- Permanent Link:
- A Predictive Model for the Antifouling Efficacy of Engineered Microtopographies
- Decker, Joseph T
- Place of Publication:
- [Gainesville, Fla.]
- University of Florida
- Publication Date:
- Physical Description:
- 1 online resource (234 p.)
- Doctorate ( Ph.D.)
- Degree Grantor:
- University of Florida
- Degree Disciplines:
- Materials Science and Engineering
- Committee Chair:
- BRENNAN,ANTHONY B
- Committee Co-Chair:
- BATICH,CHRISTOPHER D
- Committee Members:
- PHILLPOT,SIMON R
- Graduation Date:
- Subjects / Keywords:
- Adhesion ( jstor )
Antifouling ( jstor )
Biofouling ( jstor )
Eris ( jstor )
Fouling ( jstor )
Modeling ( jstor )
Seas ( jstor )
Topography ( jstor )
Ulva ( jstor )
Zoospores ( jstor )
Materials Science and Engineering -- Dissertations, Academic -- UF
antifouling -- combinatorial -- microtopographies -- modeling
- bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Materials Science and Engineering thesis, Ph.D.
- We have developed a model, called the Surface Energetics Attachment (SEA) model, that relates the work of adhesion for an organism to the probability for attachment. The model was used to examine the predicted attachment density of four organisms to a variety of topographies from data available in the literature. The results showed the model was capable of predicting relative attachment density to a high degree of accuracy (R2 = 0.83). Additionally, local effects of topographic configuration were identified and predicted by the model.
An additional tool, the radial distribution function, was applied to zoospores of Ulva linza attached to topographies. This analysis was combined with previously developed mapping techniques to help identify local topographic configuration effects on Ulva attachment that may by missed the SEA model analysis. The radial distribution function showed differences in Rmax and screening distance dependent on the location of the spore on the topography. Additionally, the contributory effects of a single spore extended only 20 micrometers from the reference which indicated an ideal size for the topography.
Topographies were characterized for their adhesion properties through Atomic Force Microscopy (AFM) measurements, contact angle measurements and cell adhesion experiments. The AFM measurements showed site dependence for the work of adhesion for a colloidal probe on a wetted topography as predicted by the SEA model. The contact angle measurements configuration dependence for static, advancing and receding contact angle measurements for DI water. The cell adhesion experiments showed lower adhesion for Madin Darby Canine Kidney cells on topographies regardless of configuration.
Bioassays with Ulva linza and Balanus amphitrite showed go agreement with the SEA model predictions. A high throughput assay was developed to test the Ulva attachment and was successfully used to discriminate differences between 32 simultaneously tested patterns. A series of topographies ranging from 5 micron to 200 micron in size were evaluated for the barnacle cyprid. These were found to inhibit attachment in line with the model predictions. Additionally, tracking experiments showed disruption of the cyprid surface probing dependent on the size of the topography. ( en )
- General Note:
- In the series University of Florida Digital Collections.
- General Note:
- Includes vita.
- Includes bibliographical references.
- Source of Description:
- Description based on online resource; title from PDF title page.
- Source of Description:
- This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
- Thesis (Ph.D.)--University of Florida, 2014.
- Adviser: BRENNAN,ANTHONY B.
- Co-adviser: BATICH,CHRISTOPHER D.
- Electronic Access:
- RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-08-31
- Statement of Responsibility:
- by Joseph T Decker.
- Source Institution:
- Rights Management:
- Applicable rights reserved.
- Embargo Date:
- LD1780 2014 ( lcc )
This item has the following downloads:
LETTERSLarge-scalepatterngrowthofgraphenefilmsfor stretchabletransparentelectrodesKeunSooKim1,3,4,YueZhao7,HoukJang2,SangYoonLee5,JongMinKim5,KwangS.Kim6,Jong-HyunAhn2,3, PhilipKim3,7,Jae-YoungChoi5&ByungHeeHong1,3,4Problemsassociatedwithlarge-scalepatterngrowthofgraphene constituteoneofthemainobstaclestousingthismaterialindevice applications1.Recently,macroscopic-scalegraphenefilmswere preparedbytwo-dimensionalassemblyofgraphenesheetschemicallyderivedfromgraphitecrystalsandgrapheneoxides2,3. However,thesheetresistanceofthesefilmswasfoundtobemuch largerthantheoreticallyexpectedvalues.Herewereportthedirect synthesisoflarge-scalegraphenefilmsusingchemicalvapour depositiononthinnickellayers,andpresenttwodifferentmethods ofpatterningthefilmsandtransferringthemtoarbitrarysubstrates.Thetransferredgraphenefilmsshowverylowsheetresistanceof 280 V persquare,with 80percentopticaltransparency. Atlowtemperatures,themonolayerstransferredtosilicondioxide substratesshowelectronmobilitygreaterthan3,700cm2V2 1s2 1andexhibitthehalf-integerquantumHalleffect4,5,implyingthat thequalityofgraphenegrownbychemicalvapourdepositionisas highasmechanicallycleavedgraphene6.Employingtheoutstanding mechanicalpropertiesofgraphene7,wealsodemonstratethemacroscopicuseofthesehighlyconductingandtransparentelectrodes inflexible,stretchable,foldableelectronics8,9. Graphenehasbeenattractingmuchattentionowingtoitsfascinatingphysicalpropertiessuchasquantumelectronictransport4,5,a tunablebandgap10,extremelyhighmobility11,highelasticity7and electromechanicalmodulation12.Sincethediscoveryofthefirstisolatedgraphenepreparedbymechanicalexfoliationofgraphitecrystals6,manychemicalapproachestosynthesizelarge-scalegraphene havebeendeveloped,includingepitaxialgrowthonsiliconcarbide (refs13,14)andruthenium(ref.15)aswellastwo-dimensional assemblyofreducedgrapheneoxides3,16andexfoliatedgraphene sheets2.Epitaxialgrowthprovideshigh-qualitymultilayergraphene samplesinteractingstronglywiththeirsubstrates,butelectrically isolatedmono-orbilayergraphenefordeviceapplicationshasnot beenmade.Ontheotherhand,theself-assemblyofsolublegraphene sheetsdemonstratesthepossibilityoflow-costsynthesisandthe fabricationoflarge-scaletransparentfilms.However,these assembledgraphenefilmsshowrelativelypoorelectricalconductivity owingtothepoorinterlayerjunctioncontactresistanceandthe structuraldefectsformedduringthevigorousexfoliationandreductionprocesses.Inthiswork,wedevelopatechniqueforgrowingfewlayergraphenefilmsusingchemicalvapourdeposition(CVD)and successfullytransferringthefilmstoarbitrarysubstrateswithout intensemechanicalandchemicaltreatments,topreservethehigh crystallinequalityofthegraphenesamples.Therefore,weexpectto observeenhancedelectricalandmechanicalproperties.Thegrowth, etchingandtransferringprocessesoftheCVD-grownlarge-scale graphenefilmsaresummarizedinFig.1. Ithasbeenknownforover40yearsthatCVDofhydrocarbonson reactivenickelortransition-metal-carbidesurfacescanproducethin graphiticlayers19.However,thelargeamountofcarbonsources absorbedonnickelfoilsusuallyformthickgraphitecrystalsrather thangraphenefilms(Fig.2a).Tosolvethisproblem,thinlayersof nickelofthicknesslessthan300nmweredepositedonSiO2/Sisubstratesusinganelectron-beamevaporator,andthesampleswerethen heatedto1,000 uCinsideaquartztubeunderanargonatmosphere. Afterflowingreactiongasmixtures(CH4:H2:Ar 5 50:65:200standard cubiccentimetresperminute),werapidlycooledthesamplestoroom temperature( , 25 u C)attherateof , 10 u Cs2 1usingflowingargon. Wefoundthatthisfastcoolingrateiscriticalinsuppressingformation ofmultiplelayersandforseparatinggraphenelayersefficientlyfrom thesubstrateinthelaterprocess20. Ascanningelectronmicroscope(SEM;JSM6490,Jeol)imageof graphenefilmsonathinnickelsubstrateshowsclearcontrastbetween areaswithdifferentnumbersofgraphenelayers(Fig.2a).Transmission electronmicroscope(TEM;JEM3010,Jeol)images(Fig.2b)showthat thefilmmostlyconsistsoflessthanafewlayersofgraphene.After transferofthefilmtoasiliconsubstratewitha300-nm-thickSiO2layer,opticalandconfocalscanningRamanmicroscope(CRM200, Witech)imagesweremadeofthesamearea(Fig.2c,d)22.Thebrightest areainFig.2dcorrespondstomonolayers,andthedarkestareais composedofmorethantenlayersofgraphene.Bilayerstructures appeartopredominateinbothTEMandRamanimagesforthis particularsample,whichwaspreparedfrom7minofgrowthona 300-nm-thicknickellayer.Wefoundthattheaveragenumberofgraphenelayers,thedomainsizeandthesubstratecoveragecanbecontrolledbychangingthenickelthicknessandgrowthtimeduringthe growthprocess(SupplementaryFigs1and2),thusprovidingawayof controllingthegrowthofgraphenefordifferentapplications. Atomicforcemicroscope(AFM;NanoscopesIIIaandE,Digital Instruments)imagesoftenshowtheripplestructurescausedbythe differencebetweenthethermalexpansioncoefficientsofnickeland graphene(Fig.2c,inset;seealsoSupplementaryFig.3)19.Webelieve thattheseripplesmakethegraphenefilmsmorestableagainstmechanicalstretching23,makingthefilmsmoreexpandable,aswewill discusslater.Multilayergraphenesamplesarepreferableintermsof mechanicalstrengthforsupportinglarge-areafilmstructures,whereas thinnergraphenefilmshavehigheropticaltransparency.Wefindthat a , 300-nm-thicknickellayeronasiliconwaferistheoptimalsubstrateforthelarge-scaleCVDgrowththatyieldsmechanicallystable, transparentgraphenefilmstobetransferredandstretchedafterthey areformed,andthatthinnernickellayerswithashortergrowthtime yieldpredominantlymono-andbilayergraphenefilmformicroelectronicdeviceapplications(SupplementaryFig.1c). 1DepartmentofChemistry,2SchoolofAdvancedMaterialsScienceandEngineering,3SKKUAdvancedInstituteofNanotechnology,4CenterforNanotubesandNanostructured Composites,SungkyunkwanUniversity,Suwon440-746,Korea.5SamsungAdvancedInstituteofTechnology,POBox111,Suwon440-600,Korea.6DepartmentofChemistry,Pohang UniversityofScienceandTechnology,Pohang790-784,Korea.7DepartmentofPhysics,ColumbiaUniversity,NewYork,NewYork10027,USA. Vol457|5February2009|doi:10.1038/nature07719706 Macmillan Publishers Limited. All rights reserved2009
