3d data.pdf

THREEDIMENSIONALOPTICALDATASTORAGEINPOLYMERICSYSTEMSBy

CHRISTOPHERJ.RYAN

Submittedinpartialfulfillmentoftherequirements

ForthedegreeofDoctorofPhilosophy

DissertationAdviser:Dr.JieShan

DepartmentofPhysics

CASEWESTERNRESERVEUNIVERSITY

May,2012

CASEWESTERNRESERVEUNIVERSITYSCHOOLOFGRADUATESTUDIES

Weherebyapprovethethesis/dissertationof ChristopherJamesRyan

candidatefortheDoctorateofPhilosophydegree Dr.JieShan Dr.KennethSinger Dr.RolfePetschek Dr.LeiZhu January20,2012

1

TableOfContentsListofTables 4ListofFigures 5Acknowledgements 7Abstract 8Chapter1IntroductiontoOpticalDataStorage 9

1.1Motivation 91.2FeaturesofOpticalDataStorage 91.3ABriefHistoryofOpticalDataStorage 121.4NewTechniquesforOpticalDataStorage 151.53DOpticalDataStorage 191.6MultilayeredFilmsasStorageMedia 221.7CoextrudedPolymericFilms 231.8ChapterContent 24

Chapter2TwoPhotonInducedAggregateSwitchingofExcimerFormingDyes 252.1Introduction 252.2Materials 272.3TPAofC18RGDye 302.4Experiment 312.5ResultsandAnalysis 322.6ChapterConclusion 34

Chapter3HighDensityOpticalDataStorageinMultilayerPolymerFilms 36 3.1Introduction 36 3.2SampleFabrication 39

2

3.3FilmProperties 41 3.4OpticalPatterningandReading 42

3.5DeterminationoftheCrosstalk 47 3.6ModelingoftheLayerCrosstalk 48

3.7ComparisontoCrosstalkModel 503.8ChapterConclusion 52

Chapter4TheeffectofMultilayeringontheContrastof3DDataStorageMedia 53 4.1Introduction 53 4.2GeometricRestrictiontotheDataDensity 55 4.3DeterminingtheSignalContrastandBackgroundNoise 57 4.4ComparingMultilayeredFilmstoMonoliths 61

4.5Results 634.6ShotNoiseandDarkCurrent 664.7ChapterConclusion 68

Chapter5ThermalInfluenceonBiexcitonAnnihilationinZincPhthalocyanine 69 5.1Introduction 69 5.2Materials 70 5.3Experiment 72

5.4Results 735.5PhysicalInterpretationoftheTimeDependenceoftheCollisionRate 765.6ThermalDependenceoftheZnPc 80

5.7Conclusion 83AppendixAPowerDependenceofPhotopatterninginC18RGdye 85 A.1Introduction 85

3

A.2ReadingfromSubdiffractionSystems 89 A.3SamplePreparation 90 A.4PhotopatterningattheTPAWavelength 90

A.5PhotopatterningwithLinearAbsorption 94 A.6AppendixConclusion 97

Bibliography 98

4

ListofTablesTable5.1 71

5

ListofFiguresChapter2TwoPhotonInducedAggregateSwitchingofExcimerFormingDyes 25

Figure2.1 27Figure2.2 28Figure2.3 29Figure2.4 29Figure2.5 33Figure2.6 33

Chapter3HighDensityOpticalDataStorageinMultilayerPolymerFilms 36Figure3.1 40Figure3.2 42Figure3.3 42Figure3.4 44Figure3.5 47Figure3.6 48

Chapter4TheeffectofMultilayeringontheContrastof3DDataStorageMedia 53Figure4.1 55Figure4.2 57Figure4.3 60Figure4.4 60Figure4.5 62Figure4.6 64Figure4.7 65

6

Figure4.8 66Figure4.9 67

Chapter5ThermalInfluenceonBiexcitonAnnihilationinZincPhthalocyanine 69Figure5.1 71Figure5.2 74Figure5.3 74Figure5.4 82Figure5.5 83Figure5.6 83

AppendixAPowerDependenceofPhotopatterninginC18RGdye 85FigureA.1 91FigureA.2 92FigureA.3 93FigureA.4 95FigureA.5 96

7

Acknowledgements

Abroadrangeoftechniques,skills,andprinciplesarerequiredtocreateandrefinenewideasasrelatedtothesemultidisciplinaryprojects. Mycontributionstothefieldexistonlyasenabledbymy interactions. Throughoutthecourseoftheseexperimentsandinventions,IcollaboratedwithmanyindividualsfromthevariousdepartmentsatCWRU.Here ispresenteda listofthose individualswhosecontributionswerepalpable:Dr.JieShan,BrentValle,AnujSiani,Dr.CoryChristenson,Dr.JackJohnson,Dr.JosephLott,Dr.KennethSinger,Dr.EricBaer,Dr.AnneHiltner,Dr.DavidSchiraldi,andDr.ChristophWeder.

8

ThreeDimensionalOpticalDataStorageinPolymericSystems

Abstract

by

CHRISTOPHERJ.RYAN

Sincethelate1980sopticaldatastoragehasbeenastapleforthecirculationofdigitalinformation. Through the years the storage capacity of these devices has grown tomatchnewdemandsandapplications. However, fundamentaloptical limitationsexistwhichinhibitthegrowthofthecurrentparadigmofdevices.Thisworkiscomprisedofexperiments and demonstrations related to new optical data storage techniques.Variousresultsarepresentedtoaugmentandoptimizefutureiterationsofsuchdevices.Most notably, a 64 layer disk is fabricated and used to store data. This device isfashioned using a polymer coextrusion technique and stores information at a highdensityon23ofits64fluorescentlayers.Tounderstandthesignificanceofsuchdevices,a simulation is used to quantify the benefits of multilayered storage disks overmonolithicdevices.Noiseisshowntobedrasticallyreducedinmultilayeredstructures,while the signal contrast grows under the influence of confinement effects. In theprocess of making this device, an aggregrochromic dye was chosen as a candidatematerial.Furtherexperimentscharacterizehowthedyechangesphasesasaresponsetophotopatterning. Aspresented,theseprojectscitespecific issueswithopticaldatastoragetechnologyandofferoptionsforcomplexityandgrowthwithinthefield.

9

Chapter1:IntroductiontoOpticalDataStorage

1.1Motivation

Since the late 1980s, threedimensional (3D) optical data storage has become asignificantareaof interest to the scientificandengineering communities. As formatsprogressedfromLaserdisctoCompactDisctoDigitalVideoDevicetoBlueRayDisc,2Dstoragedeviceshave remaineda standard forcheap, stabledata storage. Witheachnewgenerationofdevicescameinincreaseinoverallstoragedensity.[1]

However, the wave nature of light has imposed fundamental limits to the storagedensity of such optical devices. There are newmethodologies to circumvent thesebottlenecks. Expanding storage into the third dimension has produced devices thatpushdensityofopticalmediaoveraterabyteperdisk[2,3].

1.2FeaturesofOpticalDataStorage

The general operating principles of 2D optical storage formats rely uponmodulatedreflectionpatterns. Mostcommercialdisksstore informationonathinaluminumfilmthatishousedwithinatransparentplasticdisk.Informationisstoredonsuchadeviceduringthefabricationprocessasthealuminumfilmsarestampedwithapattern. The

10

information is laterreadbyreflectingafocused laserbeamatthesurface. Thedisk isspun about its axis,which translates the pattern relative to the laser. The resultantreflection from thedisk ismodulatedwith the information from thepattern,and thereflection is captured by a photodiode. As a result the photodiode produces amodulated electrical signal which conveys the information to the next step in theprocess[3].

These aluminum based disks are the most prevalent kind of writeoncereadmany(WORM)disk.Othermaterialshavebeenusedforvariationsofthisconcept.Cyanine,phthalocyanine, and azobaseddyeshavebeen layered adjacent to the aluminumorevenusedasareplacementforit.Inthiskindofdisk,thereisnostamping,andthediskismanufacturedwithoutapattern.Instead,writingisdonebymodulatingalaserbeamthatisfocusedonthismaterial.Absorptionofthemodulatedlightcausesheating,andas a result there is a spatialmodulation created in the phase of the disk (typicallypolycrystallineoramorphous).Thereisadifferenceintherefractiveindexofthetwophases,sotheresultisthatthediskispatternedwithamodulatedreflectioncoefficient.Thistypeofdiskisreadinthesamewayasitsstampedcounterparts.Thebenefitisthatthediskiswritablepostfabrication[3,4].

For later versions of this device, the disk is also erasable. In such disks, the activematerialistypicallyasemimetalalloysuchasGeSbTe.Thebasicprinciplesarethesame,

11

but by further exploitation of the materials phase behavior, the disks are maderewritable.Whenheatedabovethematerialscrystallizationtemperature(~150C),anamorphousregionbecomespolycrystallineandmorereflective. By takinganyregion,polycrystallineoramorphous,aboveitsmeltingtemperature(~600C),itmeltsandcoolsrapidlytoanamorphoussolid.Duringphotopatterning,thelaserpoweriscontrolledtoutilizetheseproperties. A lowpowermode isusedtowritedataonablankregionofthe disk, and a high power setting is used to erase written regions. The resultantphotopatternislaterreadwithreflectionbasedmethods[5].

Thewavenatureof lighthas imposeda fundamental limittothedatastoragedensity(DSD)ofallopticalstorageformatsthusfar[6].Theradiusofthenarrowestpartofthebeam is called thebeamwaist (0). Diffraction limits theminimum sizeof0baseduponthewavelength()ofthelightbeingfocusedandthenumericalaperture(NA)ofthelensthatisusedtofocusit.Thereexistsasimpleproportionalityrelationbetweenthem(Eq1)

(1)

Thebeamwaistalsodefinestheresolutionofatypicalreadingsystem. Featureswithseparation smaller than 0 cannot be easily discriminated with linear microscopymethods. TheminimumresolutionofthesystemthensetsanupperlimitontheDSD,asdatapackedmoredenselythanthebeamwaistcannotberesolved. Themaximum

12

DSDofa2Dstoragesystemisthusproportionalto.Notethatforagivendisk,thewritingandreadingprocessesdonotnecessarilyusethesameso0maybedifferentforeachprocess.

As the standard format of optical storage has changed, the DSD has increasedsignificantly. Since 0 limits the DSD, each successive format has reduced thewavelengthandincreasedtheNAoftheopticsinvolved.

1.3ABriefHistoryofCommercialOpticalDataStorage

Abriefhistoryofmainstreamopticaldata storage startswith the Laserdisc(LD)[7,8].Originally produced by Pioneer, LDs and their hardwarewere the first commerciallyavailableopticaldatastoragesystems.TheLDitselfwasanaluminumsheetof30cmindiameterwhichwasmountedinaplasticcylinder.Datawasstoredinthedeviceduringfabricationaseach aluminumdiscwas stampedwith apattern. For consumer videostorage, thispattern contained an analog video signal and adigital audio signal. Thepatternisreadwitheithera632HeNelaseror800nmdiode.

The LD device familywas replaced by the Compact Disc (CD) player. CDs occupy asignificantlysmallerformfactorthanLDs.TheCDisathinplasticcylinderwithdiameterof12cmandheightof1.2mm.ThebulkofCDsareWORMdiskswithanaluminumfilm.

13

Laterwritableandrewritabledisksweremanufactured.CDplayersusea780nmdiodelaserandopticswithaNAof0.45. Thefeaturesonthediskare largerthan0.5um ineachdimensionandseparatedby1.6um.Theoverallcapacityis~700MB[9].

