedbytheproductsofphotochemicalreactions.WedescribeintheElectronicSupplementaryInformationhowthecetylpyridiniumchloridesurfactantmoleculesexhibitaphotochemistrymainlyin uencedbythephoto-inducedcleavageofthepyridineringthatyieldsanunstablealde-hydeenamine,whichfurtherdecaysbythermallyacti-vatedprocesses.Theproductsofthereactionpossiblybuildupalubricationlayerresponsibleforpathological owdynamics.Overall,theuseofasingle owgeometrytoperform2Dopticalvisualisationand1Dvelocimetrydemonstratesanunequivocalcorrespondencebetweentheturbidbandob-servedusingtheformertechniqueandthehighshearratebandobservedusingthelatter.Thiscorrelation,to-getherwiththeidenticationofasourceofdiscrepanciesbetweensamples,enablesustore-interpret uctuationsobservedin1Dvelocityproles.Ourresultsdemonstratethattheshear-banding owofthe10%CPCl/NaSalisunstableduetotheelasticinstabilityofthehighshearrateband.Thebandisfoundtoundergosuccessivein-stabilitiesastheappliedshearrateisincreased,withthedevelopmentofa3Dcoherentsecondary owfollowed 3byatransitiontowardaturbulentstate.Theparticu-larsofthescenarioaredierentfromtheonesseenintheCTAB/NaNO3system,butitcanbeverywellra-tionalizedinthesameelasticinstabilityframework.Thepresenceofwallslipisconrmed,butitsroleinexplain-ingthe uctuationsisshowntobesecondary.Thepaperisorganizedasfollows.InsectionI,wehighlightthepreparationofthesamples,theirradiationprotocolsandthedierentexperimentaltechniquesem-ployed.SectionIIisdedicatedtoadescriptionofpre-liminaryobservations,whichconrmthepossibilityofdiscrepanciesbetweenfreshandoldersamplesandshowthatsamplealterationcanbearticiallyacceleratedbyUVlightirradiation.SectionIIIdescribesindetailthebehaviorsoffreshandirradiatedsamples,showingthecorrespondencebetween2Dopticalvisualisationand1Dvelocimetry.Comparisonswithspecicpointsofthelit-eraturealsoappearinthissection.InsectionIV,wediscusstheconsequencesofourstudiesinclarifyingthedierencesbetweentheshear-banding owofthe10%CPCl/NaSalsystemwithrespecttotherationaleprevi-ouslybuiltaroundtheCTAB/NaNO3system.Wedis-cusstheroleofelasticinstabilitiesastheprincipalcauseof uctuations,aswellasthelesserbutgenuineimpactofwallsliponshear-bandinganditsinterplaywithelasticinstabilities.Finally,weconcludeinsectionV.I.MATERIALSANDMETHODSA.MaterialsThemicellarsamplesweremadeof8.09wt:%(0.238M)cetylpyridiniumchloride(CPCl)with1.91%(0.119M)sodiumsalycilate(NaSal)inwaterwith0.5Msodiumchloride(NaCl).Intherheologyliterature,thissolutionisusuallycalledCPCl10%forthesumofthesurfactant(CPCl)andco-surfactant(NaSal)weightfractions[2,3].TheCPClandNaClwerepurchasedfromSigma-AldrichandtheNaSalfromAcros-Organics.TheCPClisinamono-hydratedform.Therefore,wetookintoaccountthewatermoleculetocomputetheweightfraction.CPClisasurfactantwhosehydrophilicheadisessentiallyformedbyapyridinering,C5H5N,whichisknowntobeparticularlysensitivetolight.Inaqueoussolution,thepyridineringcanbeopenedbyUVradiation,closetothepyridineabsorptionbandat253:7nm[42,43].Photohy-drationofpyridinecleavestheringandproducesalde-hydeenamine(5-amino-2,4-pentadienal).Thephoto-inducedcleavagereactionisusuallyreversibleinaqueoussolution,leadingtoanequilibriumbetweenringcleav-ageandringclosure.Howeverinveryviscouspolymersolutionsthecleavagereactionbecomesirreversible[44].Thephoto-inducedcleavagecanbemeasuredontheUV-visiblespectragiveninFig.1andisfurtherdiscussedintheESI,wherewedemonstratethatanessentiallyirre-versiblereactionalsooccursinCPClsurfactantsolutions. FIG.1.Photochemicalkineticsforvariousirradiationdura-tions.