SiO2 (300 nm) Ni/C layer CH4/H2/Ar ~1,000 C Ar Cooling ~RT Patterned Ni layer (300 nm) FeCl3(aq) or acids Ni-layer etching HF/BOE SiO2-layer etching (short) Ni-layer etching (long) PDMS/graphene Downside contact (scooping up) Graphene on a substrate HF/BOE Stamping Floating graphene/ Ni Floating graphene Graphene/Ni/ SiO2/Sia b cPDMS/graphene/ Ni /SiO2/Si Ni Si Figure1|Synthesis,etchingand transferprocessesforthelargescaleandpatternedgraphene films.a,Synthesisofpatterned graphenefilmsonthinnickellayers.b,EtchingusingFeCl3(oracids) andtransferofgraphenefilmsusing aPDMSstamp.c,Etchingusing BOEorhydrogenfluoride(HF) solutionandtransferofgraphene films.RT,roomtemperature ( , 25 u C). 1,500 2,000 2,500Intensity (a.u.)Raman shift (cm) >4 layers 3 layers Bilayer Monolayera c5 m 5 me5 m = 532 nm 2 m 3 layers Bilayer 4 layers 0.34 nmb>10 layers G 2D D 5 md>5 4 3 2 1 Figure2|Variousspectroscopicanalysesofthelarge-scalegraphenefilms grownbyCVD.a,SEMimagesofas-growngraphenefilmsonthin(300-nm) nickellayersandthick(1-mm)Nifoils(inset).b,TEMimagesofgraphene filmsofdifferentthicknesses.c,Anopticalmicroscopeimageofthe graphenefilmtransferredtoa300-nm-thicksilicondioxidelayer.Theinset AFMimageshowstypicalrippledstructures.d,AconfocalscanningRaman imagecorrespondingtoc.Thenumberoflayersisestimatedfromthe intensities,shapesandpositionsoftheG-bandand2D-bandpeaks.e,Raman spectra(532-nmlaserwavelength)obtainedfromthecorresponding colouredspotsincandd.a.u.,arbitraryunits. de gh 2 cm 2 cm StampingPatterned grapheneab f c 5 mm Figure3|Transferprocessesforlarge-scalegraphenefilms.a,A centimetre-scalegraphenefilmgrownonaNi(300nm)/SiO2(300nm)/Si substrate.b,Afloatinggraphenefilmafteretchingthenickellayersin1M FeCl3aqueoussolution.Aftertheremovalofthenickellayers,thefloating graphenefilmcanbetransferredbydirectcontactwithsubstrates.c,Various shapesofgraphenefilmscanbesynthesizedontopofpatternednickellayers.d,e,Thedry-transfermethodbasedonaPDMSstampisusefulin transferringthepatternedgraphenefilms.AfterattachingthePDMS substratetothegraphene(d),theunderlyingnickellayerisetchedand removedusingFeCl3solution(e).f,GraphenefilmsonthePDMSsubstrates aretransparentandflexible.g,h,ThePDMSstampmakesconformalcontact withasilicondioxidesubstrate.Peelingbackthestamp(g)leavesthefilmon aSiO2substrate(h).NATURE|Vol457|5February2009LETTERS707 Macmillan Publishers Limited. All rights reserved2009
Etchingnickelsubstratelayersandtransferringisolatedgraphene filmstoothersubstratesisimportantfordeviceapplications.Usually, nickelcanbeetchedbystrongacidsuchasHNO3,whichoften produceshydrogenbubblesanddamagesthegraphene.Inourwork, anaqueousiron(III)chloride(FeCl3)solution(1M)wasusedasan oxidizingetchanttoremovethenickellayers.Thenetionicequation oftheetchingreactioncanberepresentedasfollows: 2Fe3 z( aq ) z Ni ( s ) ? 2Fe2 z( aq ) z Ni2 z( aq ) Thisredoxprocessslowlyetchesthenickellayerseffectivelywithina mildpHrangewithoutforminggaseousproductsorprecipitates.Ina fewminutes,thegraphenefilmseparatedfromthesubstratefloatson thesurfaceofthesolution(Fig.3a,b),andthefilmisthenreadytobe transferredtoanykindofsubstrate.Useofbufferedoxideetchant (BOE)orhydrogenfluoridesolutionremovessilicondioxidelayers, sothepatternedgrapheneandthenickellayerfloattogetheronthe solutionsurface.Aftertransfertoasubstrate,furtherreactionwith BOEorhydrogenfluoridesolutioncompletelyremovestheremainingnickellayers(SupplementaryFig.5). Wealsodevelopadry-transferprocessforthegraphenefilmusing asoftsubstratesuchaspolydimethylsiloxane(PDMS)stamp24.Here wefirstattachthePDMSstamptotheCVD-growngraphenefilmon thenickelsubstrate(Fig.3d).Thenickelsubstratecanbeetchedaway usingFeCl3asdescribedabove,leavingtheadheredgraphenefilmon thePDMSsubstrate(Fig.3e).Byusingthepre-patternednickel substrate(Fig.3c),wecantransfervarioussizesandshapesofgraphenefilmtoanarbitrarysubstrate.Thisdry-transferprocessturns outtobeveryusefulinmakinglarge-scalegrapheneelectrodesand deviceswithoutadditionallithographyprocesses(Fig.3fâ€“h). Microscopically,thesefew-layertransferredgraphenefilmsoften showlinearcrackpatternswithanangleof60 u or120 u,indicating aparticularcrystallographicedgewithlargecrystallinedomains (SupplementaryFig.1b)25.Inaddition,theRamanspectrameasured forgraphenefilmsonnickelsubstratesshowastronglysuppressed defect-relatedD-bandpeak(SupplementaryFig.3).ThisDpeak growsonlyslightlyafterthetransferprocess(Fig.2e),indicating overallgoodqualityoftheresultinggraphenefilm.Furtheroptimizationofthetransferprocesswithsubstratecontrolmakespossible transferyieldsapproaching99%(SupplementaryTable1). 83.7% UV for 6 h 80.7% UV for 4 h 79.1% UV for 2 h 76.3% Initial at 550 nm 0101102103104105106107108Stretching (%)Flat 0 1 2 3 4 5 6 7 8 9Resistance (k )Bending radius (mm)RyRxBending0.0 100101102Anisotropy, Ry/RxCurvature, ! (mm)y x y x RxRyx y1011021031043rd 2nd 1stResistance ( )Stretching (%)RyRxStable Stretching cycles0 0.2 0.4 0.6 0.8 1.0 1.2 76 78 80 82 84TrTr (%)RsRs (k per square)Time (h)400 60 65 70 75 80 85 90Transmittance (%)Wavelength (nm) 0 0 0Vg (V)Magnetoresistance (k )0 10 54 2 0 0060Resistance (k )Vg (V)0 60 40 20 1,200 1,000 800 600654 3 21Recovery1.2 0.8 0.4Flat 0.8 1.0 2.3 2.7 3.5Resistance ( )30 25 20 15 10 56 3 0 6 3 0 6 3 0ab cd Figure4|Opticalandelectricalpropertiesofthegraphenefilms. a,Transmittanceofthegraphenefilmsonaquartzplate.Thediscontinuities intheabsorptioncurvesarisefromthedifferentsensitivitiesoftheswitching detectors.Theupperinsetshowstheultraviolet(UV)-inducedthinningand theconsequentenhancementoftransparency.Thelowerinsetshowsthe changesintransmittance,Tr,andsheetresistance, Rs,asfunctionsof ultravioletilluminationtime.b,Electricalpropertiesofmonolayergraphene devicesshowingthehalf-integerquantumHalleffectandhighelectron mobility.Theupperinsetshowsafour-probeelectricalresistance measurementonamonolayergrapheneHallbardevice(lowerinset)at 1.6K.Weapplyagatevoltage, Vg,tothesiliconsubstratetocontrolthe chargedensityinthegraphenesample.Themainpanelshowslongitudinal ( Rxx)andtransverse( Rxy)magnetoresistancesmeasuredinthisdevicefora magneticfield B 5 8.8T.ThemonolayergraphenequantumHalleffectis clearlyobserved,showingtheplateauxwithfillingfactor n 5 2at Rxy5 (2 e2/ h )2 1andzerosin Rxx.(Here e istheelementarychargeand h isPlanckâ€™s constant.)QuantumHallplateaux(horizontaldashedlines)aredeveloping forhigherfillingfactors.c,Variationinresistanceofagraphenefilm transferredtoa , 0.3-mm-thickPDMS/PETsubstratefordifferentdistances betweenholdingstages(thatis,fordifferentbendingradii).Theleftinset showstheanisotropyinfour-proberesistance,measuredastheratio, Ry/Rx, oftheresistancesparallelandperpendiculartothebendingdirection, y .The rightinsetshowsthebendingprocess.d,Resistanceofagraphenefilm transferredtoaPDMSsubstrateisotropicallystretchedby , 12%.Theleft insetshowsthecaseinwhichthegraphenefilmistransferredtoan unstretchedPDMSsubstrate.Therightinsetshowsthemovementof holdingstagesandtheconsequentchangeinshapeofthegraphenefilm.LETTERSNATURE|Vol457|5February2009708 Macmillan Publishers Limited. All rights reserved2009
Forthemacroscopictransportelectrodeapplication,theoptical andelectricalpropertiesof1 3 1cm2graphenefilmswererespectively measuredbyultravioletâ€“visiblespectrometerandfour-probeVander Pauwmethods(Fig.4a,b).Wemeasuredthetransmittanceusingan ultravioletâ€“visiblespectrometer(UV-3600,Shimazdu)aftertransferringthefloatinggraphenefilmtoaquartzplate(Fig.4a).Inthevisible range,thetransmittanceofthefilmgrownona300-nm-thicknickel layerfor7minis , 80%,avaluesimilartothosefoundforpreviously studiedassembledfilms2,3.Becausethetransmittanceofanindividual graphenelayeris , 2.3%(ref.26),thistransmittancevalueindicates thattheaveragenumberofgraphenelayersissixtoten.Thetransmittancecanbeincreasedto , 93%byfurtherreducingthegrowthtime andnickelthickness,resultinginathinnergraphenefilm(SupplementaryFig.1).Ultraviolet/ozoneetching(ultraviolet/ozonecleaner, 60W,BioForce)isalsousefulincontrollingthetransmittanceinan ambientcondition(Fig.4a,upperinset).Indiumelectrodeswere depositedoneachcornerofthesquare(Fig.4a,lowerinset)tominimizecontactresistance.Theminimumsheetresistanceis , 280 V per square,whichis , 30timessmallerthanthelowestsheetresistance measuredonassembledfilms2,3.Thevaluesofsheetresistanceincrease withtheultraviolet/ozonetreatmenttime,inaccordancewiththe decreasingnumberofgraphenelayers(Fig.4a). Formicroelectronicapplication,themobilityofthegraphenefilm iscritical.Tomeasuretheintrinsicmobilityofasingle-domaingraphenesample,wetransferredthegraphenesamplesfromaPDMS stamptoadegeneratedopedsiliconwaferwitha300-nm-deepthermallygrownoxidelayer.Monolayergraphenesampleswerereadily locatedonthesubstratefromtheopticalcontrast26andtheiridentificationwassubsequentlyconfirmedbyRamanspectroscopy22. Electron-beamlithographywasusedtomakemulti-terminaldevices (Fig.4b,lowerinset).Notably,themulti-terminalelectricalmeasurementsshowedthattheelectronmobilityis , 3,750cm2V2 1s2 1ata carrierdensityof , 5 3 1012cm2 2(Fig.4b).Forahighmagneticfield of8.8T,weobservethehalf-integerquantumHalleffect(Fig.4b) correspondingtomonolayergraphene4,5,indicatingthatthequality ofCVD-growngrapheneiscomparabletothatofmechanically