Thenextmajorformatof2DopticaldatastoragetobewidelyproducedwastheDigitalVersatileDisc (DVD). TheDVD follows the same size and form as the CD. SimilarlywritableandrewritableformatsareavailableforDVDs.ThemechanicsofaDVDplayerarethesameasaCDplayer;however,ashorterwavelength laserandstrongeropticsareused. ForaDVDsystema650nmdiode laser isusedwitha lensof .65NA. Thefeatureson thedisk are typically 0.32mwide, 0.4m long and 0.12m tall. Thebumps are separated by about 0.74mwhich leads to a device capacity of 4.7GB.Versions of the DVD are now made with 2 layers of storage rather than a singlealuminum layer,asecond layerofsemitransparentreflectivematerial isused. Eitherlayer can be selected by changing the position of the focal point. Intuitively, theadditionofasecond layermultiplies thestoragebya factorof2. DVDdisksarealsosometimesmade2sided;byessentiallytaking2disksandgluingthembacktoback,thecapacityisdoubledyetagain.Howevertoreadtheothersideofsuchadisk,itmustbeturnedoverintheplayer.

Themost recent commercial format foropticaldata storage is theBluRayDisk (BD).Thisdiskfollowsthesameformasthepreviousstandards.Theyareplasticcylindersof

14

the samedimensionsofCDsandDVDs. BluRay technology takes itsname from theultraviolet laserthat isused in it. Byreducingthewavelengthofthebeamto405nmandincreasingtheNAoftheopticsto0.85,theBDoffersasignificant improvement inDSDoveritspredecessors.Thebeamwaistforsuchadeviceisabout0.15um.Assuch,thefeaturesonthediskarenosmallerthan0.15um,andtheoverallcapacityis25GB.There are also dual layer BD that are analogous to the dual layer DVDs describedabove[10].

Asmentionedabove,thetrendofdecreasingthewavelengthofthelaserandincreasingtheNAof the focusingopticshaspractical limitations. Thebestmultielement lensesavailablehaveaNAofnearly1.4ifoilimmersedand0.95ifdry.Thismeansthatthereis lessthana factorof2 in improvement in theDSDof futuredevices from increasingtheNA.Thepracticeofdecreasingthewavelengthofthelaserisalsorestricted.Whilelightexistswithwavelengthsmuchshorterthan405nm,therearenogooddiodelasersourcesofsuchlight.Tocontinuethetrendtowardsmallereithernewhighbandgapsemiconductor materials must be developed, or other low cost methods of lasergenerationmustbefound. Whileusingthecurrentparadigmofdevices,there is littleopportunitytoincreasetheDSD[2,6].

15

1.4NewTechniquesforOpticalDataStorage

New techniques are being developed to circumvent the limitations of the currenttechnology. However, theseapproaches stem from threebasic ideas. Firstly, ifeachfeatureonthediskcouldrepresentawiderangeofnumbers,ratherthanjust0or1,thedensitywouldincreasesignificantly.Secondly,ifthesizeofthefeaturescouldcontinuetoshrink,theDSDwouldincrease.Finally,iftheentirevolumeofthediskcouldbeusedratherthanjustasingleplane,thentheDSDwouldriseaswell.

Regardlessof the techniquesused, thegeneral task is thesame forallopticalstoragedevices.Firstonemustcreatelocalizedchangestotheopticalpropertiesofamaterial.These changesmust bemade so that they can later be detected and resolved. Torealizesuchdevices,propermechanismsforreadingandwritingofdataareimperative.As such, this research requires significant contributions frommultiple fields includingchemistry,physics,andmaterialscience.Becausethisworkissomultidisciplinary,thereare amultitude of unique approaches to such a device. However, the general taskimposesconstraintsandthuscommonalityarisesbetweendesigns.

The information on the current disks is stored as binary. Each feature on the diskrepresentseither a 1or a 0. By changing the features to instead represent awiderrangeofnumbers,theDSDcanbegreatlyenhanced. Forexample, ifeachbumpona

16

current disk could be replaced by feature on the disk had 20 different detectibleconfigurations, then theDSDwould increase10 fold. Data throughputwouldalsobesignificantly increased,aseachnumberwrittenandreadholdsmore informationthanthat of a binary system. Of course, such a change would require a more refineddetection system. Two methods which utilize this idea are multilevel storage andholographicstorage[1113].

InMultilevelStoragesystemstheoverallstructureofthediskanddriveareverysimilartothealuminumstampedopticaldisksthathavebeenastandardforyears.However,thebinaryfeaturesarereplacedbybumpsofmultiplediscretemagnitudes.Onewaytoachievethisistouseanarrayofdifferentheights;eachonewouldreflectthebeamtoauniquepointinspace.Thedetectorfromatypicalopticaldriveisreplacedbyanarrayofdetectors, so thepositionof the reflection is accurately resolved. An alternativewould be to replace each of the bumps with features of varied discreet reflectioncoefficient. Thenthedetectorcould interpretthemagnitudeofthereflectedbeamtorepresentanarrayofnumbers. Clearlythismethod increasesthestoragedensityofagivendiskbyamultipleofthenumberofdifferentheightsthatituses[13].

In a holographic storage system features within the disk are written to a disk byinterfering2beamsoflight.Oneofthesebeamsisencodedwith informationthroughspatiallightmodulation.Thematerialholdstheinformationintheformofadiffraction

17

grating.Later,areadingbeamispassedthroughthisgratingtoreproducethespatiallymodulatedpattern. By imaging thisbeamontoaCCD, thespatialpattern is resolvedandalargedigitalnumber.Thisprocesstypicallyallowsthefeaturestoholdbetween8and 16 bits of information, rather than a single bit. The overall storage density isthereforemultipliedbythisfactor[11,12].

Another approach to denser data storage involves decreasing the size of the datafeaturesevenfurther.Asstatedearlier,improvementstoandNAareneartheirlimits.However,byusingmaterialswithnonlinearresponsessuchasmaterialswiththreshold,thefeaturescouldbewrittenwithsizesmallerthanthebeamwaist.ThistechniquehasbeenwellappliedtophotoresiststogetfeaturesassmallasWhilethereareafewopticalstoragemechanismsthatarethresholdprocesses,noneofthemhavebeenfullyexploitedyet[14,15].

TheODSdensity isalso limitedby the readingprocess.Onewayof resolving featuresbelowthediffraction limit istousenearfieldmicroscopy. Byplacing itsdetectorveryclose to its specimen, a nearfield scanning optical microscope (NSOM) uses thepropertiesofevanescentwavestogetresolution farsmallerthanthediffraction limit.Insuchasystem,theresolvingpower is limitedbythesizeofthedetectorratherthanthewavelengthoflightthatisusedforillumination.NSOMhasbeendemonstratedtodetectandresolvefeatureswith0.02umlateralspacingand0.002umheight.However,

18

thiskindofdetection is limited in its scope. Todetect features theymustbeon thesurfaceofthesample.

Anotherwaytothebeatthediffraction limit isthroughstimulatedemissiondepletionmicroscopy(STED).Thistechniqueisrestrictedtosystemsthatcontainfluorescentdye,so the reflection paradigm of disks would need to change to utilize STED. STEDimprovestheresolutionofamicroscopebyquenchingthefluorescence inpartsofthefocalvolume thatarenotat itscenter. Whenanexcitationpulse isabsorbedby thefluorescentdye,itcreatesanexcitedcarrierpopulationwithinthedye.Thispopulationinversion typically decays exponentially with time on the nanosecond scale as theexcitationsspontaneouslyemitfluorescence. However,asecondpulsecanbeusedtostimulateemission,andbyselectiveshapingofthispulse,thespontaneousfluorescenceisquenchedoutsideofthecentralregion.Thistechniquehasresolvedfeaturesontheorderof6nm. STEDmicroscopy isnot limited in the sameway asNSOM, as STEDtechniquescanpenetratethesurfaceofthedisk[6,16].

Structured illumination microscopy has achieved similar resolutions. The sample isimaged while being illuminated with a spatially periodic light source. Images arerecordedasthesourceistranslatedandrotatedthroughthesample.Aseriesofimagesareproducedthatareconvolutedwiththeknownperiodiclightsource.Byperformingadeconvolutionanimageofthesampleisproducedwithveryfinedetail,andresolutions

19

similartoSTEDhavebeendemonstrated.Imageprocessingprovidesresolutionbetterthan10nm.Becauseitiscomputationallyheavy,suchamethodisnotusefulforopticaldatastorage[17].

1.53DOpticalDataStorage

Since the detectionmethods for subdiffraction data features bringproblemsof theirown, itremainspertinenttoconsiderotheroptions. AsseenwithDVDs,theDSDofadisccanbedoubledbyaddingasecondplaneofdata.Infact,thedensityismultipliedbythenumberoflayersofinformation.Thismethodofusingthedepthofthediskasastorageparameteristhebasisof3Dopticaldatastorage.Ratherthanstoringdataonaplanewithinadisk,thegoalof3Ddatastorageistousetheentirevolumeofthedisk.Clearly,an issueariseshere,aswhen readingdeep intoadisk the information that isstoredonotherplaneswillcontributetothesignal[1].

The principles of confocal microscopy are necessary to resolve data from denselypackedvolumes. Assuch,theonlystoragemechanismswith3Dutilityarethosethatcreate localizedchanges toeither fluorescenceorreflectance. Amultitudeofdevicesexistasexamples foreachof thesedetectionmethods. However, fluorescencehasagreaterpotentialforalargenumberbecausetherefractiveindexmismatchinreflectionbasedsystemsleadstostrongeropticalaberrationswhenreadingandwriting[1].

20

Theproblemof3Ddatastorage,however, ismorecomplexthan justreadingthedisk.Toproperlydemonstratetheprinciple,onemustproduceadiskwithdifferentialopticalproperties that are distributed throughout the volume of the medium. There aregenerally twoapproaches to thisproblem. Inmany cases,data iswrittenduring thefabrication process, and the disk cannot be changed afterward. However formoreflexibility in use, writable disks are fabricated blank and data is later written byphotopatterning[1].

Thecurrent industrystandardWORM (writeonce readmany)disksoperatebasedonreflection. Data is stamped onto disks as they are injection molded. The bestcommercialdisksavailablecanstore30GBperlayerandhave2layers.Usingthesameprocess, Pioneer has fabricated amultilayered disk of 20 layers of datawith bufferlayersinbetween.Suchadevicehasthepotentialtostoreover500GB.Theabilitytofocusontoeach layer isdemonstrated,neither thewritingdatanor readingdatahasbeenshowninsuchadisk[3].

Alldesignswhichdemonstrate the ability to create30ormoreplanesofdatautilizemultiphoton absorption. These disks are produced blank, and data is written postfabrication.Amultiphotonprocessisnecessaryasitlocalizestheabsorptionoflightto

21

a single plane. Most schemes involve 2photon absorption, and a few claim higherordereffects. Inmostof these cases, themultiphoton absorptiondirectly induces achange inopticalproperties. However, therearemanydeviceswhere theabsorptioninsteadinitiatesasequenceofeventsthatresultinsuchachange.

Most of the designs in the scientific literature utilize photochromicmolecules. As ageneral term, amolecule is photochromic if it can be transformed from species tospeciesby the absorptionof aphoton. Indata storagedevices, aphotochromicdyewithasignificanttwophotoncrossectionisdissolvedintosomepolymermatrix.Uponillumination,thedyechangesitsopticalproperties.Insomecases,theabsorptionandfluorescencepropertiesofthematerialarelocallychangedenoughtomakeabasisfordatastorage[1821].

Foreaseofproduction,mostofthedevicesinthisfieldarefabricatedashomogenous,monolithicdisks.Highdatastoragedensities,stablestoragelifetimes,andrewritabilityarealldemonstratedinavarietyofpapersusingsuchdevices.Evenwithsuchsuccess,thereismotivationtomakeinhomogeneous,multilayereddisksinstead[2224].