(a)UV-visiblespectraof0.05wt:%CPClsolutionsforirradiationtimesof0,15,30,45,60,90and180min.(b)Op-ticaldensity(OD)asafunctionoftheirradiationdurationforfourwavelength216nm,259nm,369nmand410nm.Therststepisthecleavageoftheringgivenbythefol-lowingreaction: Likeinthepurepyridinecase,thecleavageyieldsanunstablealdehydeenamine,whichfurtherdecaysbythermallyactivatedprocesses.Thenalproductsseemtoincludethefreefattyhexadecanetailsoftheoriginalsurfactant.Toquantifytheeectsoftheproducsofthephotochem-icalreactions,wedistinguishedthreekindofsamples,dependingonthelightexposureconditions.the`original'sample(OS)keptinacontainerpre-ventingambientlightexposure.the`oldoriginal'sample(O-OS)placedinatrans-parentcontainer.Thedesignation`old'referstothefactthat,incourseoftime,thissampletakesa 4lightyellowcolouring(seeESI),duetotemporaryexposuresofthetransparentcontainertoambientlightprecedingeachrheologicaltest.the`irradiated'sample(IS)takenfromtheOSbatchthenexposedtoanUVlightirradiationat254nmcorrespondingtotheabsorptionbandofthepyridinering[42].Theirradiationwasperformedat40Cforfourhoursona30mLmicellarsolutionlaidoutinacrystalliser,withaUVpenraylampplacedhorizontallyatapproximately1-2cmofthesolutionstirredbyamagneticbar.Wehomogenizetheexposureandincreasetheirradiationintensitybycoveringtheset-upwithaluminiumfoil,bril-liantfaceturnedtowardstheinterior.IntheESI,werefertothisprotocolas`protocol2'.Theaimoftheirradiationprotocolistoreproducearticially,onshortertimescales,thechangesobservedintheO-OS.TheISisthenkeptinacontainerpreventingambientlightexposure.Allthesampleswerestoredat35Cinanoven.Inthisstudy,thetemperatureforexperimentsisxedatT=21:5C.Notealsothatiftheresultspresentedinthepapermostlycomefromasinglesetofexperiments,thebehaviordescribedarereproducibleandhavebeenreproducedatseveraloccasionsinthepasttwoyears,andbydierentexperimenters.B.Methods1.CylindricalCouettegeometryExperimentswereperformedintwoidenticaltranspar-entsmall-gapcylindricalCouettedeviceswithsmoothwalls,alsoreferredtoasTaylor-Couette(TC)cellsinthefollowing.Thesecellswereadaptedeitherfordirectob-servationsofthevelocitygradient-vorticityplane(r;z)orforultrasonicvelocimetryoftheprimary owinthegapv(r)atagivenlocationalongthevorticityaxis.Inallexperiments,onlytheinnercylinderwasrotatinganditsaxiswasadaptedtoastress-controlledrheometer(Phys-icaMCR301).AprecisedescriptionoftheTCdeviceusedforopticalvisualisationsisgiveninSup.Fig.1.Inthecelladaptedforultrasonicvelocimetry,theonlymodicationofthedesignistheenlargementofthewa-terthermostaticbatharoundtheouterxedcylinder,toallowenoughspacefortheultrasoundtransducer.Inallexperiments,thetopofthecellwasclosedbyasmallplugwhichlimitsthedestabilizationofthefreesurfaceofthe