cleavedgraphene(SupplementaryFig.6)6. Inadditiontothegoodopticalandelectricalproperties,thegraphenefilmhasexcellentmechanicalpropertieswhenusedtomake flexibleandstretchableelectrodes(Fig.4c,d)7.Weevaluatedthefoldabilityofthegraphenefilmstransferredtoapolyethyleneterephthalate (PET)substrate(thickness, , 100 m m)coatedwithathinPDMSlayer (thickness, , 200 m m;Fig.4c)bymeasuringresistanceswithrespectto bendingradii.Theresistancesshowlittlevariationuptothebending radiusof2.3mm(approximatetensilestrainof6.5%)andareperfectly recoveredafterunbending.Notably,theoriginalresistancecanbe restoredevenforthebendingradiusof0.8mm(approximatetensile strainof18.7%),exhibitingextrememechanicalstabilityincomparisonwithconventionalmaterialsusedinflexibleelectronics27. Theresistancesofgraphenefilmstransferredtopre-strainedand unstrainedPDMSsubstratesweremeasuredwithrespecttouniaxial tensilestrainrangingfrom0to30%(Fig.4d).Similartotheresultsin thefoldingexperiment,thetransferredfilmonanunstrainedsubstraterecoversitsoriginalresistanceafterstretchingby , 6%(Fig.4d, leftinset).However,furtherstretchingoftenresultsinmechanical failure.Thus,wetriedtotransferthefilmtopre-strainedsubstrates28toenhancetheelectromechanicalstabilitiesbycreatingripplessimilar tothoseobservedinthegrowthprocess(Fig.2c,inset;Supplementary Fig.4).ThegraphenetransferredtoalongitudinallystrainedPDMS substratedoesnotshowmuchenhancement,owingtothetransverse straininducedbyPoissonâ€™seffect29.Topreventthisproblem,the PDMSsubstratewasisotropicallystretchedby , 12%beforetransferringthefilmtoit(Fig.4d).Surprisingly,bothlongitudinalandtransverseresistances( Ryand Rx)appearstableupto , 11%stretchingand showonlyoneorderofmagnitudechangeat , 25%stretching.We supposethatfurtheruniaxialstretchingcanchangetheelectronic bandstructuresofgraphene,leadingtothemodulationofthe sheetresistance.Theseelectromechanicalpropertiesthusshowour graphenefilmstobenotonlythestrongest7butalsothemostflexible andstretchableconductingtransparentmaterialssofarmeasured26. Inconclusion,wehavedevelopedasimplemethodtogrowand transferhigh-qualitystretchablegraphenefilmsonalargescaleusing CVDonnickellayers.Thepatternedfilmscaneasilybetransferredto stretchablesubstratesbysimplecontactmethods,andthenumberof graphenelayerscanbecontrolledbyvaryingthethicknessofthe catalyticmetals,thegrowthtimeand/ortheultraviolettreatment time.BecausethedimensionsofthegraphenefilmsarelimitedsimplybythesizeoftheCVDgrowthchamber,scalingupcanbereadily achieved,andtheoutstandingoptical,electricalandmechanical propertiesofthegraphenefilmsenablenumerousapplications includinguseinlarge-scaleflexible,stretchable,foldabletransparent electronics8,9,30.Received5October;accepted8December2008. Publishedonline14January2009.1.Geim,A.K.&Novoselov,K.S.Theriseofgraphene. NatureMater. 6, 183â€“191 (2007). 2.Li,X. etal. HighlyconductinggraphenesheetsandLangmuirâ€“Blodgettfilms. NatureNanotechnol. 3, 538â€“542(2008). 3.Eda,G.,Fanchini,G.&Chhowalla,M.Large-areaultrathinfilmsofreduced grapheneoxideasatransparentandflexibleelectronicmaterial. Nature Nanotechnol. 3, 270â€“274(2008). 4.Novoselov,K.S. etal. Two-dimensionalgasofmasslessDiracfermionsin graphene. Nature 438, 197â€“200(2005). 5.Zhang,Y.,Tan,J.W.,Stormer,H.L.&Kim,P.Experimentalobservationofthe quantumHalleffectandBerryâ€™sphaseingraphene. Nature 438, 201â€“204(2005). 6.Novoselov,K.S. etal. Electricfieldeffectinatomicallythincarbonfilms. Science 306, 666â€“669(2004). 7.Lee,C.,Wei,X.,Kysar,J.W.&Hone,J.Measurementoftheelasticpropertiesand intrinsicstrengthofmonolayergraphene. Science 321, 385â€“388(2008). 8.Kim,D.-H. etal. Stretchableandfoldablesiliconintegratedcircuits. Science 320, 507â€“511(2008). 9.Sekitani,T. etal. Arubberlikestretchableactivematrixusingelasticconductors. Science 321, 1468â€“1472(2008). 10.Han,M.Y.,Oezyilmaz,B.,Zhang,Y.&Kim,P.Energybandgapengineeringof graphenenanoribbons. Phys.Rev.Lett. 98, 206805(2007). 11.Bolotin,K.I. etal. Ultrahighelectronmobilityinsuspendedgraphene. SolidState Commun. 146, 351â€“355(2008). 12.Bunch,J.S. etal. Electromechanicalresonatorsfromgraphenesheets. Science 315, 490â€“493(2008). 13.Ohta,T.,Bostwick,A.,Seyller,T.,Horn,K.&Rotenberg,E.Controllingthe electronicstructureofbilayergraphene. Science 313, 951â€“954(2006). 14.Berger,C. etal. Electronicconfinementandcoherenceinpatternedepitaxial graphene. Science 312, 1191â€“1196(2006). 15.Sutter,P.W.,Flege,J.-I.&Sutter,E.A.Epitaxialgrapheneonruthenium. Nature Mater. 7, 406â€“411(2008). 16.Dikin,D.A. etal. Preparationandcharacterizationofgrapheneoxidepaper. Nature 448, 457â€“460(2007). 17.Stankovich,S, etal. Graphene-basedcompositematerials. Nature 442, 282â€“286 (2006). 18.Li,D.,Muller,M.B.,Gilje,S.,Kaner,R.B.&Wallace,G.G.Processableaqueous dispersionsofgraphenenanosheets. NatureNanotechnol. 3, 101â€“105(2008). 19.Obraztsov,A.N.,Obraztsova,E.A.,Tyurnina,A.V.&Zolotukhin,A.A.Chemical vapordepositionofthingraphitefilmsofnanometerthickness. Carbon 45, 2017â€“2021(2007). 20.Yu,Q, etal. GraphenesegregatedonNisurfacesandtransferredtoinsulators. Appl.Phys.Lett. 93, 113103(2008). 21.Reina,A. etal. Largearea,few-layergraphenefilmsonarbitrarysubstratesby chemicalvapordeposition. NanoLett. articleASAPat http://pubs.acs.org/doi/ abs/10.1021/nl801827v (2008). 22.Ferrari,A.C. etal. Ramanspectrumofgrapheneandgraphenelayers. Phys.Rev. Lett. 97, 187401(2006). 23.Khang,D.-Y. etal. Individualalignedsingle-wallcarbonnanotubesonelastomeric substrates. NanoLett. 8, 124â€“130(2008). 24.Yang,P. etal. Mirrorlesslasingfrommesostructuredwaveguidespatternedby softlithography. Science 287, 465â€“467(2000). 25.Li,X.,Wang,X.,Zhang,L.,Lee,S.&Dai,H.Chemicallyderived,ultrasmooth graphenenanoribbonsemiconductors. Science 319, 1229â€“1232(2008). 26.Nair,R.R. etal. Finestructureconstantdefinesvisualtransparencyofgraphene. Science 320, 1308(2008). 27.Lewis,J.Materialchallengeforflexibleorganicdevices. Mater.Today 9, 38â€“45 (2006). 28.Sun,Y.,Choi,W.M.,Jiang,H.,Huang,Y.Y.&Rogers,J.A.Controlledbucklingof semiconductornanoribbonsforstretchableelectronics. NatureNanotechnol. 1, 201â€“207(2006). NATURE|Vol457|5February2009LETTERS709 Macmillan Publishers Limited. All rights reserved2009
29.Khang,D.-Y.,Jiang,H.,Huang,Y.&Rogers,J.A.Astretchableformofsinglecrystalsiliconforhigh-performanceelectronicsonrubbersubstrates. Science 311, 208â€“212(2006). 30.Ko,H.C. etal. Ahemisphericalelectroniceyecamerabasedoncompressible siliconoptoelectronics. Nature 454, 748â€“753(2008).SupplementaryInformation islinkedtotheonlineversionofthepaperat www.nature.com/nature. Acknowledgements WethankJ.H.Han,J.H.Kim,H.Lim,S.K.BaeandH.-J.Shin forassistingingraphenesynthesisandanalysis.Thisworkwassupportedbythe KoreaScienceandEngineeringFoundationgrantfundedbytheKoreaMinistryfor Education,ScienceandTechnology(CenterforNanotubesandNanostructured CompositesR11-2001-091-00000-0),theGlobalResearchLabprogramme (KoreaFoundationforInternationalCooperationofScienceandTechnology),the BrainKorea21project(KoreaResearchFoundation)andtheinformation technologyresearchanddevelopmentprogrammeoftheKoreaMinistryof KnowledgeEconomy(2008-F024-01). AuthorContributions B.H.H.plannedandsupervisedtheproject;J.-Y.C.supported andassistedinsupervisionontheproject;S.Y.L,J.M.K.andK.S.K.advisedonthe project;K.S.K.andB.H.H.designedandperformedtheexperiments;B.H.H.,P.K., J.-H.AandK.S.K.analyseddataandwrotethemanuscript;Y.Z.andP.K.madethe quantumHalldevicesandthemeasurements;andH.J.andJ.-H.A.helpedwiththe transferprocessandtheelectromechanicalanalyses. AuthorInformation Reprintsandpermissionsinformationisavailableat www.nature.com/reprints.Correspondenceandrequestsformaterialsshouldbe addressedtoB.H.H.(email@example.com)orJ.-Y.C. (firstname.lastname@example.org). LETTERSNATURE|Vol457|5February2009710 Macmillan Publishers Limited. All rights reserved2009
A PREDICTIVE MODEL FOR THE ANTIFOULING EFFICACY OF ENGINEERED MICROTOPOGRAPHIES By JOSEPH THOMAS DECKER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMEN TS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014
Â© 2014 Joseph Thomas Decker
4 ACKNOWLEDGMENTS I would first like to thank my advisor Dr. Anthony Brennan for his guidance and support throughout the process of my graduate work. I would also like to thank my committee members for their advice and guidance: Dr. Christopher Batich, Dr. Josephine Allen, Dr. Simon Phillpot and Dr. Thomas Angelini. I would also like to extend thank s to our collaborators who helped with the bioassay work: Dr. John Finlay, Dr. Maureen Callow and Dr. James Callow for their help with the Ulva assays and Dr. Nick Aldred and Dr. Anthony Clare for their help with the barnacle assays. The help of my fellow graduate students was invaluable through out this process. I would like to thank the past members of the Brennan Research Group: Dr. Chelsea Kirschner, Dr. Jiun Jeng Chen, Dr. David Jackson, Dr. Angel Ejiasi, Dr. Scott Cooper, Dr. Julian Sheats and Ms. Ad woa Baah Dwomoh. I would also like to extend my thanks to the current members of the Brennan Group for their support: Dr. Canan Kizilkaya, Mr. Clayton Argenbright, Mr. Francisco Castro Cara, Mr. Cary Kuliasha and Ms. Laura Villada. I would also like to thank Ms. Danessa Jerome for her help with topography characterization, Mr. Steven Zehnder for his help with the cell adhesion experiments, Mr. Ryan Nixon for his help troubleshooting equipment. I would also like to express my gratitude to my parents , whos e steadfast support during my education made this achievement possible . I would also like to thank the Sagstetter family for welcoming me into their family. Finally, I would like to thank my loving wife Ann , without whom I never would have finished .