22

1.6MultilayeredFilmsasStorageMedia

Thoughtherearemany issueswiththeir fabrication,there isstillmuchresearchdoneinvolvingmultilayereddisks.Tocreatethesedevices,mostgroupsspincoatalternatinglayersofphotoactiveand inactivematerials. Thisprocess,whileeasy todo,producesfilmswithpooruniformityandalsodoesnotscalewell to largernumberof layers. Aless popular technique involves fabricating individual layers and later adhering themtogether.Uniformityofthesedisksismuchbetterthanspincoating,butthiscomesatthecostofamorelaborintensivefabricationprocess.Thismethodisalsomuchmoredifficult toscale intodisks involving thirtyormore layers. While issuesexistwith theproductionofmultilayereddisks,researcherscontinuetoworkwiththemastheyboastmanybenefitsoverthemonolithicdisk[2224].

The benefits of multilayer systems are multifaceted. The major arguments formultilayered disks include reduced materials costs, increased contrast, and reducedaberration. Becausetheactivematerial isthemostcostly ingredient inthesedesigns,spatialconfinementoftheactiveregionreducestheamountoftheexpensiveingredientandreducestheoverallcostofthedisk.Furthermorethespatialconfinementincreasesthecontrastofthewrittendataregionswhencecomparedtotheunwrittenareas.Thisin turn augments the signal to noise ratio (SNR) and doubles the potential storagedensity. Finallyspatialconfinementoftheactive layersreducesthephaseaberrationsofawavefront traveling through thedisk. Thisallows for randomaccessibilityduring

23

writing,andfurthermore,thereductionofaberrationleadstosmallerfocalvolumesandanoverallincreasedworkingdepthofthedisk[2224].

Even with the benefits of multilayering, the difficulties of established methods formanufacturing suchdisksmake them anunlikely candidate for real application. Thiscreates opportunity for other technologies and processes to attempt to solve thisproblem.Hereweusethepolymercoextrusiontechniquewithdiebasedmultiplierstocreatemultilayerpolymericfilmstouseas3Dstoragemedia.[25,26]

1.7CoextrudedPolymericFilms

In thepolymercoextrusionprocess, twopolymersareheated tomatchingviscosities.Astheyareextrudedthroughthesamenozzle,theyarespreadintoabilayerfilm.Theratioofthethicknessofone layertotheother isacontrolledbyadjustingtherateatwhich each polymer flows. The overall thickness of the bilayer is also controlled byadjustingtheflowrateofthepolymers.Diesarethenusedtocut,stackandspreadthefilm. As this is done the number of layers ismultipliedwhile the overall thicknessremainsthesame[25,26]. Furtherdescriptionofcoextrusioncanbefound insection3.2aswellasafigureoftheprocessandaresultantroll.

24

Filmsmade in suchamannerareproduced in rolls. These filmsarecharacterizedbytheiroverallthickness,numberoflayers,andratioofbilayerthicknesses.Filmvarianceisusually less than5% from layer to layer. The filmsexistwith2 to4096 layerswithbilayerthicknessassmallas10nm.Thisprocessscaleseasilytomassproductionasitisalreadyusedtomake filmsonthesquaremilescale. By focusingonthismethod,wehope toalleviatedifficultyofmanufacture thathas thus faroutweighed thepracticalbenefitsofusingmultilayeredsystemsasdatastoragemedia.[25,26]

1.8Content

The bulk of thiswork demonstrates the design and characterization of a coextrudedmultilayerpolymerfilmforuseasa3Dopticaldatastoragesystem.TheworkingWORMdisk ispresented inChapter3. Chapter2coversthecharacterizationof thematerialschosenforthisdiskandaphotoinducedaggregrochromiceffect.Chapter4containsasimulation of the contrast to noise ratio ofmultilayered andmonolithic diskswhichquantifiessomeofthebenefitsofmultilayereddisks.Chapter5containstheresultsofamostly unrelated paper on the charge transport properties of Zinc Phthalocyanine(ZnPC).Theappendixcontainsinformationonphotopatterningandfarfieldmicroscopybelowthediffractionlimitaswellasanexperimentdesignedtoyieldthiseffect.

25

Chapter2:TwoPhotonInducedAggregateSwitchingofExcimerFormingDyes

2.1Introduction

Currentcommercialopticaldatastoragetechnologiesuselinearabsorptionprocessestowriteandreaddata.[27,28] Information iswrittenbymaking localizedchangestotheopticalpropertiesofthedisktoproduceaspatiallymodulatedreflectionpattern. Theoveralldatastoragedensityofadiskisdeterminedbythespacingbetweenthewrittenfeatures. Theminimumwidthofthe featuresonadisk is limitedbythediffractionofthelightusedtowriteit.Typicallynewformatsofcommercialstorageemergebyusingshorterwavelengthsandopticswithlargernumericalaperture(NA).However,thereislittleroomtocontinuethistrendwithoutadvancesinopticsandlasermaterials.Newmethodsarerequiredto further increasethestoragedensityofdisks. Themostprevalentapproachfordoingsoinvolveswritingdataintothedepthofthedisk.SomemodelsofDVD andBluRaydiskexemplify thepotentialof this concept, as they arefabricated to have up to four individually addressable storage layers. However,complexity of producing and usingmultilayer systems increaseswith the number oflayers.Tofacilitatedenserstoragestill,othersturntononlinearoptics.Twophotonabsorption(TPA)isthemostcommontoolforenabling3Ddatastorage.[20,29] Sincethisprocessesscalesquadraticallywiththe incident light intensity,optically

26

induced changes are restricted in depth even when focusing light into a bulkmaterial.[30,31]Herewereportanovel,readilymanufacturedODSsystemthatrelieson the opticallyinduced switching of the aggregation state of an excimerforming,fluorescentTPAdyeinapolymermatrix.Welldefinedvoxelswithdimensionsof3x3x6mwerewrittenindividuallybyexposingthematerialtoafocused,modulatedlaserpulsetrain.Thelightpulseshaddurationof10ns,energyof55nJ,andwerecenteredabout675nmtocorrespondtothedyesTPAabsorptionmaximum. Oncewritten, the datawas read by confocal laser scanningmicroscopy.ThreedimensionalODSsystemsbasedonthisapproachpromiseastoragecapacityofuptoseveralTbytesonaDVDsizedisk,which istwoordersofmagnitudehigherthanthatofcurrentcommercialODStechnologies.[32] Themajorityof thesystemsdesigned for3DODSemployphotochemicalprocesses toenable storage. Typical reactions include photoisomerizations,[3335]photoinduceddimerizations,[36, 37] photodecompositions,[38] and photopolymerizations.[39, 40]Fluorescent photochromic systems have attracted particular interest, because thephotophysicalprocessesare fast,efficient,andreversible.[19,41,42] However, ithasbeen challenging to create fluorescent photochromicmaterials,which combine highstability,highfluorescencequantumyield,andlargeTPAcrosssection.

27

Herewedemonstrateanapproachto3DODSmaterialsthatreliesontheswitchingofthe aggregation state of an excimerforming fluorescent dye. The dye has anappreciableTPAcrosssection,and it isblended inan inerthostpolymer. It isshownelsewhere this material changes fluorescence colors as a response to heat [4346]chemicals,[47,48]ormechanicalforces,[4953]Theopticalchangesarisefrominducedchangesoftheaggregationstateofthedyemolecules.WesurmisedthatsuchchangesareelicitedinsmallvolumesbyTPAinducedlocalheating.2.2MaterialsWeexploredameltprocessedblendofpoly(ethyleneterephthalateglycol)(PETG)and 1.1% w/w of 1,4bis(cyano4octadecyloxystyryl)2,5dimethoxybenzene (C18RG, Figure2.1) as TPAaddressable ODSmedium.[43,45]C18RGwasselectedon accountof its significant changesin absorption and emission spectraupon aggregation/dissociation, itshigh thermal and photochemicalstability,and,asdemonstratedhere,its appreciable TPA crosssection.

28

PETGwaschosenasthematrixduetoitsglassynatureandexcellentopticalproperties.Its glass transition temperature (Tg) of 78 C, which defines the write/erasetemperature(vide infra) is sufficiently above ambient temperature and providesexcellent stability of the storage medium as discussed below. The solubility phasediagram and aggregation kinetics of C18RG/PETG blend sand similarmaterials havebeenpreviouslyinvestigated.[43,45]

Figures2.1illustratestheaggregationstateandopticalpropertiesofthe1.1%w/wC18RG/PETG blend films as a function of thermal history. The corresponding normalizedabsorption and fluorescence spectra are shown in Figure 2.3. The dissolutiontemperatureatwhichthedye isthermodynamicallysoluble isca.130Cfor1.1%w/wofC18RGinPETG,Figure2.1;notethatthedissolutiontemperatureisafunctionofthedyecontent.Thermodynamicallyunstable,butkineticallytrappedmolecularmixturesofthedyeandthepolymercanbeproducedbyquenchingathermodynamicallymiscible,meltedmixtureofthetwocomponents(230Cforablendcomprising1.1%w/wdye)to

29

below Tg. In this state, the blend film appears yellow (peak wavelength of the

absorption spectrum maxabs = 447 nm, Figure 2.3) and displays the greenfluorescence(peakwavelengthofthefluorescencespectrummaxfl =508nm,Figure2.2,2.3)thatischaracteristicofamolecularlymixedblend.Subjectingthequenchedblendtotemperatures above Tg but below the dissolution temperature leads to stable and

pronouncedchangesinabsorption(orangeappearance,maxabs =387nm,Figure2.3)andfluorescence (orange, maxfl = 542 nm), due to aggregation of the chromophoremolecules; these changes are retained if the blend is cooled back to ambienttemperature (Figures 2.2,2.3).The original state can be restored by subsequentlyheating the phaseseparated blend to above the dissolution temperature(all eraseexperiments reportedherewere carriedoutat160 C) (Figures2.1,2.2,2.3).Thus, theabove data document that the blend employed here can be used as a rewritable,optically readable storage medium, in which local exposure to welldefinedtemperatures allows one to write and erase information in two dimensions withmicroscopicresolution.