uidathighstrainrates[23].Ahome-madesolventtrapwasalsousedtolimitevaporation.ThedimensionsoftheTCdeviceareasfollows:innerradiusRi=13:33mm,heighth=40mmandgape=1:13mm.2.Rheo-opticalset-upIntherheo-opticaldevice,thegapwasvisualizedbyusingalasersheet(wavelength632.8nm)propagatingalongthevelocitygradientaxisandextendingalongthevorticityaxis.Adigitalcamerarecordedthescatteredin-tensityat90,givingaviewofthegapinthe(r;z)plane.Theeldofobservationwascentredatmid-heightandvariesfrom0.5to2cmaccordingtothechosenmagni-cation.Weappliedanumericalalgorithmtoeachframeinordertodetecttheinterface[23,24].Flowvizualizationsinthe ow-vorticityplaneusingseedinganisotropicre ectiveparticles(anisotropicmicaplateletsfromMerckatavolumefractionof6.105),werealsoperformedtoobserveapossible3Dcharacteroftheshear-banding ow.Inthisconguration,the uidisillu-minatedbyambientlightandtheintensityI(z)re ectedinthevelocitygradientdirectioniscollectedonadigitalcamera(seeref.[24]forfurtherdetails).3.Rheo-velocimetryset-upThevelocityofthesampleinthe owdirectionwasmeasuredusinghighfrequencyultrasonicspeckleve-locimetryatanaxialpositionabout15mmfromthebottomoftheTCcell.USVisatechniquethatallowsonetoaccessvelocityproleswithaspatialresolutionof40mandatemporalresolutionof0.02-2sdepend-ingontheappliedshearrate.Itreliesontheanalysisofsuccessiveultrasonicspecklesignalsthatresultfromtheinterferencesofthebackscatteredechoesofsuccessivein-cidentpulsesofcentralfrequency36MHzgeneratedbyahigh-frequencypiezo-polymertransducer(PanametricsPI50-2)connectedtoabroadbandpulser-receiver(Pana-metrics5900PRwith200MHzbandwidth).Thespecklesignalsaresenttoahigh-speeddigitizer(AcqirisDP235with500MHzsamplingfrequency)andstoredonaPCforpostprocessingusingacross-correlationalgorithmthatyieldsthelocaldisplacementfromonepulsetoan-otherasafunctionoftheradialpositionracrossthegap.Onevelocityproleisthenobtainedbyaverag-ingovertypically1000successivecross-correlations.FulldetailsabouttheUSVtechniquemaybefoundin[45].Forthesevelocimetryexperiments,0.3wt:%hollowglasssphereswereaddedtotheOSandIS,toactasultrasoniccontrastagents[45].Theglasssphereshaveanaveragediameterof6mandadensityof1:1(PottersIndustriesInc.,UK).Thesoundspeedinoursampleswasindepen-dentlymeasuredtobe1555m/s.4.TypicalprotocolsMostoftheexperimentalresultspresentedinthispa-perareobtainedforstart-up owsataknownimposedshearrate.Typically,eachstart-uptestwasperformed 5TABLEI.Summaryofthelinearviscoelasticparametersfor10%CPClOS,O-OSandIS. Sample G0(Pa) R(s) 0(Pa.s) OS 1865 0.650.05 12110O-OS 1935 0.660.05 12710IS 2105 0.520.05 10910 fortenminutes.Inbetweeneachstart-up owexperi-ment,thesamplewasallowedtorelaxandrestwithout owfortwominutes.Whenweusedtherheo-opticalset-up,thesampleswerefreeofseedingparticles,exceptifotherwisestated.Weperformed2Dopticalvisualisationsandsimultaneouslyrecordedtheglobalshearstresstimeseries.Whenweusedtherheo-velocimetryset-up,thesampleswereseededwiththeultrasoniccontrastagents.Weperformed1DUSVandsimultaneouslyrecordedtheglobalshearstresstimeseries.Notenonethelessthatalltheresultspresentedinthepaperhavealsobeendupli-catedusingstart-up