5 TA BLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 21 ABSTRACT ................................ ................................ ................................ ................... 22 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 24 Scope of Research ................................ ................................ ................................ . 24 Specific Aims ................................ ................................ ................................ .......... 24 Specific Aim 1. Develop a Mechanistic Model that C ombines the A ttachment Point Theory and ERI Models for Fouling on T opographies ....... 24 Specific Aim 2. Explore the R e lationship Between the Spatial D istributi on of Attachment Sites and the Spatial Distribution of Attached Z oospores of Ulva linza ................................ ................................ ................................ ....... 25 Specific Aim 3. Examine the Work of Adhesion on Engineered Microtopographies Using Contact A ngle and AFM ................................ ........ 25 Specific Aim 4. Systematically Probe the Relationship B etween Topographic D imen sions and Configuration on the Behavior of both Mi crofouling and Macrofouling O rganisms ................................ .................... 26 2 BACKGROUND ................................ ................................ ................................ ...... 28 Issues A ssociate d with Marine F ouling ................................ ................................ ... 28 Model Systems for Marine F ouling ................................ ................................ ......... 30 Organic F ouling ................................ ................................ ................................ 31 Microfouling ................................ ................................ ................................ ...... 32 Macrofouling ................................ ................................ ................................ ..... 34 Design of N ontox ic Antifouling C oatings ................................ ................................ . 37 Colloidal I nteractions ................................ ................................ ........................ 37 Fracture M echanics ................................ ................................ .......................... 41 Surface Energy ................................ ................................ ................................ . 44 Surface Roughness ................................ ................................ .......................... 52 Summary ................................ ................................ ................................ ................ 63 3 SEA MODEL DEVELOPMENT ................................ ................................ ............... 65 Background ................................ ................................ ................................ ............. 65
6 Analysis ................................ ................................ ................................ .................. 68 Experimental Methods ................................ ................................ ............................ 73 Materials ................................ ................................ ................................ ........... 73 Pattern Selection ................................ ................................ .............................. 73 Cell Attachment D ata ................................ ................................ ....................... 76 Cell Attachment Area C alculations ................................ ................................ ... 77 Simulation ................................ ................................ ................................ ......... 79 Results and Discussion ................................ ................................ ........................... 81 Unit Cell Modeling ................................ ................................ ............................ 81 Areal Attachment Density P redictions ................................ .............................. 82 S ummary ................................ ................................ ................................ ................ 89 4 RADIAL DISTRIBUTION OF ATTACHED ZOOSPORES AS A FUNCTION OF TOPOGRAPHY ................................ ................................ ................................ ...... 91 B ackground ................................ ................................ ................................ ............. 91 Materials and Methods ................................ ................................ ............................ 95 U. linza Zoospore Attachment M ethodology ................................ ..................... 95 Image P rocessing ................................ ................................ ............................. 96 Radial D istr ibution F unction ................................ ................................ .............. 96 Zoospore Distribution Unit Cell M apping ................................ .......................... 99 Modeling ................................ ................................ ................................ ........... 99 Results and Discussion ................................ ................................ ......................... 100 N series Distribution ................................ ................................ ....................... 100 Statistical Comparison Between N Series Surfaces ................................ ....... 103 Kinetic Study Distributions ................................ ................................ .............. 105 Simulated Distributions ................................ ................................ ................... 109 Zoo spore Distribution U nit Cell M apping ................................ ........................ 111 S ummary ................................ ................................ ................................ .............. 114 5 EXAMINATION OF THE WORK OF ADHESION ON ENGINEERED MICROTOPOGRAPHIES ................................ ................................ ..................... 116 B ackground ................................ ................................ ................................ ........... 116 Materials and Methods ................................ ................................ .......................... 123 Pattern S election ................................ ................................ ............................ 123 Contact A ngle M easurements ................................ ................................ ........ 125 Contact Line Images ................................ ................................ ...................... 126 AFM M easuremen ts ................................ ................................ ....................... 126 Statistical Analysis ................................ ................................ .......................... 127 Results and Discussion ................................ ................................ ......................... 127 Contact Angle Measurements ................................ ................................ ........ 127 Comparison of Contact Angle Data to s ................................ ....................... 132 Work of Adhesion Calculations from Contact Angle Data .............................. 137 AFM M easurements ................................ ................................ ....................... 138 S ummary ................................ ................................ ................................ .............. 144
7 6 HIGH THROUGHPUT METHODS OF EXAMIN G THE ATTACHMENT OF ULVA LINZA TO ENGINEERED MICROTOPOGRAPHIES ................................ . 145 B ackground ................................ ................................ ................................ ........... 145 Materials and Methods ................................ ................................ .......................... 149 Pattern Design ................................ ................................ ................................ 149 Slide Design ................................ ................................ ................................ ... 155 Mold Fabrication ................................ ................................ ............................. 156 Bioassay Sample Preparation ................................ ................................ ........ 157 Scanning Electron Microscopy ................................ ................................ ....... 158 Ulva Bioassay ................................ ................................ ................................ . 158 Image Processing ................................ ................................ ........................... 159 Image Data Analysis ................................ ................................ ...................... 159 Statistical Analysis of the Effect o f Pattern L ocation ................................ ....... 160 Results and Discussion ................................ ................................ ......................... 160 Assay 1 April 2013 Assay ................................ ................................ ............ 160 Assay 2 April 2014 Assay ................................ ................................ ............ 164 Effect of Pattern Location on Spore Density ................................ ................... 165 Comparison to SEA Model Prediction ................................ ............................ 170 Spore Distribution Maps ................................ ................................ ................. 173 S ummary ................................ ................................ ................................ .............. 178 7 BARNACLE ATTACHMENT AND PROBING BEHAVIOR ON TOPOGRAPHICALLY MODIFIED SURFACES ................................ .................... 180 Background ................................ ................................ ................................ ........... 180 Materials and Method ................................ ................................ ........................... 184 Pattern Design ................................ ................................ ................................ 184 Mold Fabrication ................................ ................................ ............................. 185 Bioassay Sample Preparation ................................ ................................ ........ 186 Barnacle assay ................................ ................................ ............................... 186 Barnacle video assay ................................ ................................ ..................... 186 Statistical Analysis ................................ ................................ .......................... 187 Results and Discussion ................................ ................................ ......................... 187 Attachment Assay 1 ................................ ................................ ....................... 187 Attachment Assay 2 ................................ ................................ ....................... 190 Comparison to the SEA model ................................ ................................ ....... 190 Video M icroscopy ................................ ................................ ........................... 191 Discussion ................................ ................................ ................................ ............ 199 Summary ................................ ................................ ................................ .............. 203 8 CONCLUSIONS AND FUTURE WORK ................................ ............................... 204 Conclusions ................................ ................................ ................................ .......... 204 Future Work ................................ ................................ ................................ .......... 207 APPENDIX
8 A MATLAB IMAGE ANALYSIS ................................ ................................ ................ 210 B ANALYSIS OF DIFFERENCES BETWEEN WINGS AND SHOULDERS ON SILICONE RETAIN SAMPLES ................................ ................................ ............. 215 Problem Statement ................................ ................................ ............................... 215 Analysis of Retain Samples ................................ ................................ .................. 216 ATR FTIR ................................ ................................ ................................ ....... 218 Captive Air Bubble Measurements ................................ ................................ . 223 Acrylate Grafting ................................ ................................ ............................. 225 Summ ary ................................ ................................ ................................ .............. 225 C MADIN DARBY CANINE KIDNEY CELL ADHESION TO TOPOGRAPHIES ....... 227 B ackground ................................ ................................ ................................ ........... 227 Materials and Methods ................................ ................................ .......................... 227 Results ................................ ................................ ................................ .................. 228 Summary ................................ ................................ ................................ .............. 229 D CONTACT LINE IMAGES ................................ ................................ ..................... 230 REFERENCES ................................ ................................ ................................ ............ 234 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 248
9 LIST OF TABLES Table page 3 1 Literature references for data used to validate the SEA model. .......................... 68 4 1 Pair distribution function data for individual and groups of spores on different topographies ................................ ................................ ................................ ....... 96 4 2 Ratio between the first and second peak in the pair distribution function for the n4 topography and smooth measured at different settlement times ........... 101 4 3 Pair distribution function data for individual and groups of spores as a function of time on the n4 topography and smooth PD MSe ............................. 103 4 4 Simulated results for spore distributions on different surfaces .......................... 104 6 1 Topography designations for this study. The h eight, width and spacing refer to the April 2013 assay ................................ ................................ ..................... 147 7 1 Topographies used for the first Balanus assay ................................ ................. 177 B 1 Average conta ct angle measurements on control and retain surfaces. ............ 207 B 2 Surface energy calculated from contact angle data from the control and retain samples. ................................ ................................ ................................ . 209 B 3 Samples identified for additional testing ................................ ........................... 210 B 4 RMS roughness values for different test surfaces. Colors indicate similar surfaces in terms of casting method ................................ ................................ . 214
10 LIST OF FIGURES Figure page 2 1 Hierarchy of model marine fouling organisms. ................................ .................... 28 2 2 Z oospores of Ulva linza preferentially attach in the ch annels of a 5x5 um topography ................................ ................................ ................................ .......... 30 2 3 Relationship between assay time and spore attachment density on smooth and topographically modified surfaces ................................ ............................... 30 2 4 Initial attachment of Navicula incerta with random orientation and d istribution. .. 31 2 5 Foot appendage of the cyprid of Balanus amphitrite. Images A D are different views of the foot ................................ ................................ .................... 32 2 6 Contact angle for a liquid drop on a smooth surface ................................ .......... 42 2 7 Original Baier curve showing an empirical relationship between critical surface tension ( x axis) and relative biological interaction (y axis) ..................... 45 2 8 Schematic of the Wenzel wetting state in which the entire topography (tops and side walls) are wetted. ................................ ................................ ................. 49 2 9 Schematic of the Cassie Baxter wetting state. ................................ ................... 49 2 10 Map of individually attached Ulva zoospores relative to the Sharklet unit cell .... 51 2 11 Plot of attachment density v. ERI I ................................ ................................ ....... 53 2 12 Second version of the ERI ................................ ................................ .................. 54 2 13 The Sharklet microtopography. ................................ ................................ ........... 57 2 14 Hierarchal structure for the combined repellence of barnacles and algae .......... 58 3 1 The unit cell seen here offers three different sites with different numbers of attachment points. Features that offer 1, 2, and 3 attachment po ints at each location are highlighted in yellow. ................................ ................................ ....... 61 3 2 Plot of relative settlement against E quation 3 2 3 . ................................ ............... 62 3 3 A lattice can be applied to the Sharklet AF unit cell in Figure 3 2 to accurately describe the differences between the sites. ................................ ........................ 63 3 4 ies tested against Ulva linza . ................................ .. 69