30

In principle, two differentmodes of operation are possible. The first begins with aquenchedblend (inwhich thedyemoleculesaredissolved)intowhichdata iswrittenthroughannealingaboveTgbutbelowthedissolutiontemperature,anderasedthroughheatingabovethedissolutiontemperature.Alternatively,thesecondprocessstartswithaphaseseparatedmaterial intowhichdata iswrittenbyheatingabovethedissolutiontemperatureandiserasedbyannealingaboveTg.ThecyclesillustratedinFigure2.2and2.3 show that either starting point is a viable option and in principle, manywrite/read/erasecyclesarepossible.2.3TPAofC18ThechoicetoutilizeC18RGforthepresentopticaldatastoragesystemswasbasedonthe expectation that this dye, like other cyanosubstituted oligo(phenylenevinylene)s[5357] possesses an appreciable TPA crosssection, so that the above

31

describedwrite/eraseschemescouldbeachievedbyTPAinducedlocalheating.TheTPAcrosssectionofC18RGwasmeasuredusingtheopenapertureZscanmethod.[58,59].Figure2.4 shows theTPA crosssectionsofC18RGasa functionofwavelength in therangeof625to725nm,wherethelinearabsorptionisnegligible(Figure2.3).TheTPAcrosssectionofthedyevariesbetween0and650GMwithamaximumat675nm. AverysimilarbehaviorwasfoundforC1RG(SupportingInformation).Theseexperimentsreflect a significant nonlinear absorption, which is comparable to that of similarmoleculesreportedintheliterature.[5357] 2.4ExperimentForTPAbasedwritingexperiments,1.1%w/wC18RG/PETGblend filmsofathicknessof 150mwere annealed at 90C for 2days to ensure complete aggregationof thechromophores.DatawritingwasaccomplishedbyusingaNd:YAG laser incombinationwith an optical parametric oscillator (OPO),which produced light pulses of a centerwavelengthof675nm,durationof10ns,andenergyof3mJ.Thepulsetopulseenergystability was ~ 20%. These pulses were attenuated and focused onto the storagemediumthroughanoilimmersedobjectivelenswithanumericalapertureof0.85.TheresultingGaussianbeamwasmeasuredtohaveawaistof3mandRayleighrangeof10m.Eachdataspotwaswrittenbyexposingthesamplestoasinglelaserpulse.Thesamplesweremoved in3Dbyacomputercontrolled3axistranslationstage.Ininitialexperiments,theaveragepulseenergywassystematicallyvaried.Foreachtrial,asetof

32

spotswaswrittenwhiletheaveragepulseenergywasheldconstant.Atpulseenergiesabove100nJpermanentlocalizeddamagewasobserved,whilepulseenergiesbelow25nJbroughtaboutnovisibleopticalchanges.Theenergyrangeof5065nJwasfoundtoafford thedesired changes. An averagepulse energyof55nJwasused for thedatawritingexperimentspresentedbelow. Confocal laser scanning microscopy wasused to characterize the voxelswritten into the C18RG/PETG blend films using theabove approach.A continuouswave laseroperating at awavelengthof 400 nmwasused to excite the samples and the fluorescence was recorded in two channelscorresponding to the integrated intensity in the spectral rangesof500525nmand650 800 nm, respectively. These spectralwindows are sensitive to and thuswerechosentomonitortheaggregatedanddispersedstateofthechromophores.2.5ResultsandAnalysisConfocalmicroscopyimagesofarepresentativesampleshoweightwrittendataspotsof10mbelowthesurfaceofthefilminFigures2.5and2.6.Thevariationsinthesedataspotswere caused by the pulsetopulse energy variation of the output of theOPO.Figure5a shows the raw intensity imageof the sampleafter theapplicationofa lowpass filter in a plane parallel to the film surface (XY plane) for fluorescence in thespectralrangeof650800nm.Orangeexcimeremissionofaggregateddyemoleculesis observed across the entire sample, except for thewritten spots,which appear asdarkerareas,indicativeofdispersionofthedyeaggregatesduetotheTPAinducedlocal

33

heating. The result is further confirmed by the corresponding fluorescence intensityimagerecordedforthespectralrangeof500525nm(Figure2.5).Theimageshowsacomplementarybehavior, i.e.thewrittenspotsappearbright,reflectingan increaseofthe green emission in these areas. In addition,we did not observe any appreciablephotoinduceddegradationofthefilmsduringwriting.Inpreviousstudies,theratiooftheemissionintensitiesintheaboveshortandlongwavelengthwindowshasfoundtorepresentagoodmeasureof theaggregate states.[43,45,50,53]Thus,weused theratio of the intensity of the images shown in Figure 2.5, with their respectivebackgrounds subtracted, to generate a composite image, which indeed shows asignificantly improved contrast. The image contrast can be further enhanced byapplyingalowpassfilter.Thechoiceofthelowpassfilterremovestheeffectofthedyeaggregatesfromtheimage.Furtherprocessingcouldeasilyconvertthesignaltobinary.

34

AsisevidentfromFigure2.5,thespotsizeswrittenwiththesetupemployedherehaveadiameterof~3 mintheXYplane. Figure2.6showstheemissionintensityprofileofthetop4spotsfromFigure2.5casrecordedintheZXplane.Thedimensionsofthedataspots in theZXdirection (~6m)areabit larger than those in theXYplane (~3m),whicharemostlylimitedbythesizeofthebeamfocus.Notethatbecausethewritingprocess requires a threshold temperature, it is possible to achieve a data spotsubstantially smaller than thewritingbeam size, the resultofwhichwillbe reportedelsewhere.[60] The current results reveal clearly that the TPAbasedwriting processallowsonetowritevoxelsthataremicroscopicallylocalizedinall3Ds.

35

Finallywecommentonthestabilityofthisnewstoragemedium.BelowthepolymersTg,themolecularmobilityofthesystem isnegligibleandthemorphologyofthedye/hostsystemsisstableforyears.Inthisregard,theTgofthechosenhostpolymerdictatesthestableoperatingtemperatureregimeofthesystem.Thisindicatesthatwrittenfeaturesremain intactatambientconditions,but leavesopenthepossibilityoferasingdatabysupplyingsufficient thermalenergy toheat thesampleaboveTgand reaggregate thechromophores,ashasbeenillustratedinFigure2.2.2.6ConclusionIn summary,we have demonstrated a newODS system that relies on the opticallyinitiated, thermally induced switchingof theaggregation stateofanexcimerforming,fluorescent TPA dye in a polymermatrix. Such blends can easily, inexpensively andrapidlybefabricated in largequantitiesusingsimplemeltprocessingtechniques.Welldefinedvoxelswithdimensionsof~3x3x6 mhavebeenwrittenthroughthe exposureof the blend to single laser pulses. The voxel size is comparable to the focal pointvolumeof thewriting laser,suggesting that thermal transportdoesnotplacea lowerlimittothevoxelsize,at leastnotatthe lengthscalesusedhere.Adiffraction limitedlaserbeamcanbeachievedbyoptimizingtheopticalsetup.InthatcasethevolumeforefficientTPAislimitedtoaspaceslightlysmallerthan0.4 mineachdimensiongiventhewavelengthandnumericalapertureused.SuchaschemecanbeusedtopotentiallywriteseveralterabytesofbinarydatainadiskofsizeofthecommonCDsandDVDs.

36

Chapter3:HighDensityOpticalDataStorageinCoextrudedMultilayerPolymerFilms3.1IntroductionNewtransformativeconceptsforopticaldatastorageapplicationsareneededtomeetthe future requirements of applications inmultimedia, archiving, security, andmanyothers. Commercial data storage technologies are moving to threedimensionalmaterials, but the known concepts suffer from limited addressability and highfabrication costs.We demonstrate here that storage systems based on coextrudedmultilayer films can overcome these problems and allow for terabyte level bitbybitopticaldatastorage.Stratifiedfilmswith64storageand64bufferlayers,withaperiodof3.4m,were fabricatedbya readily scalablecontinuous rolltorollprocessat200metersperhour. Information in the formofcomplexpatternsand individualbitswasrecordedinupto23superimposedlayersbyphotobleachingafluorescentdyeviaonephoton absorption. The optical resolution and crosstalkwere examined. The resultsdemonstratethatthefabricationprocess,whichisfarsimplerthancurrentapproaches,allowsonetofabricateBluraycompatible,highdensitymultilayerstoragemediawithstoragecapacitiesthatareordersofmagnitudehigherthanthestateoftheart.Highcapacityopticaldatastorage (ODS) isrequiredforrobustarchiving,securitytags,and even new media formats for threedimensional (3D) displays, and many otherapplications[61].ConventionaltwodimensionalODShasadvancedtoallow25GB/layerstorage in bluray (BR) discs, which is sufficient for high definition video storage.

37

However,thestoragecapacity is limitedbydiffractionofthewritingbeam,thecostofmanylayered media, as well as the number of layers that can be fabricated andaddressed [62]. Shorter wavelengths or higher numerical aperture (NA) optics offersome improvements,butsubstantialadvancescanonlybeachievedbyutilizingmultidimensional methods including spectral and polarization multiplexing [63, 64],holographicrecording[65],andinparticular,theefficientuseoftheaxialdimensiontoovercome the limits of surface storage [20]. The capacity of 3D storage media hascurrentlyprogressedtowardterabyte(TB)levels[2].

Localizingdata in a3D storagemedium isoften achievedby activating theparticularmaterial responseusing twophotonabsorption (TPA)byhighpower sources, suchasnearinfraredpulsedlasers.Theuseofthisnonlinearopticalprocessgreatlyreducestheoptical changes outside the region of interest [2, 18, 6670]. In these schemes, dataplanesarewritteneitherinmonolithicmaterials[2,19,71],orinmaterialswithdiscreteactive layers. The former are easy to fabricate, while the latter allow furtherconfinementofthedatawithinalayer.Thisreducescrosstalkduringwriting/reading,aswellas theamountand costof the recordingmaterial.However, themultilayer (ML)discsreportedinpreviouseffortswerefabricatedthrougheithersequentialspincoating[15,21]or lamination [3,24,72],whichare labor intensiveand cannoteconomicallyscaletolargenumbersoflayers.

Whileeasy, lowcostmultilayer fabrication isone important roadblock,other system

38

level issues are also impeding the transformation to commercial TB ODS. Thistransformation will be best addressed by a evolving the system from the presentcommercial state of the art. In particular, it would be desirable to develop a TBread/write system using existing BR laser diode technology and that conforms toexisting requirements. Optical aberrations in the read/write system limit theaddressabledepthsothatthecurrentBRspecificationofatotalthicknessofthestoragemedium to 140m significantly limits the number of layers possiblewith a reflectivestorageschemeasareflectiveschemerequireslargelayerseparationtoavoidcoherentreflectioneffects. Thus,TBmultilayerstoragerequiresamediumandstorageschemethatsimultaneouslyconfinesthedatatotightaxialdimensionsandallowsreadingandwritingwithminimumcrosstalk.

Wereporthereonanapproachthatsuccessfullyaddressesalloftheseissuesandpavestheway for futurehighperformance, lowcost,easilyscalableandmanufacturableTBODSmedia. First,wehavedevelopedanovel,robust,andsimpleapproach fordigital3DODS inMLpolymer films thatwere fabricated ina continuous,melt coextrusion,rolltoroll process. This is truly lowcost and easily scalable both in film area andnumberof layers. Inaddition,permanentstorage isdemonstrated in23superimposeddatalayersutilizingfluorescence(FL)quenchingofannoveloligo(pphenylenevinylene)dyeorganicdyeupononephotonabsorption. Thiswas achievedwith a submilliwattcontinuouswave(CW)405nmdiodelaser,enablingoperationwithcurrentcompactBRsources. Finally, a high axial data density (3 m/layer), low crosstalk scheme was

39

realizedusingaFLreadingscheme,which,incombinationwithwritingatthediffractionlimit of the BR laser, promises TB storage capacitywithin commercial disc thicknessspecifications.

3.2SampleFabricationThe coextrusion technique [73, 74] used tomanufacture these films is illustrated inFigure 3.1a. In this process, which has already been successfully applied to thefabricationofphotoniccrystals[75],lasers[76],andgradientrefractiveindexlenses[77],two thermoplastic polymers (A and B) are heated to form a melt with matchingviscosities, and then coextruded into a bilayer feedblock. The AB bilayer is sentsequentially througha seriesofmultiplicationdies.Eachdie cuts, spreads, stacks themeltanddoublesthenumberof layers.Filmswithoverfourthousandlayersand layerthickness as low as 10 nm have been produced using this technique [74, 78]. Thelaboratoryprocessemployedinthepresentstudyallowsfabricatingfilmsupto36cminwidthataspeedofapproximately200m/hr,thoughmuchhigherspeedsandwidthsarepossibleincommercialproductionlinesemployedforothercommercialapplications.