owsatimposedshearstress(i.e.creeptests).Exceptfortheparticularsoftheearlytimetransientresponse,thesamebehaviorswereobserved.II.PRELIMINARYOBSERVATIONSAttheconcentrationchosenforthisstudy,farfromtheisotropic-nematictransitionatrest,thesolutionsaresemi-diluteandmadeofhighlyentangledwormlikemi-cellesforminganelasticnetwork.TheevolutionsofthelossandstoragemoduliindicatethatthethreesolutionsOS,O-OSandIS,behaveasalmostperfectMaxwellianelementsovertheexploredrangeoffrequencies.Theelas-ticmodulusG0andrelaxationtimesRobtainedfromtstotheMaxwellmodelaregiveninTableI.Thethreesamplespresentverysimilarlinearpropertiesbutwithslightquantitativedierences:theirradiationprocessseemstoinducealargerelasticmodulusandasmallerrelaxationtime.Computationofthezeroshearratevis-cosityfrom0=G0RshowsthattheISisslightlylessviscousthantheOSandO-OS.SuchsubtlechangesinlinearpropertiesforO-OSandISaremostlikelyduetoproducsofphotochemicalreactions,whicharediscussedintheESI.Figure2displaysthecomparisonbetweenthesteadystateshearstressasafunctionoftheappliedshearrate_ fortheOS,O-OSandIS.Thelowshearratebranchesarenotstrictlysuperimposedduetotheslightdierencein0betweenthesamples,butinanycase,afteraNew-tonianregime,thesolutionsexhibitshear-thinning.Thethreeexperimental owcurvesfollowquantitativelythesametrend:theypresenttwoincreasingbranchessepa-ratedbyastressplateauatp=1202Pacharacteristicoftheshear-bandingtransition,andextendingbetweentwocriticalshearrates_ 1and_ 2.Theapparentbegin-ningofthestressplateau_ 1rangesbetween_ OS1=1.30.1s1and_ IS1=1.50.1s1,whiletheapparent FIG.2.Top:Semi-logarithmicplotofthesteadystateappar-ent owcurvesoftheOS(closedcircles),O-OS(opencircles)andIS(opensquares)measuredunderstrain-controlledconditions.Thesamplingoftheshearratesweepis120sperdatapoint.Bottom:transientresponsesatshorttimesfordierentappliedshearratesalongthe owcurve.endofthestressplateau_ 2rangesbetween_ OS2=121s1and_ IS2=171s1.Itisimportanttostressthattheexperimental owcurvesareapparent owcurves,inthesensethatthecurvatureofthegeometry,potentialsecondary owsandwallslipcanallhaveanimpactonthemeasurements,whichthendonotsolelyre ectmaterialproperties.Inparticular,thestressplateauisnot atandtheshearstressincre-mentbetweenthetwoextremities(OS25PaandIS28Pa)isonlypartiallyexplainedbythestressheterogeneityinherenttothecellcurvaturethatleadstoageometricalstressincrement=22Pa.Beyondconcentrationeectsthatcanin uencetheslope,onemustalsoconsiderthepotentialin uenceofsecondary ows[23].Also,above_ 2,theshearstressincreasesno-ticeablyfollowinganapparenthighshearratebranch.Wewillseeinthefollowingthatthebranchhasapurelydynamicalorigin.Thishighshearratebranchpresentsasharp`S'shapefortheOS.FortheIS,thisshapeseems 6broaderandtheupperpartofthe`S'isnotreachableduetoinclusionofbubblesinthesample.ThisbehaviordierssomewhatfromthebehavioroftheO-OS.Butletusrecallthattheeectsoflightexposurehavebeenar-ticiallyenhancedintheIS.AsshowninSup.Fig.2,ifasampleisirradiatedonlyfortwohours,its owcurveisinbetweentheO-OS owcurveandtheIS,whichwasirradiatedforfourhours.Thetypicaltransientresponsesoftheshearstressatshorttimesfollowingasuddenstart-upof owareshowninthesubplotsofFig.2,forvariousappliedshearratesalongthe owcurve.IntheNewtonianregion,theexpectedmonoexponentialgrowthisobserved.Atthebeginningofthestressplateau,theresponseisdomi-natedbyastressovershootfollowedbyasigmodalde-cayand/ordampedoscillationsandasmallundershootprecedingthestabilizationoftheshearstressaroundasteadystatevalue.Asdescribedindetailselsewhere[23,24],theshearstressundershootcontainstheme-chanicalsignatureoftheonsetofsecondaryvortex