11 3 5 "ERI Series" topographies, which systematically alter the r and s values of a topo graphy to alter the ERI value (E quation 3 1). ................................ ............... 69 3 6 features (n) to change the surface's ERI value. ................................ .................. 70 3 7 leaving the number of features and size of features constant. ........................... 70 3 8 (topography B) by altering the spacing between features and the width of features. ................................ ................................ ................................ ............. 70 3 9 Asym metric unit cell of the Sharklet topography.. ................................ ............... 74 3 10 Comparison between simulated and experimentally measured attachment maps for U. linza. ................................ ................................ ................................ 75 3 11 Correlation between relative permanent attachment density of zoospores of U. linza and cells of the diatom N. incerta to topographies with one unique site. ................................ ................................ ................................ ..................... 77 3 12 Correlatio n between relative attachment densities of zoospores of U. linza and cells of C. marina to topographies with narrow features. ............................. 78 3 13 Modified thermodynamic model to include contribution of num ber of features to the attachment density distribution. ................................ ................................ 80 3 14 Comparison between the SEA model and experimentally measured attachment density data for U. linza, N. incerta, C. marina and B. a mphitrite. The data can be compared to E quation 3 11. ................................ .................... 81 4 1 N series topographies analyzed in this study and described previously 32,117 . ..... 88 4 2 Representative images for zoospores of U. linza on the n1 topography (left) and a smooth surface (right) ................................ ................................ ............... 89 4 3 The pair distribution function measures the relative probability o f a zoospore attaching a distance r away from the reference. ................................ ................. 89 4 4 Zoospore mapping schematics for the radial distribution function.. .................... 90 4 5 Radial distribution function for five different topographically modified surfaces as well as a smooth surface. The plots are offset by an arbitrary amount to more clearly illustrate the shape of the curve. The horizontal lines denote a g(r) va lue of one ................................ ................................ ................................ . 94
12 4 6 Pair distribution function for groups of spores on the five n series topographies as well as a smooth surface. Distributions are stacked by an arbitrary amount for clarity ................................ ................................ .................. 95 4 7 Comparison between measures values of the radial distribution function. A) screening distance B) r max for different surfaces from Long et al. 117 Stars indicate statistically distinct gro ups. ................................ ................................ .... 97 4 8 Radial distribution function for individual spores on the n4 topography at different settlement times. Distributions are stacked by an arbitrary amount for clarity ................................ ................................ ................................ ............. 98 4 9 Pair distribution function for the spore aggregates on the n4 topography as a function of time. Distributions are stacked by an arbitrary amount for clarity .... 100 4 10 Pair distribution function for individual spores on a smooth surface as a function of time. Distributions are stacked by an arbitrary amount for clarity .... 100 4 11 Simu lated radial distribution function for 5 m zoospores on the n series topographies. Distributions stacked by an arbitrary amount for clarity ............. 102 4 12 Mapping of r max and screening distance for different locations within the n4 unit c ell. Images are normalized against global averages for n4 (top) and smooth (bottom) to show differences between sites. Colors are scaled based on intensity from red (relatively low value) to green (relatively high value) ....... 104 5 1 Channel series topographies used contact angle measurements .................... 115 5 2 N series topographies used contact angle measurements ............................... 116 5 3 Angle series topographies used contact angle measurements ........................ 116 5 4 The Sharklet topography indicating the different directions for measurement of contact an gles and adhesion data. ................................ ............................... 117 5 5 Summarized results for contact angle measurements on the eight channels series topographies. Bars represent the standard deviation of the measurements ................................ ................................ ................................ .. 120 5 6 Summarized results for contact angle measurements on the eight angle series topographies. Bars represent the standard deviation of the measurements ................................ ................................ ................................ .. 122 5 7 Summarized results for contact angle measurements on the eight angle series topographies. Bars represent the standard deviation of the measurements ................................ ................................ ................................ .. 123
13 5 8 Predicted contact angles compared to experimentally measured contact angles for the 27 patterns in this study. ................................ ............................ 124 5 9 Irregular contact line shape on the +2.6SK2x2_n9 topography. ....................... 125 5 10 Drop on the +2.6CH8x2 topography. ................................ ................................ 125 5 11 The contact line can be drawn in two different ways to estimate the local s value for the topography. A) The method used to calculate Figure 5 8 used a simple straight line relationship that underestimated the contact between the drop and surface. B) Capturing the tortuosity of the contact line led to a larger measured s val ue that more close predicted the measured contact angles. ................................ ................................ ................................ .............. 126 5 12 Predicted contact angles compared to experimentally measured contact angles for the 27 patterns in this study. The s values were calculated using the method shown in Figure 5 11 B. ................................ ................................ . 127 5 13 Normalized work of adhesion W t /W s for DI water on different topographies based on the measured advancing and receding contact a ngles. .................... 129 5 14 Map of the maximum adhesion forces measured by the AFM on the Sharklet microtopography in water. The scale bar is in units of nN ............................... 129 5 15 Map of the maximum attractive force measured by the AFM on the Sharklet microtopography in water. The scale bar is in units of nN ............................... 130 5 16 Work of adhesion at differen t locations within the Sharklet (+2.6SK2x2) microtopography. Red sites indicate sites with a lower work of adhesion than the smooth surface and green sites indicate sites with a higher work of adhesion. ................................ ................................ ................................ .......... 133 6 1 Traditional slide schematic for testing engineered microtopographies. ............ 136 6 2 Channel series topographies used for the 32 pattern array .............................. 140 6 3 N series topographies used for the 32 pattern array ................................ ......... 140 6 4 Angle series topographies for the 32 pattern array ................................ ........... 142 6 5 Asymmetric and geometric patterns for the 32 pattern array ............................ 143 6 6 Schematic for the high throughput Ulva experiments. 32 5x5 mm patterned areas are spaced 1 mm apart on a microscope slide ................................ ....... 145 6 7 Two different pattern configurations used for the April 2013 assay. ................. 145
14 6 8 Post assay images of protrud ing (left) and recessed (right) features. Images courtesy of John Finlay ................................ ................................ ..................... 150 6 9 Attachment density of U. linza to topographies on the 32 pattern array slide. Error bars correspond to the sta ndard error of the measurements. .................. 151 6 10 Ulva density results from the April 2014 assay. Error bars are Â± the standard error of the measurement ................................ ................................ ................. 154 6 11 Each different 32 pattern array has a different distribution of spore densities. A) Randomly configured array from April 2014 B) Configured Array from April 2014 C) Recessed array from April 2013 D) Protruding array from April 201 3 . 155 6 12 The correlation between the attachment density to one topography and the attachment density to the surrounding topographies on a randomly configured sample is very low (R 2 = 0.0028). ................................ ................... 156 6 13 Comparison between Sharklet patterns at different locations within the 32 pattern array shows no statistical differences within the slide ( = 0.05). ......... 158 6 14 The radial distribution function for zoospores of Ulva linza on smooth PDMSe surface from Chapter 4 showed average attachment density about 20 m from the reference spore. ................................ ................................ ................. 159 6 15 Spores (dark spots in image) fit between the channels on the 4x2 channels topography for the randomly configured array ................................ .................. 160 6 16 Plot of the relationship between are a of attachment, available attachment sites and observed attachment density for Ulva linza to the topographies in this study ................................ ................................ ................................ .......... 160 6 17 Mapped spore locations (left column) and screening distance s (right column) for the n and angle series (up to a6) for the configured array from the April 2014 assay. ................................ ................................ ................................ ...... 165 7 1 A) Both channels and Sharklet topography inhibited the attachment of B. amphit rite. Stars are statistically significant groups B) B . amphitrite and U. linza have a similar relationship between topographical aspect ratio and attachment density reduction ................................ ................................ ............ 170 7 2 Attachment data from the first barnacle cyprid attachment experiment from July 2013. Bars represent the standard error ................................ .................. 176 7 3 Contact angle of te st surfaces post assay. Samples were gently rinsed with DI water prior to testing. Error bars are the standard deviation. ...................... 177 7 4 Attachment data from the second barnacle cyprid attachment ex periment from August 2013. Bars represent the standard error ................................ ..... 178
15 7 5 SEA model predictions for the 4 timepoints measured in the two B. amphitrite attachment assays. The top trend line is for Assay 1 and 48 hours Assay 2 and the bottom trendline is for 72 hours Assay 2 ................................ .............. 178 7 6 Average step distance for the barnacle cyprid during a track. Bars indicate standard error of available me asurements. Horizontal lines indicate statistically similar groups (Tukey test, =0.05) ................................ ................ 179 7 7 Step frequency of the barnacle cyprids on the different topographies. Bars indicate standard e rror of the measurements. ................................ .................. 180 7 8 Representative tracks of surface exploration by cyprids on a smooth T2 surface. ................................ ................................ ................................ ............. 181 7 9 Represen tative tracks of surface exploration by cyprids on the +7.3SK5x5_n8 surface. ................................ ................................ .................... 181 7 10 Representative tracks of surface exploration by cyprids on a +26SK20x20_n8 surface. ................................ ................................ ................. 182 7 11 Representative tracks of surface exploration by cyprids on a +100SK200x200_n8 surface. ................................ ................................ ........... 182 7 12 Representative velocity profiles of cyprids expl oring a) Smooth T2 b) 5 x 5 c) 20 x 20 and d) 200 x 200 SK_n8 surfaces. ................................ ....................... 183 7 13 Root mean square (RMS) radius of gyration for cyprid tracks. The error bars are the standard error of the availa ble measurements. Bars show statistically similar groups (Tukey test, =0.05) ................................ ................................ .. 184 8 1 Topography in PDMSe with a 500 nm undercut. ................................ .............. 194 A 1 Bright field and fluorescent images that could be analyzed with the MATLAB Â® program. The left image shows U. linza attached to the a5 topography and the right image shows the spores imaged with fluorescent imaging. ................ 19 6 B 1 Ulva attachment density as a function of casting surface on PDMSe. A and B are different images of the same slide showing the border between the two different smooth surfaces. C shows a fluorescent micrograph of the two surfaces. Figure courtesy of John Finlay. ................................ ........................ 201 B 2 Contact angle measurements from the two samples. Bars are the standard deviation of the measurements. ................................ ................................ ........ 202 B 3 Contact angle hysteresis on the two samples ................................ .................. 203 B 4 ATR FTIR spectra for control and retain PDMSe samples. .............................. 204
16 B 5 High strain (upper) and low strain (lower) modulus values for glass cast PDMSe films ................................ ................................ ................................ ..... 205 B 6 AFM images from the glass cast shoulders of retain slides from July 2013 ..... 206 B 7 AFM images from the wafer cast center of retain slides from July 2013 ........... 206 B 8 AFM height images from a smooth control retain fro m July 2013 ..................... 207 B 9 AFM height images from the smooth shoulders of retain sample from July 2010 ................................ ................................ ................................ ................. 207 B 10 AFM height images from th e wafer cast center of July 2010 retain samples .... 208 B 11 Measured captive air bubble contact angles on the wings and shoulders from a retain sample from July 2013. Bars indicate the standard d eviation of five measurements on the same sample. ................................ ................................ 209 B 12 Acrylate grafted silicone slide with a two layer wing region (analogous to topography mounting step) and one layer shoulder region.. The optical clarity is different between the wing and shoulder regions of the slide (indicated with arrows). Photo courtesy of Cary Kuliasha ................................ 210 C 1 Adhesion strength of MDCK cells to the Shar klet microtopography. The normalized cell density refers to the number of cells in the center of the sample where the applied shear stress was 0 Pa. ................................ ............ 214 D 1 Contact line geometry on the +2.6SK 2x2_n3 topography. This image shows the parallel edge of the topography ................................ ................................ .. 215 D 2 Contact line geometry on the +2.6SK2x2_n4 topography. This image shows the parallel edge of the topography ................................ ................................ .. 216 D 3 Contact line geometry on the +2.6SK2x2_n5 topography. This image shows the parallel edge of the topography ................................ ................................ .. 216 D 4 Contact li ne geometry on the +2.6SK2x2_n7 topography. This image shows the drop shape perturbed by the pattern ................................ .......................... 217 D 5 Contact line geometry on the +2.6SK2x2_n9 topography. This image shows the drop sh ape perturbed by the pattern ................................ .......................... 217 D 6 Contact line geometry on the +2.6CH2x2 topography. This image shows the drop following the topography into a smooth curve at the parallel edge ........... 218 D 7 Contact line geometry on the +2.6CH12x2 topography. This image shows an elongated drop following the topography ................................ .......................... 218
17 LIST OF ABBREVIATIONS AFM Atomic Force Microsc opy ATS Allyltrimethoxy silane B. amphitrite Balanus amphitrite C. marina Cobetia marina DLVO Derjaguin Landau Verwey Overbeek DMT Derjaguin Muller Toporov ERI Engineered Roughness Index H. elegans Hydroides elegans HMDS Hexamethyl disilazane JKR Johnson Kendall Roberts MDCK Madin Darby Canine Kidney N. incerta Navicula incerta PDMS Polydimethyl siloxane PEG Polyethylene glycol RDF Radial distribution function Re Reynolds number SEA Surface Energetics Attachment TPU Polyurethane U. linza U lva linza Probability of a Type 1 error in statistical analysis
18 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosoph y A PREDICTIVE MODEL FOR THE ANTIFOULING EFFICACY OF ENGINEERED MICROTOPOGRAPHIES By Joseph Thomas Decker August 2014 Chair: Anthony Brennan Major: Materials Science and Engineering We have developed a model, the Surface Ene rgetics Attachment (SEA) mod el that relates the work of adhesion for an organism to the probability of attachment. The model wa s used to predict the attachment density of four organisms to a variety of topographies from data available in the literature. The results showed the model was capable of predicting relative attachment density to a high degree of accuracy (R 2 = 0.83). Additionally, local effects of topographic configuration were identified and predicted by the model. An additional tool, the radial distribution function, was applied to zoospores of Ulva linza attached to topographies. This analysis was combined with previously developed mapping techniques to help identify local topographic configuration effects on Ulva attachment that may by missed the SEA model analysis. T he radial distribution function showed differences in R max and screening distance dependent on the location of the spore on the topography. Additionally, the contributory effects of a single spore extended only 20 m from the reference which indicated an ideal size for the topography.