40

The chromophore C18RGwas synthesized as previously described [45]. PETG Eastar6763andspectroscopicgradetoluenewereobtainedfromEastmanChemicalCompanyandBurdick&Jacksonandwereusedasreceived.AblendofC18RGandPETG(nominaldyecontent2wt.%)waspreparedusingaHaakeRheocord9000batchmixerat230Cfor 5minutes. To coextrude the PETG solutionwith the PVDF, bothmaterialswereheatedto230Cwherethepolymershavematchingviscosities.Thebilayerproducedbythecoextruderwassentsequentiallythrough5dies.Eachdiecutperpendiculartothebilayers,spread,andstacked the film tomultiply thenumberof layersby2.The finalfilmproducedwasasystemof64layerswithanoverallthicknessofapproximately200m.Coextrusionalsopermits themanufactureofperiodic filmswithmore than twodistincttypesoflayers(i.e.,ABCorABCB)toaccommodatemoresophisticateddesigns[74].Usingthistechnique,wefabricatedastoragesystemconsistingof64dataand64buffer

41

layers,whichservetoconfinethebitswithindiscretelayers.AphotographoftherollofthefilmproducedinthisstudyisshowninFigure4.1b.DatastoragelayerAiscomposedofatransparenthostpolymer,poly(ethyleneterephthalateglycol)(PETG)thatisdopedwith2.0wt.%of the fluorescent chromophore1,4bis(cyano4octadecyloxystyryl)2,5dimethoxybenzene (C18RG) (13). Buffer layer B consisted of poly(vinylidenefluoride) (PVDF), is optically inactive and refractive indexmatched to layer A. Thismaterial isparticularlyuseful to limitdiffusionof thedyeduringprocessing [79].TheaveragethicknessesoflayersAandBare0.3and3.1m,respectively.Theproductionprocessandwriting/readingsystemhasmuchbroaderapplicabilitythantheparticularmaterialreportedhere.3.3FilmpropertiesC18RG is a cyanosubstituted oligo(pphenylene vinylene) dye with aggregochromicproperties [45]. The structure is shown in Figure 3.2. Previous studies onthermodynamically immiscible blends of this dye and various host polymers havedemonstratedsignificantchangesoffluorescence(FL)propertiesuponexposuretolight,heat, chemicals, ormechanical forces,which are attributed to excimer formation orbreakup[45].Theusefulnessofthisparticulardyeforopticaldatastoragehasalreadybeen demonstrated by twophoton switching of the excimermonomer transition inmonolithic films [70]. We prepared the dyedoped polymer layers in a molecularlymixed blend following the procedure described in [70]. The absorption coefficient of

42

layerAis0.1m1at405nm,whereasboththePETGmatrixandthePVDFbufferlayeraretransparentinthevisiblespectrum.

Figure3.3showstheFLspectrumofasingleactivelayerunder405nmexcitation,takenwithaspectrometerandCCD fibercoupled to theconfocalmicroscope.Anarea2.5x2.5 mwas scanned in about 100ms at 0.01mW/m2. Themonomer and excimerfluoresceat410and445nm,respectively.UponexposuretoCW lightofhighfluence,bleachingoftheFL isobservedwithnoshiftofthepeak, indicatingthat thedyedoesnotaggregatetoformexcimersundertheseconditions.3.4OpticalPatterningandReadingThedatawritingwasperformedbyFLquenchingofthestoragemediumC18RGupononephotonabsorptionofaCW laserbeamat405nm focusedontothechosen layer.ThisbecomespossiblebecauseoftheaxialconfinementofthedataintheMLfilms.Thereading, on the other hand,was done by FL detection as opposed to reflection, anapproachcommonlyusedforsingleandfewlayerstoragemedia.Asweshowbelow,FL

43

detection significantly increases the axial layer packing density and thus the storagecapacity. Ifmolecularlydispersed inPETG,C18RGdisplaysabsorptionandFL spectrawithmaxima at 445 nm and 510 nm, respectively. An intensity on the order of 0.1mW/m2orgreaterisrequiredtoobtainmeasurablequenchingwithsubmsexposures.Thechangeswereobservedtobepermanentandstableoverthetimeperiodofmorethan2years.Figure3.4depictsFLimageswrittenintothetop23storagelayersofthe64layerfilmsdescribedabove.ThewrittenregionscorrespondtoareasofreducedFLintensity(black).Towrite thedata, theoutputofaCW405nmdiode laserwas focused into the filmthroughanOlympusMPlanApochromat,100x,1.4NAoilimmersedobjective.Patternswere recorded using anOlympus FV1000 confocalmicroscope by scanning the laserbeamalongacustomizedpathatarateof75nm/ms.Writingwasperformedlayerbylayerfromthelowesttothetopmoststoragelayer.Theincidentpowerwasabout130W and the intensity was varied between 1.5 mW/m2 (topmost) to 2.0mW/m2(lowest layer).Thesameconfocalmicroscopeand lasersourcesubsequentlycollected3DFLimagesofthesampleatareducedfluenceandincreasedscanrate(0.01mW/m2at5m/ms).

44

TheuseoftheFLdetectionschemesallowssmallerlayerspacingscomparetoschemesrelying on phase changes and reflection, as discussed. Another limiting factor is theresponsefunctionofthereadingsystemitself.Theconfocalmicroscopeusedhere,witha1.4NAobjective, isanextremecase.Withtheseoptics,the intensityatthedetectorplanedropsbyhalfifthesampleismovedbyabout0.1maxiallyoutofthefocalplane(foran infinitelysmallaperture),which ismuchsmaller than the layerspacing [45]. Ifinstead the 0.85 NA objective found in BR players is used, even with an aperturediameterafactorof10 largerthanthespotsizeatthedetector,thisfigure isstillonly0.89m.Thus,whilethe factors limitingtheminimum layerspacingarerelaxedhere,theoptical limitofthereadingsystem isnotyetan issue.Thereare likelyother issues

45

thatresultsinaneedforaminimumspacing,suchasthicknessvariationandsolubilityofthedye.FromtheimagesshowninFigure3.4,itisevidentthatdatacanreadilyberecordedandretrieved fromeachof the individual storage layers. The average reduction in the FLintensitythroughoutthefilmforthewrittenareasisabout22%.Theimagesshowthatthequalityoftheretrievedimagesdecreasesforthedeeperlayersduetoaberrations.However,wedemonstrated that it is readilypossible to retrieve information from23layers,which is the largest number of recorded layers that has been reported in aheterogeneousMLODSmedium.Wenotethattheaberrationsdependontheworkingdistance of the objective, and that the quality of storage in the deep layers, andtherewith the number for layers from which information can be retrieved, can befurtherincreaseduptotheBRspecificationbyoptimizingthelenssystem.Stateoftheart,twoorfourlayerBRdiscshaveanaxialspacingofgreaterthan10minorder to limit thecoherentcrosstalk thatoccursdue tomultiple reflectionsof thereadingbeamatthereflective layerandspacer layer interfaces [80].TheFLdetectionschemeemployedheregreatlyreducesthemultiplereflectionsaswellasemittingatanondegeneratewavelength,allowingmuch smaller spacings tobeused compared tophasechangematerials.Thus, the spacingofour layers (3m) isoneof the smallestexplored[21].

46

ThearealdensityofODS isconstrainedby thebeamwaistat thediffraction limit.Toexamine the data bit dimension of our ML films, single lines were written into amonolithicfilmoftheactivelayerunderthesamewritingconditionsasdescribedabove.The resulting bleaching profile is shown in Figure 3.5. A fit yields a fullwidthhalfmaximum (FWHM) of 380 nm, which is approximately the minimum bit spacingachievable in this current system, and is consistentwith thediffractionlimitedbeamsize. This places the areal density of ourML films close to that of BR systems, theminimumbit spacingofwhich is320x150nm,owing to the thresholdnatureof thephasechangewritingprocess,allowingsubdiffractionlimitwriting.

47

3.5DeterminationoftheCrosstalkA significant factor that determines theminimum bit spacing in both the axial andlateraldimension isthecrosstalk.OneattractivefeatureofMLfilms inthecontextof3Dstorageistheconfinementofthebitsintheaxialdirection,whichreducescrosstalkbetweenneighboringbitsandlayersduringwritingandreading.Todirectlymeasurethecrosstalk, an array of bits was written into 10 successive layers and the contrastmodulation in themiddle (probe) layerwas readas informationwaswritten in the

48

others. Similarwriting conditions as described abovewere employed. The laserwasmodulatedwithasquarewavegenerator toproduceonoffbitpairsseparatedby1.0minbothlateraldirections,andthetotalareawritten(40x40m)waslargerthanthebeam diameter in any given layer, so as not to underestimate the total crosstalkbetweenanytwo layers.Thisalso leadstoresultsthatarenotdependentonwhichofthe10 layers is chosen as theprobe.A subsectionof the FLpatternandmodulationafter selectwriting stepsare shown inFigs.3.6aandb.ThemaineffectofcrosstalkappearstobeanoverallreductionintheaverageFLlevel.

3.6ModelingoftheLayerCrosstalkCrosstalk isan issueduringbothwritingandreading in3Dstoragesystems. Here,thecrosstalkduringwritingisexaminedforthecasesoflinearandtwophotonabsorption.The relevantparameter,physically, is the ratioof the intensity receivedatagivenbitlocationduringexplicitwritingofthatbitrelativetothatobtainedduringwritingofall

49

otherbitsinallotherlayers.ThesimulatedbitarrayconsistsofNzlayerswithaspacingofz,eachconsistingofNybyNxbits,withspacingsofyandx,respectively.ThebitarrayoccupiesanvolumeofsizeLxbyLybyLz.Theorigin isplacedatthecenterofthedataarray.AssumingadiffractionlimitedGaussianbeam,thereductionintheFLasinglebitlocated at the origin during explicit writing of that bit (the signal, S) should beproportionaltosomepowerofthefluence.

2

20

zpN z

eS Cw

(3.1)

whereC isaproportionalityconstant, istheabsorptioncoefficient, w0 isthebeamwaist, and p is chosen to be either 1 or 2 to simulate either a linear or quadraticbleachingresponse.TheFLreductionofthissamebitduringwritingofalltheotherbits(thenoise,N)isgivenbythesum

2 2

2 2/2 /2 2( ) 2( )/2

/2 /22

2

2

/

y xz

z y x

z

k k

N zi x j yp p

pN NN

k N j N i

w w

k NN C e ee

wS

(3.2)

andthe1/e2beamradius,wk,atthezoriginwhenwritinglayerkisgivenby

20 20

1/k

k zwn w

w (3.3)

wherenistherefractiveindex,andisthewritingwavelength.Sissubtractedfromthistoaccount for the single term in the sumwhich isdefinedas the signal.This canbegreatlysimplifiedassumingahighlyfocusedbeamanda largescanarea.However it ismoreaccuratetosimplyperformthesummationnumerically(Matlab).Theparameterswerechosentocorrespondtothoseusedintheexperiment.Thebitspacingwaschosen