ows.Suchtransientstressresponses,typicalofsystemsun-dergoingashear-bandingtransition,havebeenwidelyobservedintheliterature[3,37,46{49],andwewillnotcommentonitfurther,aswewishtofocusonthelocal owbehaviorofthesamplesatlongtimesafterstart-upof ow.AsrevealedinFig.3bydirectvisualisationsinthe(r;z)planeforvariousappliedshearratesalongthestressplateau,thebandingstructureisnotidenticalinthethreesamples.Asinotherwormlikemicellarsolu-tions[23,46,50{52],theinducedbandisslightlyturbidgivingastrongopticalcontrastbetweenthetwobands.Inthethreecases(OS,O-OSandIS),thesystemisorganizedintotwomacroscopicbandsofdieringopti-calproperties,separatedbyaninterfacethatundulatesalongthevorticitydirection.FortheOS,wecanidentifyawell-denedwavelengththatincreaseswiththeappliedshearrate,whilethepatternappearsmorecomplicatedfortheO-OSandIS.TheO-OSandISbehaviorareverysimilartoeachother,withanirregularundulationoftheinterfacialproleandcontinuousprocessesofgrowthandrelaxationofturbidity uctuationsfurtherinthegap(seephotosat3,5and8s1andmoviesinthesupplemen-tarymaterial).Althoughtheproportionoftheinducedturbidbandandthewavelengthoftheinterfaceproleseemcomparableforthethreesamplesatthebeginningofthestressplateau,theybecomemuchlargerfortheOSwhentheappliedshearrateisincreased(seephotosat8,10and15s1).Thisobservationisconsistentwiththefactthattheinterfacialwavelengthhasbeenfoundtoscalewiththeproportionoftheinducedbandh[53].Inthethreesamples,astheshearrateisfurtherin-creased,theinducedbandundergoesanotherinstabil-ity.Typicalexamplesaregivenat15and17s1fortheOS,20and25s1fortheO-OS,and25and35s1fortheIS.Inthesamesnapshot,weareabletosimultane-ouslyobserveregionswherethebandscoexistwithanundulatedinterfaceandregionswheretheinducedtur-bidbandisdestabilized,the owbeinglocallystronglydisordered.Finally,the owoftheOSbecomesfullydis-orderedaboveashearrateof_ =20s1,apicturereminis-centofelasticturbulence[25{28].SupplementarymoviesanddetailsgiveninsectionIIIAwillhelptounderstandthisnew owpatternthatwewillcallturbulentbursts.Indeed,animportantpointhereisthatthetransitiontowardselasticturbulencestartswhiletheinducedbanddoesnotlltheentiregap,incontrasttoobservationsintheCTAB/NaNO3system,wherethetransitiontotur-bulenceoccursonthehighshearratebranch[25].Notethatinthethreesamples,theshearratesthatcorrespondtotheonsetofturbulentburstsalsocorrespondto_ 2,i.e.theup-turnintheir owcurve.Therefore,ifitwasnotfortheonsetofturbulentbursts,thestressplateauwouldmostlikelyextendtomuchhighershearratesuptothetruehighshearratebranch,whentheproportionofthehighshearbandwouldhavereachedh=1.Theconclusionofthesepreliminaryobservationsistwofold.First,the10%CPClmicellarsolutionisseentoexhibitsecondary ows.Atthebeginningofthestressplateau,thesecondary owsarereminiscentoftheTaylor-likevortex owpreviouslyidentiedintheCTAB/NaNO3system[22{25].Butathighershearrates,andbeforetheendoftheplateau,weobservetheonsetofturbulentburstswhichtemporarilyandlocallydisturbthebandingstructure(Fig.3).Second,wehaveobservedsomedierencesintheparticularsofthisscenariobe-tweenfreshsamples(OS)andagedsamples(O-OS).AndUV-lightirradiationcanreproducethesamplealteration,sincethebehavioroftheO-OSandISareessentiallyidentical.Sincethealterationofasampleisdiculttocontrolandcanariseoververylongtimes(typicallysev-eralmonths),makingasystematicstudydicult,inthefollowing,wefocusonlyonthecomparisonbetweenthe