We study the kinetics of the isotropic-smectic phase transition in a colloidal rod/polymer mixture by visualizing individual smectic layers. First, we show that the bulk isotropic-smectic phase transition is preceded by a surface freezing transition in whi

thesetwomorphologiesissmallandwillexaminetheirrelativestabilityelsewhere.

Realspaceimagesenablesustostudytheki-neticpathwayfortheformationofcolloidalmem-branes.Theycaneitherhomogeneouslynucleatefromthemetastableisotropicsuspensionorcanhet-erogeneouslynucleateatthesurface-inducedsmecticphase(Fig.4aandb).Acolloidalmembranenucle-atedattheinterfacegrowsintotheisotropicphaseeitherasatwistedribbonora atdisk.Overape-riodofafewdaystwistedribbonscanreachalengthsofseveralhundredsmicrons.Fluorescenceimagesindicatethattherearenorodsintheisotropicsolu-tion.Thereforecolloidalmembranes(ribbons)mustelongateduetorodsthatdi usefromametastablenematicphasethroughasurfacesmectictoamorestablecolloidalmembrane.Thefactthatthereisatransportofrodsacrosstheinterfaceshowsthatthecolloidalmembranesarestructureswithlowerfreeenergythanthenematicphaseorbulksmecticphase.Atlowerdegreesofsupercoolingwemostlyobserveheterogenoussurfaceinducednucleationin-steadofhomogeneousnucleationofcolloidalmem-branes.Thisshowsthatatwo-stepkineticpathwayhasalowernucleationbarrierfortheformationofcolloidalmembranes.Tosummarize,thephasesep-arationinregion3ofthephasediagramproceedsintwosteps.Inthe rststepontimescaleofsecondstominutesweobservetheformationofnematictac-toidswithsurfacesmecticphaseidenticaltothoseobservedinregion2.However,thesetactoidsaremetastable.Inthesecondslowsteponatimescaleofhourstomonthsweobservethenucleationofcol-loidalmembranesatthesurfacefrozensmecticphaseandtheirsubsequentgrowthintotheisotropicphase.Afewcommentsareinorderregardingthestruc-turesobservedinregion3.First,toourknowledgethisisthe rsttimethatnon-amhiphilicobjectswithverysimpleexcludedvolumeinteractionshavebeenselfassembledinto2Dmembrane-like(Fig.4c)and1Dpolymerlikestructures(Fig.2f)[26].Wespec-ulatethatthesestructuresarestabilizedbyprotru-sionlike uctuations[27].Second,itseemsplau-siblethatisolatedcolloidalmembranesobservedinregion3arehighlyswollenlamellarphasesprevi-ouslyobservedinmixturesofnematicfdandhardspheres[4].Theswellingofthelamellarphaseis

predictedtheoretically,buthasyettobeobservedinexperiments[28].Third,astheosmoticpres-sureisincreasedthereisatransitiontoregion4inwhichsmallcolloidalmembranesirreversiblystackupontopofeachothertoformelongated laments(Fig.2g).Thenatureofthetransitionfromiso-latedmembranestoasmecticphaseremainsun-explored.Fourth,uallythereversee ectisob-servedpathwayswhereastablenucleusiswettedbyametastablephase[5,29]

Inconclusion,therearetwoimportantresultsthatcanbededucedfromourexperiments.The rstsur-prisingresultisthatarod/polymermixtureexhibitssurfacefreezinginwhichaquasi2DsmecticphasewetstheI-Ninterface.Thise ectoccursatrodcon-centrationof235mg/mlwhilebulkI-Sphasetran-sitionoccursat255mg/ml.Toourknowledgethisisthe rsttimethatthesurfacefreezinghasbeendirectlyvisualizedinasystemwhosephasebehaviorisdominatedbyentropicrepulsiveinteractions.Thesecondresultofthisworkistodemonstratethere-lationshipbetweenthesurfacefreezingandthebulkisotropic-smecticphasetransition.Acomplextwostepkineticpathwayforthenucleationofthesmec-ticphaseoutoftheisotropicsolutionhasbeeniden-ti ed.Inthe rststepametastablenematicdropletwithasurfacefrozensmecticphasenucleatesintheisotropicsolution.Inthenextstepisolatedmono-layers(colloidalmembranes)ofsmecticphasenucle-ateatthesurfacesmecticphaseandsubsequentlygrowintotheisotropicphase.Duetothesimplicityandgeneralityoftheexcludedvolumeinteractionswhichdominatethephasebehavioroffd/Dextranmixture,theresultspresentedhereshouldberele-vanttoamuchwiderclassofsystemsthanthosestudiedhere.

IwishtothankSethFraden,GerhardGompper,ArjunYodh,TomLubensky,DanielChen,PeterLangandPavlikLettingaforusefuldiscussions.IamparticularyindebtedtoJanDhontforhishos-pitalityatFZ-JuelichandtheAlexandervonHum-boldtfoundationfor nancialsupport.PartofthisworkwasdoneatBrandeisUniversitywherethisresearchwassupportedbytheNSF-DMRgranttoSethFraden.

[1]D.Frenkel,H.Lekkerkerker,andA.Stroobants,Na-ture332,822(1988).

[2]G.J.VroegeandH.N.W.Lekkerkerker,Repts.on

Prog.Phys.8,1241(1992).

[3]P.N.PuseyandW.vanMegen,Nature320,340

(1986).

[4]M.Adams,Z.Dogic,S.L.Keller,andS.Fraden,

Nature393,349(1998).[5]P.R.tenWoldeandD.Frenkel,Science277,1975

(1997).

[6]J.Zhu,M.Li,R.Rogers,W.Meyer,R.H.Ottewill,

W.B.Russel,andP.M.Chaikin,Nature387,883(1997).

[7]U.Gasser,E.R.Weeks,A.Scho eld,P.N.Pusey,

andD.A.Weitz,Science292,258(2001).

[8]V.J.AndersonandH.N.W.Lekkerkerker,Nature