50

as1.0minbothlateraldimensionswithallbitsbeingon(numericallyequivalenttotheonoffpatternof0.5mspacingproducedbythesquarewavegenerator), z =3m, Nx=Ny=40, Nz=10, Lx=Ly=40m, Lz=27m, and w0=0.32m. A beam waistcorresponding to the experimentally observed value of 0.32 m is used. The resultplotted inFigure4 is the ratioS/N.Scorresponds to themodulationsignal,while thetotalNresults inoverallconstantbleaching,so thisratiocanbedetermined from the

experimentaldatabycalculating ,wheremaxistheaverageofthepeakvalues

inthemodulationandministheaverageofthetroughs.This calculation is intended only as an orderofmagnitude comparison, as there aremanyotherphysicalprocessesthatmustbetaken intowhendesigninganoptimalMLstructure (29), such asmultiple reflections.One of the primary differences betweenexperimentandtheoryhere isthefactthatthebeam isscannedcontinuouslyandnotdiscretely.Furthermore,forlargeintensitiesthebleachingwillbecomesublinear,whichisnotaccountedforinthetheory.Thelightscatteredattheinterfacesandtheinabilitytocontrolallaspectsoftheconfocalwritingsystemonsmallscales(suchastheretraceandsamplepositioning)alsocontributetotheCBR.3.7ComparisontoCrosstalkModelThe ratioof signalmodulation to thebackgrounddepletion FL (carriertobackgroundratio,orCBR) isused toquantify the crosstalk.TheCBRafterwritingeachof the10layers (startingwith the probe layer) is plotted in Figure 3.6c (triangles). The value

max min1max

51

decreases from2 to0.15with increasingnumberof layers,and is ingoodagreementwithnumerical simulations.While this isnot insignificant, thisCBR ratio ismore thansufficienttoresolve individualbit information,asshown inFigure3.4.DuetothehighNAofthewritingobjectiveandtheinertbufferlayers,thefluenceinthelayeradjacenttoone that isbeingwritten, is reducedbymore thana factorof10.Two theoreticalcurvesfortheCBRarealsoplotted,oneassumingthebleachingisrelatedtothelinearpowerofthefluenceandoneassumingaquadraticdependence (e.g. inTPAscheme).The theory, which is consistent with the experimental results, indicates nonlinearfluenceresponsesyieldsignificantenhancementsintheCBR.Thesimilarityofthebleachedspotsizeandcrosstalkmeasurementstothetheoreticalresultsbasedonthebeamparameterssuggestthatthebleachingprocessmaydependinanearly linearfashiononthefluence;however,asthefilmreportedherehasbeenstoredundermercurycontainingfluorescent lampswithweakblueviolet linesfortwoyearswithout ameasurable decrease in the FL, onemay speculate that a thresholdexists, belowwhich no bleaching occurs.Onephoton absorptionwith a nonlinear orthreshold fluenceresponse is thepreferredmethod forcommercialization, incontrastto the nonlinear optical processes such as TPA which require complex pulsed lasersystemsor veryhighpowerCWdiode lasers, and the longerwavelength required towrite increases the spot size. Other materials such as Au nanorods [15] or organicphotopolymers[81]havealreadyshownpotentialforaonephoton initiatednonlinear

52

orthresholdresponse.Workiscurrentlyongoingtoincorporatesuchmaterialsintothecoextrusionprocess.Inthefuture,wewillincorporatetheseonephoton,nonlinearfluenceresponsewritingschemesintothemediumdescribedheresothatwecanexpecta40layer1TBcapacity,0.8TB/cm3densitydiscinthestandardarealformat.Theresulting136mthickfilmiswithin the 140 m bluray specification compatible with the optics of compact,commercialwritingsystems,whichaccount foraberrationand tilt tolerances[82].ThispresentsafeasibleapproachtofabricatingaTBlevelstoragemediuminalargescaleatlowcostwithinthematerialspecificationsofcurrentwritingsystems.3.8ConclusionInsummation,wehave shown thatcoextrudedML filmsare feasible foruseasa3DODSmedium. Withthecommercialsystemspresentlyavailable,theabilitytoincreasethedensity is limitednotonlyby theopticsbut alsoby the costneeded to add andmanufactureadditional layers.Coextrusion removes this serious constraint.Wehavewrittenfilmscontaining23active layerswith independent images,the largestnumberof layers of any stratified storage medium. The crosstalk between layers is alsoexamined,andwhilenotnegligible,issmallenoughtopermitdemonstrationofthisfilmasastoragedevice.Thelateralbitspacingislimitedbythediffractionof405nmlaser,andtheaxialbitspacingallowsTBlevelrecordingwithinthethicknessspecificationofcurrentdiscplayers.

53

Chapter4:TheEffectofMultilayeringontheContrastandNoiseof3DStorageMedia

4.1Introduction

As new 3D optical data storage technologies are developed, a variety of new diskschematicsarecreated.Typicallythesedisksfitintooneoftwostyles.Monolithicdisksare made from homogenous materials and as a result are easier to fabricate thanmultilayereddisks.Ontheotherhandtheoperationaladvantagesofmultilayereddisksareoftendemonstratedbythegroupswhomakethem[15,21,24].Somemultilayereddevices have been shown to offer data storage densities that are not attainable bymonolithicdisks [21]. Themostcommonclaim is thathighercontrast tonoise ratios(CNR)areachievedthroughmultilayering[24,83]. However,theprecisenatureofthebenefitsisnotwellunderstood.

Previous works show that the contrast to noise ratio(CNR) of a layer within amultilayered disk changes very littlewhen the adjacent layers arewritten[24]. Thisresult isexpectedformultiplereasons. Inamultilayeredsystemmore lightshouldbedeliveredtothetarget layerbecause there is lessparasiticabsorption fromthe layersabove.Furthermore,thereshouldbelessbackgroundandnoiseproducedfromoutoffocuslayerswhenreadingamultilayereddisk.

54

It is the focus of thiswork tomodel and compare both the signal contrast and thecontrast of the background noise from out of focus regions in bothmultilayer andmonolithic fluorescent disks. In doing so, the benefits of reduced crosstalk andparasitic absorption are quantified and can act as a guide for the design of futurestorage media. The effect is examined for various intraplane spacings and variousdegreesof confinement. Furthermore, theenhancementsareexamined in the caseswhenShotnoiseandphotodiodedarkcurrentarethedominantnoisesources.

Thenatureofthesignalcontrastandbackgrounddependheavilyuponthewritingandreadingmechanismsof thedisk. Formost3Ddatastoragesystems, the fluorescencepropertiesofamaterialaremodifiedby interactionwithawritingbeam. Typicallyasample isnotfluorescentataparticularwavelength,andtheapplicationofthewritingbeamcreatesfluorescence.However,otherdisksexistwherefluorescenceisquenchedby the application of thewriting beam. Often the change is created by a nonlinearabsorption event, however some materials exist which linear absorption leads tononlinearmaterials responses. In such systems, themechanism for change is oftenthermalorphotochemical. Often thesemechanismscreate thresholdbasedchanges,andproducebinaryspatialmodulations.Thisisthetypeofdevicethatisconsideredinthissimulation.

55

Athreestepapproachistakentoevaluatethesignalcontrastandnoisecontrastofsuchdisks.Firstthesignalgeneratedbyasinglevoxeliscalculatedforafluorescentreadingprocess. Next, the total background noise from fluorescent regions surrounding thisvoxel iscalculateddeterminethemaximumpossiblecontrastofthebackgroundnoise.Finally, the interplane data spacing (D=a+b) and the ratio of the active layer to thepassive layer(Ra=a/D)arevaried tocompare thesignalandnoisecontrasts forvariousdatadensitiesandsystems[figure4.1].

4.2GeometricRestrictiontotheDataDensity

In this simulation, signal contrast is compared to the noise contrast as the maindeterminant of the effect of multilayering. However, geometry suggests a rough

56

estimateofthelimittothedensityofthedisksbasedsimplyuponthesizeofthewrittenspots.Thelengthscalesofinterestarethethicknessofanactivelayer(a),thethicknessofapassive layer(b),andthezradiusofthewrittenspots(z0). Theparametersofthewritingbeamdeterminez0.Whenwriting,theintendedlayershouldbeatthecenterofthe beam while adjacent layers should be outside of z0. Under this constraint 2 2.Twolimitingcasesemergeforthissystem.Inthemonolithiclimit,b>0,so 2. Inthis limit,theplanesofdataarespacednocloserthanthetwicez0. Intheveryconfinedmultilayeredstructurelimit,a>0so .Intheselimits,itisseenthatthemultilayeredstructureallowstheplanestobespacedabouttwiceascloselyasthemonolithiclimit[figure4.2].Thisestimateisbasedsolelyontheconstraintthattheadjacent layers lie outside of the written spots potential diameter. By calculatingcontrastofthesignalandnoise,theproblemisaddressedinamorerigorousmanner.

57

4.3DeterminingtheSignalContrastandBackgroundNoise

The contrast and noise values are calculated as sums over all fluorescentcoordinates[equ4.1].ThesearecombinedtocalculatethecontrastCNR.Thecenterofthebitofinterestisdefinedattheoriginandthepositionoftheconfocalpinholeisfixedaccordingly. M(r,,z) is the spatialdistributionof fluorescent regionswithin thedisk.I(r,z)representstheintensityprofileofthereadingbeam,whichiscenteredattheorigin.Theconfocaltransferfunction,T(z),determineshowmuchoftheemittedlightfromanypointiscapturedbythedetector.Otherimportantquantitiesarethequantumyield(Y)

58

andabsorbance()ofthedye,andthesensitivity(r)ofthedetector.Sothenthecurrentproducedinthedetector(J)canbefoundbyevaluatinganintegral:

, , , (4.1)

TheterminbracketsdescribesthedistributionoftheabsorbedlightandtheapplicationofYT(z)determineshowmuchisreemittedintothedetector.Theintegralisdoneoverall space to accumulate the contribution from every point. Application of thephotodioderesponsivity,rp,convertsthecapturedphotonstoelectriccurrent.Theformforeachof these functionswillbedescribed lateras the signalcontrast(JS)andnoisecurrent(JN)arecalculated. WhencalculatingtheCNR=JS/JN,thematerialspropertiesand detector sensitivity cancel out, and the result is the ratio of the integral of thespatial properties of the system[equation 4.2]. However, the absorbance of thematerialstillaffectstheshapeandamplitudeofI(r,z).

,,, ,,,

(4.2)

WhereMS andMN are the formsofM for regionsproducing signal andbackground,respectively.Forthesimulationofthemicroscopesillumination,thereadingbeamwasset to awavelength() of 400nm and a numerical aperture(NA) of 0.85. I(r,z)wasassumed to be the shape of a Gaussian beam with decaying intensity in absorbingregions[equ4.3].

59

,

(4.3)

HereL(z) is theeffectivepath length through thematerialwithabsorptioncoefficientatlocationz.Thisisusedtoapproximatetheattenuationofthelightasitpenetratesthesample.z0istheRayleighrangeofthebeamandw0isthebeamwaist.Thesetwoparameters are directly related to NA and by equation 4.4. The absorbance andquantumyieldofthedyewaschosentomatchthedyeusedforourdatastoragebasedexperimentsintheotherpapers[60,84]

(4.4)

GeometricopticsdeterminethetransferfunctionT(z). Theamountof lightfromeachpointsourcethatmakes ittothedetector isdeterminedbycomparingthesizeoftheimageofeachpointsourceattheconfocalpinholetothesizeoftheconfocalpinhole.Interference effects in the point spread function were ignored, and the resultingapproximationofT(z)wasaLorentzianinz,centeredonthefocalplane[figure4.3].ToexaminetheeffectofthepinholeonCNR,simulationwasdoneoncewiththepinholematchedtothesizeofAirydiskofthefluorescedlightandlaterwiththepinhole5timeslargerthantheAiryDisk.

60

To find Jsand JN ,M(r,z)wasdesigned foramaterialwitha thresholdresponse totheillumination.[15, 68] As such, the functions are populated with values of 1 incoordinates that arewritten and 0 for the unwritten coordinates. The CNRwill beidenticalforsystemswherewrittenregionsarerepresentedbyquenchedfluorescence.Nearthebeamfocus,equipotentialsurfacesofI(r,z)areroughlyellipsoidal,sothespotswere assumed to be ellipsoidal with radius w0 and z0 in the r and z directions,respectively. For this disk, only seven layers of data are simulated. The noisecontributionsfromadditionallayersdecreaseastheinversesquareofthedistancefromthe focusbecause T(z)has as a Lorentzian form. Contributionsbeyond the first fewlayersareverysmall,andthesumconvergesrapidly[figure4.3].