owbehaviouroftheOSandIS.III.RESULTSA.2Dopticalvisualisation{Secondary owpatternsWenowwishtocompareindetailtheinterfacedy-namicsfollowingasuddenstepshearratefromrestforthetwosamples(OSandIS).Fromtheinterfaceprolesdetectedoneachframe,wecanbuildaspatio-temporaldiagramthatdisplaysingreylevelstheinterfacialevolu-tionasafunctionoftimeandspacecoordinates.Fig.4gatherssomeofthepatternsthatwehaveidentied.Thezcoordinatecorrespondstothedirectionofthecylinderaxis.Thediagramcapturesboththetransientregimeandtheasymptoticbehaviour.Typically,theinnercrestsoftheinterfaceprole(closertotheinnercylinder)arecodedindarkgreywhiletheoutercrests 7 FIG.3.ViewofthegapoftheTCcellinthe(r;z)planeilluminatedbyaradiallasersheetfordierentappliedshearrates,(a)fortheOS,(b)ISand(c)O-OS.Thesnapshotsareextractedfromthelast100sofeachstepshearrate.Theleftandrightsidesofeachpicturecorrespondrespectivelytotheinner(rotating)andoutercylinders.Thehorizontalspatialscaleisgivenbythegapsizewhilethevertical-oneisgivenbythe1mmwhiteline.arecodedinlightgray.Notethatinallcases,thein-terfacialinstabilityisassociatedwiththeexistenceofasecondaryvortex owasillustratedby owvisualisa-tions.Theevolutioninspaceandtimeoftheamplitudeoftheinterfacealongzhasbeenshownpreviouslytobecorrelatedtosecondary ows[24].Whentheinter-faceexhibitsundulations,eachwavelengthoftheinter-facecorrespondstoapairofcounter-rotatingTaylor-likevortices,mainlylocalizedinthehighshearrateband,withinward owsco-localizedwiththeinterfaceinner 8 FIG.4.Spatiotemporalevolutionofthepositionoftheinterfacebetweenbandsinresponsetoquenchesatdierentshearratesalongthe owcurve.Thepositionoftheinterfaceinthegapisgiveningreylevels,withtheorigintakenattheinnermovingwall.Thezaxisrepresentsthespatialcoordinatealongthecylinderaxisandthesizeoftheeldofobservationisgivenattheleft-handsideofeachdiagram.Thehorizontalaxisisthetimefromtheonsetofstart-up ow.Eachspatiotemporaldiagramcorrespondsto590s.NotethatfortheOSat12,15and17s1,thehorizontalblacklineonthetopofthespatiotemporaldiagramsareartefactsduetoabubblestuckinthethermostataroundtheTCcell.(Bottomleft)IntensitydistributionI(z;t)re ectedintheradialdirectionbyanisotropicmika akesseededintheOSforastepshearratefromrestto_ =7s1.crestsandoutward owsco-localizedwiththeinterfaceoutercrests.ThespatiotemporaldiagraminthebottomleftcornerofFig.4givesanillustrationofthepatternonecanobtainbyseedingre ectiveparticlesinthesam-pleandobservingfromtheoutercylinder.Thistech-nique,usedinparticularbyAnderecketal.[54]forthestudyoftheinertialTaylor-CouetteinstabilityandlaterbyLarson,Shaqfeh,Mulleretal.[55,56]forthestudyoftheelasticinstability,issensitivetovariationsoftheradialvelocitycomponent[57].Thesuccessionofdarkandbrightstripesstackedalongtheverticaldirectionin-dicatesrespectivelyradial owand owperpendiculartothedirectionofobservation.Thisqualitative owvisu-alisations,isagoodandquickwaytogaininformationonthethree-dimensionalnatureofthe ow.1.Taylor-likevortex owandturbulentburstsLetusnowturnbacktoadescriptionoftheinterfa-cialdynamics,whichisclearwaytogaininformationaboutsecondary owpatterns.First,wedescribe