19 Topographies were characterized for their adhesion properties through Atomic Force Microscopy (AFM) measureme nts and contact angle measurements. The AFM measurements showed site dependence for the work of adhesion for a collo idal probe on a wetted topography as predicted by the SEA model. The contact angle measurements configuration dependence for static, advancing and receding contact angle measurements for DI water. Bioassays with Ulva linza and Balanus amphitrite showed go od agreement with the SEA model predictions. A high throughput assay was developed to test the Ulva attachment and was successfully used to discriminate differences between 32 simultaneously tested patterns. A series of topographies ranging from 5 m t o 200 m in size were evaluated for the barnacle cyprid. These were found to inhibit attachment in line with the model predictions. Additionally, tracking experiments showed disruption of the cyprid surface probing dependent on the size of the topography .
20 CHAPTER 1 INTRODUCTION Scope of Research Marine biofouling is a significant issue in any industry that makes use of the oceans, lakes or rivers. Engineered microtopographies have recently been proposed as a nontoxic alternative to traditional biocida l methods of controlling marine fouling. Well controlled patterns have been shown to be able to control and prevent the attachment of a variety of marine fouling organisms. A number of strategies have been explored to predict fouling on microtopographies . These strategies are universally based on empirical observations of the attachment behavior on different surfaces and as such do not offer much in the way of insight into the mechanism through which topographies can prevent and influence biofouling. In this work, the relationship between engineered topography and cell attachment will be explored in a way that utilizes existing models to create a universal model for organism attachment to topographies. This model will then be used to further explore prev iously unexamined aspects of fouling on topographies and will utilize newly acquired insights to design and optimize antifouling topographies. Specific Aims Specific Aim 1. Develop a Mechanistic Model that C ombines the A ttachment Point Theory and ERI M od els for Fouling on T opographies There are currently two widely used models for cell attachment to topographies: the Attachment Point Theory and the Engineered Roughness Index. The parameters in these models suggest that they can be rationally combined t o yield a more mechanistic model for marine fouling. It is hypothesized that such a model can be created that will suggest new relationships between cell attachment and surface roughness and will also predict fouling density with a high degree of accuracy (R 2 > 0.80)
21 Specific Aim 2. Explore the Relationship Between the Spatial Distribution of Attachment S ites and the S patial D istribution of Attached Z oospores of Ulva linza Previous work by the Brennan group has indicated effective antifouling microtopog raphies inhibit the ability of zoospores of Ulva linza to attach in the preferred gregarious manner. It has also been suggested that the presence of a topography has an influence on the spatial distribution of marine fouling organisms, including the U. li nza. It has been hypothesized that the presence of preferential attachment sites, as predicted by the Attachment Point Theory and measured experimentally may contribute to the antifouling character of certain topographies, as well as cause a measurable ch ange in spatial distribution of the attached zoospores. This aim will examine the local effects of the topography on spatial distribution of U. linza in order to gain further insight into how the configuration of the topography influences cell attachment and also to investigate the hypothesis regarding topographic configuration and spatial distribution. Specific Aim 3. Examine the W ork of Adhesion on E ngineered Microtopographies Using Contact A ngle and AFM The work of adhesion is directly related to the attachment of biofouling organisms through the SEA model. One of the assumptions when calculating the work of adhesion for the organisms on the topography is that the area for attachment is the driving force for the differences between the patterned surf aces. Contact angle measurements can be used to measure the work of adhesion on the topographies. The work of adhesion will be measured on the topographies using contact angle to verify the assumptions used in the SEA model calculations . The SEA model a lso predicts local differences in the attachment of organisms based on the local configuration of the topography. AFM can be used to probe these
22 differences and relate the measured adhesive forces back to the SEA model predictions. These measurements wil l be performed to provide measurable confirmation of the SEA model predictions. S pecific Aim 4. Systematically Probe the Relationship Between Topographic Dimensions and C onfigu ration on the Behavior of Both Microfouling and Macrofouling O rganisms The ra nge of marine fouling organisms make predicting their behavior on engineering topographies very difficult. In particular the difference between unicellular microfouling organisms and multicellular macrofouling organisms is substantial. Previous studies h ave indicated that there is a need for species specific topography, and that a combination of two different topographic sizes is not effective at preventing fouling. However, microtopographies have been used to effectively deter both macrofouling and micr ofouling under specific circumstances . Microfouling in the marine environment, particularly in terms of U. linza , has been studied extensively in terms of attachment to topographies. These results show a relationship between topographic configuration and attachment and have identified several promising antifouling technologies. However, the current method of screening topographies is cumbersome, and the ability of these assays to quickly probe several different aspects of the topography for their relevan ce to antifouling is lacking. A study is therefore proposed to create a method of quickly screening these topographies and also to probe new aspects of the topography and compare these results to any relevant model predictions. Macrofouling has also been examined with the barnacle cyprid of Balanus amphitrite as a model organism. The complexity of the cyprid suggests that there may be a more complicated relationship between the topography and attachment. Therefore
23 we will examine not only the attachment density of this organism as a function of topography size and configuration but also the manner in which the organism probes these different surfaces. These two pieces of information combined will yield important new insight into the manner through which topography influences cyprid attachment.
24 CHAPTER 2 BACKGROUND Issues Associated with M arine F ouling Biofouling is the accumulation of unwanted organic material on a surface. The buildup of organic material, from the molecular level up to un icellular and finally multicellular organisms, inevitably lea ds to the failure of the device. In the marine environment, failure includes loss of efficiency for naval vessels, transportation and introduction of invasive species, reduced aquaculture output as well as other serious environmental and economic consequences. The impacts of biofouling have le d to considerable efforts to develop an understanding of the mechanism of the biofouling process in order to design new materials for antifouling coatings. Marine fouling impacts any industry that utilizes rivers, lakes or oceans for operation. These include naval and shipping vessels, aquaculture, power generation and others. Each industry faces unique issues specifically related to biofouling that lead to increased operation costs and environmental concerns. Ship hulls present an enormous amount of surface area, approximately 438 million m 2 in the United States alone 1 . Fouling organisms have a measurable effect on fuel efficiency due to increases in drag, leading to significant increases in fuel consumption and costs associated with hull cleaning and biofouling prevention 2,3 . Schultz et al . 4 have estimated that fouling from algal slime can increase resistance by 20%, and that he avy barnacle fouling can increase resistance up to 80% for heavily fouled surface. The increase in resistance due to fouling on the Arleigh Burke class destroyer DDG 51 contributes $56 million in additional costs to the navy for issues related to biofouli ng, totaling $1 billion over the next 15 years 2 . The cost is considerably
25 higher when examining the entire fleet, as the DDG 51 comprises only 20% of the naval fleet. Ship hull f ouling also presents a significant issue in terms of the transfer of invasive species. An overwhelming majority of ship hulls (up to 96%) have biofouling from species not native to the port in which they are docked, with an additional 38% of ballast water and 57% of sediment samples also showing non native species 5 . Invasive species transport has been documented as a function of shipping lanes as well as static device fouling that results by chance 6,7 . For example, a dry dock sample that arrived in Hawaii from the Philippines in 1992 has resulted in a significant infestation of Chama macerophylla and G elliodes fibrosa Â¸ a moll usk and sponge respectively that have invaded Hawaiian ports 7 . Invasive species are also a significant issue in freshwater ports as evidenced by the invasion of zebra mussels in the Grea t Lakes region 8,9 . The economic and environmental impact is difficult to measure however the scale of the problem, and the contributions from biofouling to this issue, are plainly seen anywhere in the world. Marine vessels are at an advant age in terms of reducing fouling since they are mobile for some of their usage time. The motion of the vessel provides a natural mechanism to shear weakly adhered organisms from the hull. Static surfaces which must remain functional present a considerable extra challenge where biofouling is concerned. Aquaculture presents an important component to the antifouling equation. Approximately 50% of all seafood is produced through aquaculture 10 , clearly indicating the importance of maintaining a healthy environment in this industry. However, fouling of aquaculture nets and cages is one of the most serious problems in t he industry and is
26 a key contributor to production costs. Approximately 5 10% of all costs associated with bringing aquaculture products to market arise from biofouling related issues, and can be as high as 15% for certain types of organisms such as shell fish. These additional costs total $1.5 3 billion annually 11,12 . Another example of fouling is in the field of power generation . Heat exchangers are often placed underwater to increase efficiency, however the high temperatures associated with these dev ices make them excellent locations to attract biofouling. T he flow of water over the surface in t hese system s imparts a natural barrier to fouling . However, colonization is inevitable for these devices, and efficiency losses from biofouling of heat excha ngers costs approximately $2 million per year for a single 550 MW power plant. Considering there are approximately 7,000 power plants in the United States alone the problem of biofouling in heat exchangers has a significant economic impact 13 15 . Model Systems for Marine F ouling A major challenge in the design of antifouling surfaces is the diversity of fouling organisms in the marine environment. Fouling in the marine environment is a dynamic process that spans sev eral orders of magnitude in terms of both the size of the fouler and the time in which a certain species can be expected to colonize a surface ( Figure 2 1 ). Substantial remodeling of a surface can occur between i nitial fouling of organic molecules and final fouling by hard macrofoulers. As a result, several systems have been selected as models for marine fouling. These models are divided into three categories based on temporal and size scale: organic molecules, microfouling (bacteria, algae), and macrofouling (barnacles, tubeworms).