TosimulateJs,onlythecentervoxelispermittedtofluoresce[figure4.4A].Thebackgroundnoise,JN,isdeterminedbyexaminingthevarianceofthebackground.To

61

calculatethemaximumpossiblebackground(Jmax),allsurroundingvoxelsareconsideredtobewritten[figure4.4B].Inareal,writtendiskeachvoxelhasa50%chanceofbeingwritten.TheaveragebackgroundisthenJmax/2,ashalfofthesurroundingvoxelsareexpectedtobewritten.Thevariationofthisbackgroundisthemainsourceofnoise.Becausethebitsareeitheronoroff,theycanbedescribedbyabinomialdistribution.I(r,z)andT(z)aresharplypeaked,sothevoxelsthatcontributethemosttothebackgroundanditsvarianceareadjacenttothebitofinterest.Thisisseeninfigure4.3asthecentralpeaksaremuchlargerthantheothers.Mostofthebackgroundcomesfromthesetwobits,sothebackgroundisapproximatedbyabinomialdistributionwith2trials.ThevarianceinthebackgroundrelatedtothemaximumbackgroundasJN=Jmax/2

3/2.Insystemswithlargercontributionstothenoisecomefromotherlayers,JNwillstillbeproportionaltoJmax.However,theproportionalityconstantwillbesmaller.

62

4.4Comparingmultilayeredfilmstomonoliths

Thisprocessisrepeatedfordiskswithvaryingzaxisseparationbetweenvoxelsandalsovaryinglevelsofconfinementinthezaxisthroughmultilayering.TheratioRawasvariedfrom 100% to 1% in increments of 0.1%. This provided CNR values for monolithicsystemswasaswellasmultilayeredsystemsofvariedconfinement. Ineachcase,themultilayereffectwasaccomplishedbyapplyingaspatial filter in the formofasquarewavetothecorrespondingM.VaryingvaluesofRaweresimulatedbyvaryingthedutycycleofthesesquarewaves.[figure4.3].DatadensitywasvariedbysimplydecreasingD.Dwasvariedfrom2z0toz0sothensimulationwouldspanthemostcommondesignsfromexperimentsintheliterature.

63

4.5Results

Tobetterunderstandwhathappensastheparametersarechanged,onecanexaminethe signal and crosstalk noise individually [figure 4.4A]. In the figure, the simulatedcontrast and background noise are normalized to the simulated contrast andbackgroundnoise(respectively)ofamonolith in thegeometric limiting caseofD=2z0.WhileholdingDconstant,afewtrendsareapparentinfigure4.6.ForlargevaluesofRa,JS increasesasRadecreases. This isbecausemoreofthereadingbeam isdeliveredtothefocalregionastheotherregionscauselessparasiticabsorption.Thatis,forsmallerRa,thelighthasashorterpathlengththroughthedyeinlayersabovethebitofinterest.HoweveratsmallRa,JSreachesamaxandbeginstodecreasewithdecreasingRa.Thishappens when the bit of interest becomes smaller than the confocal region. JN,however,continuouslydecreasesastheRa isdecreased. Thenoisedecreasesbecausethereislessfluorescentmaterialoutsideofthefocalregiontocontributetothenoise.As the pinhole size is increased, the behavior of JN remains unchanged. JS behavessimilarlyforboth largeandsmallpinholes. However,forthe largepinholethepeak isshiftedtolargerRa.Thisisadirectconsequenceofthewideningoftheconfocalregion.

64

In theabsenceof theanyothernoise sources, theCNRcontinuously increasesas theactive layersaremadethinner[figure4.7]. There isamonotonic increase intheCNRwithdecreasingvaluesofRa.ItisalsonotablethattheCNRdecreasesmonotonicallyasDisdecreased.Evenwithmultilayering,morelightfromadjacentlayersisdeliveredtothedetectorsasthespacingisdecreased.Thebehaviorissimilar inboththe largeandsmallpinholecases.However,theenhancementisstrongerwhenthepinholeissmaller.In the figure, theCNRvaluesarenormalized to theCNR foramonolithicdevice in itsclosely packed geometric limit. Thesemonolithic CNR values are 625 for the smallpinholeand127forthelargepinhole.

65

SofromtheaboveitisseenthatmultilayeringcanproduceapalpableenhancementtotheCNR.Thisenhancementcaninsteadbespentonincreaseofdatastoragedensity.Amonolithbecomesunreadableasthedataspacingismadecloserthantheradiusofthedataspots.However, therearemultilayeredsampleswithreadableCNRvaluesat thesamedatadensity.ForeachRathereisacorrespondingseparationthathasaCNRthatmatchesthemonolithicdevicewithnospotoverlap.Assuch,themultilayeringprocesscan increasethedatastoragedensitywhile leavingtheCNRconstant. Thisproperty isexemplifiedinfigure4.6.Thedensityiscappedbecausetheseparationhasalowerlimitof approximatelyD/z0=1. Evenwithmultilayering,when theplane spacing is smallerthan thewritten spot diameter, the information becomes imprintedon the adjacentlayers duringwriting. This adds noise to the layer at the focus of the beam, and itcannotbefilteredbytheconfocalsystem.

66

4.6ShotNoiseandDarkCurrent

Thus far, theonlynoise termdescribedhasbeen thecrosstalk. This result isan ideallimitratherthananexpectation.Othersignificantsourcesofnoiseforthesedeviceswillcome from the detector. The photodiode dark current and the Shot current bothcontributetotheoverallCNR.TheShotnoisewillvarydirectlyasthesquarerootofthegeneratedphotocurrent.Thephotodiodedarkcurrentnoisevaluesareconstantanddonotscalewiththeamountoflighttakenin.AslongasJNislargewhencomparedtothedarkcurrentandShotNoise,figure4.7shouldpredicttheeffectofaxialconfinementinmultilayered disks. However, when the JN is similar in size to the other terms,multilayeringproducesmuchlessenhancement[figure4.9]

67

In figure4.9whenothernoisesourcesaredominant, theCNRenhancementsseen infigure4.7areallbuteliminated.Thisisbecausemostoftheenhancementinfigure4.7istheresultofreducingJN.Whilereductionoftheactivelayerhaslittleeffectinthesecases, increasingthepowerofthereadingbeamorthesensitivityofthedetectorwillboosttheCNRofsystems.However,theCNRofamonolithscalesinthesamemannerastheCNRofamultilayereddiskasthebeampowerisincreased,sothereremainsnoenhancement frommultilayering. Although, JN scales linearlywith beam power anddetectorsensitivitywhileDCisconstantandShotnoisescalesasthesquareroot.Fromthis, it ispossibletoretaintheenhancementsofmultilayeringby increasingthesignal

68

andnoisecurrents so that theyare largewhencompared to theShotnoiseanddarkcurrent.

4.7Conclusion

OverallaclearimprovementinCNRisseenastheconfinementeffectsareincreasedinthese simulations. When the crosstalk noise is the dominant term, themultilayereddisksoffera considerableenhancement to theCNRofadisk. Increasing the readingbeampowerdoesnotoffer anybenefit in this casebecause the signal and crosstalknoise both scale linearly with power. When the pinhole size is increased, theenhancement is not as strong. However, increasing the pinhole size increases thephotocurrentssignificantly.WhendarkcurrentandShotnoisearethedominantterms,there is no enhancement to the CNR from multilayering. However increasing thegeneratedphotocurrentcanmakethebackgroundnoisethedominantterm.

69

Chapter5:ThermalInfluenceonBiexcitonAnnihilationinZincPhthalocyanineFilms

5.1Introduction

Metallophthallocyanine(MPC)dyesarenotedfortheirnonlinearopticalpropertiesandelectronic structure. Common features include large values of chi3, a columnarcrystalline form, liquid crystallinemesophases, and absorbance bands that span thevisiblespectrum.

While inorganic semiconductorshave filled the role in thepast, themassproductioncapabilityoforganicmaterialsplacesthemasstrongcandidatesforactivemediainthenextgenerationofoptoelectronicdevices. Otherworkspresent thesedyesasactivemedia for femtosecond Kerr gates, photovoltaics, optical data storage, and opticallimiters.[8588]

For each application, carrier transport and lifetime are of paramount importance.Various authors have measured ultrafast exciton dynamics in MPC samples.[8991]Many reportbiexcitonannihilationathighexcitationdensities. In theiranalyses, theexcitonexcitoncrossectionsareextracted fromdynamicsmeasurementsandarethenusedtocalculatetheintermolecularhoppingtimes.Inallcases,thecrossectionisfoundtovaryliketime1/2,whiletheintermolecularhoppingtimesrangefrom10to400fs.

70

These calculated hopping times are derived from the interpretations of thedimensionality of the biexciton interaction. The phenomenon of 1D diffusion ofexcitonsalongMPCchainsisoftenpresentedasanexplanationoftheresult;howeveraquasistatic population of excitons interacting in 3D can produce the same timedependence. Themainpurposeofthiswork istoproperlydiscriminatethesemodels.Indoingso,Iintendtoexplaintheexcitonicbehaviorinbothcrystallinephaseandthefirstmesophase.

Thefemtoseconddynamics forthin filmsofzincphthalocyanine (ZnPC)arepresented.This work characterizes the temperature dependence of exciton behavior in bothcrystallineand liquidcrystallinephasesoftheZnPC. Themeasuredexcitonpopulationdynamicsdonotscale linearlywith the initialpopulationdensity,and thisbehavior isaccuratelydescribedbyabiexcitonrecombinationmodel.Thebiexcitonrecombinationcrossectionand intermolecularhopping timearepresentedat temperatures from90400K.Heretheexcitonhoppingtime isreportedtovarysignificantlywithtemperatureinbothphases. This thermaldependence is indisagreementwith themodel thatthebiexciton annihilation arises from 3D interaction of static carriers. The dependencestronglysuggeststhatthe interaction isrestrictedtoanexcitonpopulationundergoing1D diffusion. From this, the role of temperature and structural order in the excitonhoppingtimeisclear.

71

5.2Materials

ZnPCpowderwaspurchased fromAldrichandpurifiedvia thin layerchromatography.The resultant powderwas placed onto a thin sheet of quartz. Then itwas heatedbeyonditsmeltingpointandpressedagainstanothersheetofquartz.Thethicknessofthesamplewassetbyaonemicronspacer. Thefilmwasthencharacterizedopticallywith linearspectrummeasurements.Thephasetransitiontemperatureofapowderofthissamplewasthenmeasuredtobenear375Kwithdifferentialscanningcalorimetry.

TheabsorbancespectrumofthefilmwasfitusingaLorentzoscillatormodel.[9298]Afit of seven states was chosen as there are seven distinguishable features in theabsorptionspectrum(figure5.1). Theextractedresonant transitionenergiesand theirbroadeningfactorsarelistedintable5.1.

72

Linear spectrometer measurements confirm a band structure similar to MPC filmsstudiedinotherpapers[9298].Thelowestenergypeakisunderstoodtobethefirstpi>pi*moleculartransition.[98]Thesymmetryoftheabsorptionandphotoluminescencespectral lines confirms that the next highest energy peak is electronically the same.However,thisstateincludesanopticalphononmode.[98,99]Bothofthesestatesarecommonly present in ZnPC solutions and films.[92, 96, 98, 99] Photoconductivitymeasurements of ZnPC films have confirmed this state to function as a Frenkelexciton.[97,98]

73

5.3Experiment

Ultrafastpumpprobe spectroscopywasused to study theexcitondynamicsof theseZnPC films. An amplified titanium sapphire laser provided illumination for timedependantmeasurements. Byuseofathermalstage,thetemperatureofthesamplewascontrolledandmonitored.

Carrierlifetimemeasurementsweremadewithexcitationfluencefrom110mJ/cm2attemperatures from 80450 K. A fundamental output beam of 800 nm (1.55 eV)wavelengthwasused as aprobewith a frequencydoubled400nm (3.1eV)pumpingpulse to produce dynamic absorption measurements. Pump and probe beamsintersectedata30oangleandhadradiiof200umand50um,respectively. Datawasmeasuredoverarangeof180pswitha260fsresolution.