27 Figure 2 1 . Hierarchy of model marine fouling organisms. Organic fouling Organic molecule fouling occurs very quickly after submersion of a material, us ually in a much shorter timespan than that of fouling organisms 16 . These molecules represent both the shortest time scale and smallest species related to fouling. Proteins and other organic material adsorb nonspecifically and can lead to radically diffe rent conditioning layers depending on the surface and the molecule content in the solution. As such, protein models such as fibronectin and bovine serum albumin are commonly used as models for organic fouling 17 21 . Fouling from small organic molecules can lead to surface properties that are different from the properties measured before submersion. The subsequent fouling of organisms on a surface is, in part, driven by the initial organic fouling. Therefore, it is imperative to take organic fouling into account when evaluating antifouling coatings. Microfouling Microfouling is attachment and colonization on a surface by organisms that are relatively small, on the order of one cell. Microfouling organisms consist of bacteria, algae and diatoms, which make up the initial layer present in marine biofouling ( Figure
28 2 1 ). As such, the microfouling layer can have the effect of either enhancing or inhibiting fouling by larger o rganisms 22 25 . The US Navy and various marine biology research groups use a number of model microfouling organisms as indicators of the efficacy of novel antifouling and fouling release coatings. T hree of these o rganisms, the bacteria Cobetia marina, the algae Ulva linza and the diatom Navicula incerta , are discussed here . These three organisms are in the 1 to range and are among the most prevalent fouling organisms in the marine environment 26 . Cobetia marina is a gram negative marine bacteria that has been extensively studied as a model for marine bacteria fouling 27 29 . These bacteria settl e out of the water column and attach to a surface, at which point they proliferate to form a biofilm 28 . C. marina is tested in a flow cell to better mimic the conditions under which they would attach to a surface 28 . The proliferation of the bacteria allows the C. marina to extensively remodel a surface after coloniza tion, which must be taken into account when testing a material . Ulva linza is the one of the most prevalent fouling algae for ocean faring vessels. These organisms attach to a surface as flagellate zoospores and as such will actively preferential . Preferential sites can be, for example, channels that have dimensions similar to the body of the spore ( Figure 2 2 ) . Callow et al. 31 showed that zoospores of U. linza generally sample several sites before forming permanent attac hments to a surface. This surface sampling takes place over several hours in culture, during which the density of spores on the surface increases following an Arrhenius relationship 31,32 .
29 Figure 2 2. Zoospores of Ulva linza preferentially attach in the channels of a 5x5 um topography . [Reprinted with permission from Hoipkemeier Wilson, L. et al . Antifouling potential of lubricious, micro engineered, PDMS elastomers agai nst zoospores of the green fouling alga Ulva (Enteromorpha). Biofouling 20 , 53 63 (2004) ] Figure 2 3 . Relationship between assay time and spore attachment density on smooth and topographically modified surfaces. [ Reprinted with permission from Cooper, S. P. et al . Engineered antifouling microtopographies: kinetic analysis of the attachment of zoospores of the green alga Ulva to silicone elastomers. Biofouling 27 , 881 892 (2011 ) ] Zoos pores of U. linza tend to settle gregariously, with up to 80% of spores attaching in groups of two or more 31 33 . Attachment occurs with the secretion of a hydrophilic adhesive composed of polysaccharides that will have different degrees of spreading dependent on the wettabi lity of the substrate 34,35 . Unlike C. marina, U. linza
30 do not proliferate once attached to surface. Instead, m ature sporelings grow from the attached zoospores which leads to the familiar algae plant. Navicula incerta , like the C. marina , falls through the water column to a surface, leading to a random attachment pattern ( F igure 2 4 ). Once attached the diatoms adhere using a polysaccharide adhesive, will proliferate in addition to moving along the surface 36 . The Navicula are then mobile on the surface, leading potentially to remodeling. Most strategies for controlling N. incerta are focused therefore on preventing a strong bond, i.e. fouling release materials, as it is almost impossible to keep N. incerta off the surface in a reasonable fashion. Figure 2 4 . Initial attachment of Navicula incerta with random orientation and distribution . Macrofouling Macrofouling organisms are in general several orders of magnitude lar ger than microfouling organisms. Macrofoulers are multicellular, and will often have several different stages in life and can be significantly more complicated than their microfouling counterparts. The most studied model organisms are the barnacle cyprid Balanus amphitrite 3,24,37,38 and the tubeworm Hydroides elegans 22,23,25,39 , however many groups have examined other types of organisms as well 40,41 .
31 The barnacle cyprids are much more complex than the unicellular algae and bacteria species discussed previously. B. amphitrite is a multicellular barnacle species that is capable of probing a surface and actively discriminating between attachment sites. The cyprid has a foot which it uses for both motion on a surface as well as sensory information. The foot appendage of the cyprid is able to probe the surface for complex physical and chemical cues. The foot is an extremely complex organ comprised of many sensory components at various size scales. These cyprids will also probe a surface either by searchi ng using a foot like appendage or by swimming near the surface ( Figure 2 5 ). This searching behavior, and the surface characteristics that lead to a preferential attachment surface have been studied extensively 42 . T he cyprid state attaches directly to the surface, at which point it metamorphoses into a mature barnacle 43 . The metamorphosed barnacle presents a significant obstacle and must be addressed in the d esign of antifouling materials. Figure 2 5 . Foot appendage of the cyprid of Balanus amphitrite. Images A D are different views of the foot. [ Reprinted with per mission from Maruzzo, D., Conlan, S., Aldred, N., Clare, A. S. & HÃ¸eg, J. T. Video observation of surface exp loration in cyprids of Balanus amphitrite: the movements of antennular sensory setae. Biofouling 27 , 225 239 (2011)].
32 The tubeworm H. elegans is also used as a model organism for marine macrofouling. The tubeworm is notoriously difficult to target with nontoxic antifouling technologies. This is partly due to the fact that many tubeworm species, including H. elegans , require a bacterial biofilm in order to attach to a surface 23 . There is no known way to target the se tubeworm s directly with nontoxic strategies and as such any effective strategy must be focused on the bacteria microfouling. The mechanism through which the bacteria aids in the attachment of the tubeworm was only recently discovered, whi ch will likely lead to more specified antifouling technologies in the future 44 . Design of nontoxic antifouling coatings T raditional antifouling strategies have focused on the use of biocides to prevent attachment. Copper and tin are often used as antifouling compounds in the marine environm ent 16,26 . However, m any concerns related to environmental issues have led to a decline in the use of tox ic antifouling strategies. T ributyltin from antifouling paints has been observed to buildup in harbors where many ships are docked. Tributyltin has also been detected in large concentrations in indigenous marine life that was the unintended target of leaching antifoulants, causing significant damage to the environment 45 48 . The environmental concerns surrounding the use of tributyltin led to a ban on its use in antifouling coatings , with a ban on copper biocides likely to follow 49 . As a result, a demand has arise n for environmental neutral coatings for antifouling and fouling release. The design of nontoxic antifouling coatings requires careful consideration of the properties of the organism surface interface as well as the forces involved in the approach and atta chment to the surface. The ability for an organism to attach to a
33 surface depends on the attachment mechanism of the organism, the properties of the surface and the properties the surrounding media. As a result colloidal interactions, fracture mechanics, surface energy and surface roughness all must be considered when evaluating the potential for a surface to be used as an antifouling coating. Colloidal interactions A collection of unicellular organisms in solution, prior to attachment, resembles a colloi dal suspension. Some authors have attempted to model the attachment of fouling organisms to a surface using theories for colloid stability, namely the Derjaguin Landau Verwey Overbeek (DLVO) theory 50 54 . The DLVO theory evaluates the interaction energy between colloidal particles and can predict whether a colloid will aggregate or remain dispersed in solution 55 interactions as well as electrostatic forces aris ing from ions in the solution to calculate the total energy of interaction E tot ( Equation 2 1 ) . ( 2 1 ) In Equation 2 1 , E VdW refers to contributions fr om Van der Waals forces to the total energy and E ES refers to electrostatic interactions. The exact values of the Van der Waals and electrostatic interactions for a system can be calculated through the geometry, change and solution properties of the syst em. The Van der Waals forces in particular are impacted by geometry. Exact solutions for the Van der Waals interaction force have been solved for many simple geometries for the interacting species. Particularly relevant for the case of cells interacting with surfaces is the sphere plate geometry. The organism can be thought of as a sphere and the surface as a plate, thus allowing the use of the simple relationship between the two when calculating the Van
34 der Waals force. The Van der Waals force arise f rom dipoles, either induced or permanent, in the materials. These dipoles interact with each other at very short distances, leading to either attraction or repulsion depending on the interacting materials. It should be noted that similar materials should always have attractive Van der Waals interactions, however this does not necessarily have to be the case for dissimilar materials 55 . The E VdW parameter for this geometry is given by the radius of the sphere R , the distance between the sphere and the plat e D and the Hamaker constant for interaction between the two materials A ( Equation 2 2 ) . ( 2 2 ) The second term in Equation 2 1 , the electrostatic interaction te rm, arises from the so charged ions. A surface has an inherent charge to it oppositely charged ions present in a solution will congregate at the surface to neutralize that char ge. The properties of the double layer, as well as the magnitude of the forces generated from the charge, are a function of the surface properties, the geometry of the solution and the properties of the ions in solutions. The electrostatic forces from a sphere interacting with a flat plate are given by Equation 2 3. ( 2 3 ) Equation 2 3 includes two terms to describe the interaction energy between the particle and the surface, Z and . The Z term is analogous to the Hamaker constant in the in Van der Waals energy equation. This term is a function of the vacuum permittivity o , the relative permittivity , the temperature T , the valence of the interacting ions z , the surface potential o , and the number of charged species on a surface e ( 2 4 )
35 ( 2 4 ) The term in E quation 2 3 is dependent on the properties of the solution. The term 1/ ( corresponding to the exponent in E quation 2 3 ) is commonly referred to as the Debye length and corresponds to the characteristic length at which the e lectrostatic forces are expected to decay. The Debye length is then only dependent on the temperature of the solution and the ratio between cations and anions. It is possible to calculate the Debye length from the terms in E quation 2 4 . Since E quation 2 4 puts these constant terms in to the Z parameter, the calculation for the Debye length is much simpler. For a 1. 1 cation to anion solution (NaCl for example), the relatio nship is given by E quation 2 5 55 ( 2 5 ) The Debye length can also be calculated from the parameters used to calculate the Z parameter. In this case, the relationship is Equation 2 6. ( 2 6 ) The DLVO interactions between a cell and a surface can be difficult to quantify due to the inherent diversity within a cell population. As a result, cell attachment results predicted from the DLVO theor y did not always repres ent ex perimental realities. Redman et al . 56 reported that, when the barrier for attachment of Escherichia coli ( E. coli ) was too high theoretically , attachment to the surface still occurred . They state d the cause was a secondary minimum in the free energy curve at a greater distance which et al . 52