Fromthesamplesabsorption,heatcapacity,andthermalconductivity,itwasestimatedthateachpulseheatsthesampleby0.3K.Thecharacteristictimeforthermaldiffusionacrossthe laserspotof200umwasestimatedtobe3ms. Thereforethepumpbeaminducedheatingwasexpectedtohavenegligibleeffectsonthepsdynamics. Thebulktemperature of the sample was expected to be within 3 K of the reading of thethermocouple.

74

5.4Results

Toquantifytheexcitondynamics,thequantumefficiencyoftheconversionofabsorbedphotons to excitons was assumed to be near 100%. Observed photoluminescencemeasurementsbyBalaetal suggest that thisefficiencyvaries littleas temperature isvaried.[98] From the pulse energy and sample absorbance it was clear howmanyphotonswere absorbed, and thiswas taken to be equal to the number of excitonscreated. Thetimedependantphotoabsorbancemeasurementswerecalibratedtothisstandard.

At293K the initialexcitonpopulation is linearwithpower. However, thepopulationdynamics do not scale linearlywith initial concentration (figure 5.2). This indicatesexcitonexcitonannihilationispresent.[8991]

Similar features are observed for all temperatures from 90K400K. There is a fastbiexciton annihilation over 510 ps. A slower decay from single exciton behavior is

75

noticedover100ps. Finallyamuchslowerbackgroundtermremainsnearlyconstantoverananosecond,andthislikelyfromselftrappedpi*states.[8991]

Becausethepopulationdecaydoesnotscalelinearlywithpopulation,abiexcitondecaymodel(equation5.1)isusedtofitthedata.[8991]

12)( nntn (5.1)

Here n is the exciton concentration, is the single exciton lifetime, and )(t is thebiexcitoncrossection.

Thetimedependenceof )(t isbaseduponthespatialconstraintsofthesystem[91].In

each case, )(t takes the form of ptt 0)( where p is some real number and 0 is aconstantthatdependsuponthepropertiesofthesystem.Forhighpopulationdensities

( 1)( nt ) the linear term in equation 5.1 is negligible. This short timeapproximationisseeninequation5.2.

101101 )1( ntpn p (5.2)

This illustratesawayto fit thetimedependenceof0 throughsimplegraphmethods.Givencorrectselectionofp,aplotof1/nvs.tp+1shouldbeastraightline.Fittingn1toapower lawshowedthatp= 1/2. Aplotofn1vst1/2 forallofthemeasureddatasetsproducesaseriesoflinesasseeninfigure5.3.

76

So with the result of p = 1/2, the form of )(t is determined to be 2/10)( tt .Substitution intoequation5.1allows thecarrierpopulation tobe solved forall timesandconcentrations.Solvingtheequation5.1yieldsequation5.3.

terfn

enp

t

10

10

(5.3)

Thisbecamea tool toextract thedynamicscoefficientsunder theconditionp= 1/2.Note that in equation 5.3 the carrier concentration goes to zerowhen t >> . Thedifferential absorption, however, persists on timescales much larger than excitonlifetimes. As such, thisportionof the signal isneither from singleexcitondecaynorbiexciton annihilation. Therefore, this long term behavior was subtracted beforeapplyinganyofthefittingroutines.[91]

The time dependence of )(texists because the excitonswhich are closest to each

other at t=0 aremost likely to annihilate first. At later times, the closest pairs ofexcitonshavealreadyannihilatedand theaverage time forany remainingexcitons toannihilate is longer. Theformof )(t isdeterminedbytheexcitonexciton interactionstrengthandthetranslationalfreedomoftheexcitons.

77

5.5PhysicalInterpretationoftheTimeDependenceoftheCollisionRate

Intheprevioussectionsitwasshownthatthedifferentialabsorptionsignalwascausedmainly by the presence of excitons, and that these excitons undergo a nonlinearannihilationwitheachother. Thebiexcitonannihilationcoefficient,(t)wasshowntohavethetimedependenceof(t)=0t0.5atpumpingintensitieslower10mJ/cm2.

Thereare two interpretationsof theexcitonickinetics thatpredict this result. Sucharesponse is characteristic of quasistatic excitons with long range interactions in 3dimensions.Thisdependenceisalsoproducedbyapopulationofexcitonswithashortrangeinteractionthatareallowedtodiffusein1dimension.[91]

Tobuildthesemodels, it isassumedthattheexcitonsannihilatethroughdipoledipoleinteraction. The phenomenological description of the annihilation rate for a singleexciton,(t),istheintegratedinteractionbetweenthatexcitonandanyotherexcitoninthesample[91,100].Thisisrepresentedas

),()()( trgrrdt d (5.4)Weredisthedimensionalityofthesystem,(r)isthedipoledipoleinteractionstrengthof the excitons, and g(r,t) is the pair correlation function between two excitations

78

spacedbyadistancerattimet.[91,100]Thedipoleterm,(r)iseasilyunderstoodasaForstertypeinteractionwhere

6

)(

rRkr Aop (5.5)

HereRAistheForsterRadiusofannihilationandkopistheopticalexcitationdecayrate.Thecorrelationfunctioncanbefoundwithadiffusionequation[91,100]

)),((),()(2),(2),( 2 trgFtrgrtrgDttrg

(5.6)

Where D is the diffusion constant for the excitons, and F describes higher orderinteractions.[91,100]Togoodapproximation,

),()),(1)((2)),(( trgtrgtntrgF (5.7)

Onesuchinterpretationisthestaticcase.Inthisvieweachexcitoniswelllocalizedanddoes not move significantly over the course of its lifetime. Since the excitons areconsideredquasistatic, thedivergent term is ignored. If thehigherorder interactionsarealsoneglected,thenthesolutiontoequation5.6issimplyintegrated.[91,100]

tretrg )(2),( (5.8)

Placementofthis function into thedefinitionof(t)providesan integrableexpressionandreducestotheform

160)(d

tt (5.9)

79

Soclearlythisinterpretationreproducesthemeasuredbehaviorwhend=3.Inmakingtheseapproximations,itisimportanttofindtheconstraintsunderwhichtheyholdtrue.Thehigherorder interaction term isnegligible ifF(g(r,t))

80

Awellknown resultof randomwalker calculations is that thenumberof sitesvisitedincreaseswiththeelapsedtimetel.[101]

S~telf (5.13)

Herefisdeterminedbythedimensionalityofthesystem.Forsystemsofintegervalued,fhaspiecewiseform.[102]

21

22dfor

dfordf (5.14)

So,fromequations5.125.14

ftSn ~~ 1 (5.15)

For short times and high concentrations, the linear decay term in equation 5.4 isnegligible. Substitution of equation 5.18 allows the time dependence of (t) to beshown

)(~)( 212 ttttnn ff (5.16)

Inordertoproducethesametimebehavioronbothsidesoftheequation,itisclearthat(t)~tf1.Sousingthepiecewisedefinitionoffnowdefines(t)[91],

22)(

0

12

0

dfordfortt

d

(5.17)

The experimental time dependence of (t)~t0.5 shown in the previous section isexpectedforaonedimensionalsystem(d=1)inthisinterpretation.

81

Sowe see that there are two interpretationsof the exciton kinetics thatpredict theobservedresult. Suchbehavior ischaracteristicofquasistaticexcitonswith longrangeinteractions in 3dimensions. [91] A population of excitons with a short rangeinteractionalsoproducesthisresultwhenconstrainedto1dimensionaldiffusion.[91]To differentiate between these cases, it is necessary to consider the temperature

dependenceof 0 .

5.6ThermalDependanceoftheZnPc

For thenext setofmeasurements the temperaturedependenceof the carrierdecayparameterswasexamined.Thesamplewasheldattemperaturesfrom90K415Kwhilethepumpprobedynamicsweremeasuredatbothahighanda low fluence. The lowfluencemeasurements resulted in sufficiently lowexcitationdensity so thatbiexcitonannihilationwasnotobserved.Thesefitswereusedtoaccuratelydetermineateachtemperature. The high powermeasurementswere then fit to equation 5.3with

constrained toextract 0 at the same temperature. The resultingvaluesof 0 and wereplottedinfigure5.4.

82

Herea trend is clear in thebehaviorof 0 .It is seen to risewith temperature in thecrystallinephaseand thendecreaseswith temperatureas thedisorderof the systemincreases.Asseen infigure5.4, dropsby~20%asthetemperaturegoesfrom365Kto370K.Thisshowsthatthereisasignificantlylargernumberofwaysforeachexcitontoscatterandannihilateastheliquidcrystalgoesthroughthephasetransition.

For a quasistatic system of interacting excitons, the temperature should have littleeffect on the rate of interaction. Themain contribution from a rising temperaturewouldbethermalexpansion.Astheaverageintermolecularspacingincreases,therateofreactionwoulddecrease.Thispredictiondoesnotmatchtheseresults.Theobservedrateofreaction increaseswithtemperature inthecrystallinephase. FurthermoretheintermolecularspacingofZnPCchangesverylittleasitisheatedthroughthecrystallinephase.Thusheatingshouldhavelittleornoeffectontherateofbiexcitondecayifthe

83

systemwereinsuchaquasistaticconfiguration.Assuch,weknowtheexcitonsdonotadheretothequasistatic3Dinterpretation.[89,91]

In this interpretation, the intermolecular hopping time, h is intrinsically linked to 0

andthemoleculardensity,N.Assuch, h wasdeterminedforeachtemperatureunder

thecondition that 20-14 = Nh [91]. Since thedensityof the filmvariesvery littlewithtemperature,thehoppingtimewasdetermineddirectlyfrommeasurementsof 0 andplottedinfigure5.5.

It should also be noted that the intermolecular hopping time and D, the diffusion

constantareinverselyrelated. That isD~ h 1~ 0 2.[91,101,102]FromthisrelationthediffusionconstantDisdetermineduptoaproportionalityconstant(figure5.6).

Totestthisinterpretation,welooktotheArrheriusequation.Thisrelationisoftenusedtomodelthermallyactivatedprocesses. Formostsystemsthatundergodiffusion,the

84

temperaturedependenceof some rate constant,R, iswelldescribed as an activatedprocessofactivationenergyEaandcharacterizedbytheArrheniusequation,

TkE

B

a

eRR

0 (5.18)

whereTistheabsolutetemperatureofthesystem,kBistheBoltzmannconstant,andR0isthatrateatT=0K.Inthiscase,thediffusionconstant,D,willbefittotheArrheniusequation(figure5.6). Afitforanactivatedprocessdoesnotworkwellovertheentirerangeofdata. However, if split into2 segments that correspond to thephases, thismodel fits 0well. Themodel showsbiexciton recombination tohaveanactivationenergyof 217 meVinthissystem.

5.7Conclusion

So it is clear that the temperature affects both linear exciton decay and biexcitoncollision rates in ZnPC. From this, the experiment determined that the excitonrecombinationdoesnotoccurunder theconstraintsofaquasistaticsystemwith longrangeinteractions.Instead,thesampleisinteractionisfoundtohappenatshortrangeforexcitonstravellingin1D.Itisalsofoundthatwithincreasingtemperaturetherateofbiexciton collision increases. This change has corresponding effects on theintermolecularhopping time anddiffusion constant. Furthermore, the complexityof

0 s temperature dependence shows that different phenomena dominate theannihilationrateineachphase.

85

AppendixA:PowerDependenceofPhotopatterninginC18RGdye.

A.1Introduction

Theabilitytopatternhighdensity3Dopticalstoragedisks is limitedbytheshapeandsizeof thewrittenspots. Understandingbeamseffecton thesample isan importantpartofoptimizinganysuchdisk. Hereaseriesofspotsarepatterned intoapolymercontaining a two photon absorbing dye. In separate experiments, the sample ispatterned with wavelengths corresponding to its linear and two photon absorptionpeaks. Byvarying thepowerof thepatterningbeams, the sizeof these spots isalsovaried. As thepowerof thepatterningbeam isdecreased, thespotsbecomesmallerandinsom