36 indicate d that specific interactions of secondary ions cause the observed deviations from the DLVO theory predictions . Still other authors have shown that DLVO theory and extended DLVO theory (including surface energy arguments) can be used without any extra explanations for bacterial fouling 53,54 . Hermasson 51 provide d an excellent review of examples in which foulin g is predicted by the DLVO and extended DLVO theo ries . The mixed results regarding the use of the DLVO theory to predict cell attachment indicate that the interactions used to calculate the energy for attachment may not include all possible interactions between the organism and the surface and that org anism specific effects may need to be factored into the final calculations. F racture m echanics Biofouling can be prevented by either eliminating the bond between the organism and the surface or by limiting the adhesion strength of the organism after attach ment. Fouling release coatings focus on the latter strategy . The weak bond between the organism and the fouling release surface leads to a low critical stress for detachment. The shear stress to remove the organism is then applied during the normal use of the device, such as when a naval vessel is in motion . Currently, fouling release coatings based are the only commercially available coatings that are considered non toxic. Companies such as International Paint have developed silicone based paints that have been shown to be effective fouling release coatings for some organisms 57 . Several theories based in fracture mechanics have been used to explain and predict fouling release behavior. Brady performed extensive early work on marine fouling organisms 58 . His work was based in Griffith crack c riterion . The Griffith crack criterion relates the energy required to form two new surface s with the modulus of the material . The energy required to form a new surface is equivalent to the surface
37 energy 59 . The critical stress to propagate a crack is related to the elastic modulus E , the Poisson ratio o f the material , the size of the crack a and the fracture energy G c ( Equation 2 7 ) . In real materials, particularly soft materials such as plastics, viscoelastic effects at the edges of the crack can cause the fracture energy t o be much higher than the G c predicted by surface energy arguments 60 . ( 2 7 ) Soft materials, such as silicone elastomers with elastic moduli of 1 2 MPa, are often used as fouling release coatings. In the case of fracture of the bond between an organism and a so ft material it is necessary to examine the pull off force necessary to break the adhesive bond. Kendall 61 derived an equation for the critical stress for removal of a disk from a thin elastomeric film. This relation depends on the thickness of the film t , the bulk modulus E and the fracture energy G c ( Equation 2 8 ) . ( 2 8 ) Kendall also examined the relationship between thick elastomeric films and the pull off force of a disk. In this case the relationship does not depend on the thickness and is only a function of the material properties of the interface and elastomer and the ation for critical stress ( Equation 2 9 ) . ( 2 9 )
38 Brady and Singer used these relationships to define an empirical relationship between surface mechanical properties, surface energy, and the likelihood of fouling release. Specificall y, the E and G c terms were used as the defining terms for the critical removal stress as they are the driving terms in the Kendall relation. The fracture energy can be further related to the surface energy which allows for the relation between surface p roperties, modulus and the effectiveness of a coating for fouling release 58,62 . Chaudhury et al . would later extend this work to applications of soft elastomers and relating the coating thickness to the likelihood of fracture 63 . This study found that increasing the elastic modulus of a silicone elastomer decreased the percentage of spores and sporelings of U. linza that were remove d at 55 Pa shear stress. Additionally, increasing the thickness of the elastomer coating from 16 m to 100 m increased the percentage of spores and sporelings removed at 55 Pa shear stress. Both of these observations agree with Equation 2 8 and Equation 2 9 above, highlighting the effectiveness of the fracture mechanics model for organism release. Unsurprisingly , based on the fracture mechanics analysis for low adhesion materials, the most effective fouling relea se surfaces tend to be low surface energy , low modulus materials . The Wooley lab has developed combinations of fluoronated and PEG based polymers that have shown excellent fouling release and antifouling properties due to their incorporation of both antif ouling and fouling release components. The Ober lab has taken a similar approach with block copolymers 18 . Another method of creating fouling release materials is t o create an active coating. These are materials that change their surface properties in an attempt to remove any adhered organisms from the surface. The change in surface properties can
39 come in a number of forms. Some authors, such as the Aizenberg grou p, have utilized solvent filled materials to promote fouling release on their surfaces 64 . The organisms only have a liquid surface to attach to for this type of surface. The surface therefore has an extremely low E and thus would make for an excellent fouling release surface. The material developed by the Aizenberg group , which they have named SLIPS (Slippery Liquid Infused Porous Surfaces) infuses a perfluoronated oil into a Teflon matrix, creating a low surface energy surface that essentially consists of the low surface energy o il that can be replenished if damaged. The SLIPS surface has shown to be extremely effective against both pathogenic bacteria ( Pseudomonas aeru ginosa, Escherichia coli and Staphylococcus aureus ) 65 as well as the previously discussed U. linza and B. amphitrite 66 . Active fouling release coatings also include surfaces that dynamically change topography in order to break the organism surface bond. The surface will provide the stress necessary to break the or ganism surface bond and therefore act as a fouling release surface. Zhao used both pneumatic and electric field stimulation to cause predictable deformations in soft materials (such as silicone) to remove bacteria from the surface 67 . This technology has been demonstrated effective in a urinary catheter model with a mature biofilm of Proteus mirabilis 68 . The Balazs group recently demonstrated theoretical viability of actuated surfaces to prevent attachment and remove d attached cells 69 71 . The simulations in these works refers to swimming bacteria and algae and describe the potential effectiveness of ciliated materials, with the cilia either actively moving to repel the materi als or passively moving with the current of the water.
40 However, these materials have yet to be demonstrated effectively in an experimental bioassay. Surface Energy The surface energy of a material is the excess energy resulting from the interface betwe en the material and some other medium such as air or a liquid. The energy necessary to form a bond between an organism and a surface can be described by the work of adhesion W adh . The work of adhesion is a function of the surface energy of the each compo nent of the system ( 1 and 2 ) and the interfacial energy between the two adhering surfaces 12 ( Equation 2 10 ) ( 2 10 ) Equation 2 10 gives a measure of the effect of substrate surface energy on the energy necessary for a cell to adhere to the surface. The most common method for measuring surface energy is the use of contact angle measurements. A contact angle experiment measures the angle between a drop of liquid and a surface in some surrounding medium ( Figure 2 6 ) . A typical experiment will use a water drop on a surface in air, although variations including other liquids such as ethyle ne glycol and interfacial energy between the solid and gas sg , the solid and the liquid sl , and the liquid and gas lg as well as the contact angle formed betwe en the drop and the surface ( Equation 2 11 ) ( 2 11 )
41 Figure 2 6 . Contact angle for a liquid drop on a smooth surface Equat ion 2 11 is valid under the assumption that there are no other forces acting smooth surface. Differences between the two s ides of the drop caused by changes in topography or tilting of the drop are not addressed. Additionally, the interfacial energies in Equation 2 11 do not attempt to elucidate the relative contributions from different forces on the overall energy. The rela tive contribution from different forces to the surface energy measured by Equation 2 11 have been explored . Owens and Wendt 72 introduced the concept of a composite surface energy that includes contributions from both dispersive ( d ) and polar ( p ) interactions ( Equation 2 12 ) . ( 2 12 ) The values for the polar and dispersive components of the surface energy are related to the contact angle of a liquid with the surface much like in Equation 2 11 . In the case of the Owens Wendt method, the surface energy values are obtai ned by using two references liquids such as water and methylene iodide. Water and methylene iodide are used as references because they are well characterized in terms of their surface tension. Methylene iodide provides a reference with no contributions f rom the
42 polar component of the surface energy and water provides a reference with well characterized polar interactions. Van Oss and coworkers 73 fu rther broke down the polar component of the surface energy into Lewis acid ( + ) and Lewis base ( ) interactions, and attributed the dispersive component of the surface energy to Lifshitz Van der Waals forces ( LW ) 73 ( Equation 2 13 ) . ( 2 13 ) Again, the different components of the surface energy are related to the measured contact angle on the surface. The separation of the components of the surface energy allows for the complete calculate of the energy involved in wetting the surface, and can help predict the relative contributions of the different molecules and charges on a surface towards the hydrophobicity. Much like the Owens Wendt method, the Van Oss method requires multiple contact angle measurements with well characterized liquids. This usually involves adding glycerol or ethylene glycol to the two Owens Wendt liquids as a third well characterized probe liqui d. The relationship between surface energy and biofouling has been extensively studied , both in terms of empirically relating surface energy to attachment and attempting to calculate the thermodynamics of adhesion in terms of surface energy 17,27,35,74 86 . The most direct method is to calculate and relate the work of adhesion for different organisms to the observed biofouling. Absolom et al . 75 were among the first to quantitatively relate surface energy, and by extension surface thermodynamics, to bacterial attachment. They found good correlation between the surface energ y of a polymeric substrate and the number of attached bacteria after 30
43 minutes of culture. This study related the surface energy of the bacteria to the surface energy of the material, and used the work of adhesion W adh to predict whether or not attachmen t would occur. The Busscher lab later expanded on the work on surface energy related bioattachment 77,83 91 . Their work has included extensive analysis of the surface properties of microorganisms, in particular how the properties of the bacteria cell wall influence attachment. The Busscher group also attempted to identify the contributing factors in terms of force as they relate to bacterial adhesi on 86 . This analysis combined the calculations from the DLVO theory as well as surface thermodynamics to describe the interactions between the bacteria and the surface in terms of adhesive streng th. One of the drawbacks to the work of adhesion method is that it requires the calculation of the W adh term for every organism. This is possible for a laboratory assay but is not feasible for a surface that is deployed in the field due to the number o f different potential fouling organisms. As a result, empirical relationships between substrate surface energy and biofouling have been developed. Hydro phobic surfaces (surfaces with low surface energy ) are conventionally thought to have the best antifou ling properties 92,93 . An empirical relationship, dubbed the Baier Curve, was drafted based upon fouling retention of proteins and cells on materials that range from hydrophobic to hydrophilic. It attempted to set a minimum for fouling at a surface ene rgy of approximately 22 mN/m, which corresponds with the surface energy of silicone 74 .
44 Figure 2 7 . Original Baier curve showing an empirical relationship between critical surface tension (x axis) and relative biological interaction (y axis). [ Reprinted with permission fr om Baier, R. E. The role of surface energy in thrombogenesis. Bulletin of the New York Academy of Medicine 48 , 257 (1972)]. The relationship between fouling and surface energy described by the Baier curve can be used as a general guide for designing antifouling surfaces. However, the curve must be recalibrated depending on the organisms u nder investigation. Ista 27,94 and Bennet 95 both systematically showed that attachment density of U. linza increased as the surface tension of the substrate was lowered. Conversely, Finlay 17 showed a relationship similar to the Baier curve in terms of adhesive st rength of N. incerta and bovine serum albumin to xerogel surfaces with varying surface tensions. The specific interactions between fouling organisms and surfaces mean the relationship between surface tension and fouling density must be calibrated for each individual species.
45 The Baier curve showed the inability to predict the biofouling response for very hydrophilic, approaching the surface energy of water at 72 mN/m. These hydrophilic materials include polymers such as polyethylene glycol (PEG) that are used as the basis for hydrogels. These very hydrophilic materials tend to be effective antifouling materials for the entire range of organic fouling species, provided the surface has a neutral charge 26 . As a result many of the new nontoxic antifouling coatings are based on these surfaces. The ability for hydrophilic surfaces to interact with water appears to be the primary reason for their ability to resist fouling. These are so calle described by Rosenhahn et al. 96 . A systematic study by Grunze et al . 19 showed a relationship between the size of self assembled monolayers (SAMs) of PEG oligomers and their ability to resist fouling, with a minimum size necessary to prevent attachment. They also showed that density of their SAMs directly a ffected the amount of fouling present on the surface. Jiang carried out a similar study with sulfobetaine methacrylate ( SBMA ) based polymers. In that study, the distance between charges on the zwitterionic molecule was determined to be the factor most re levant for the prevention of fouling 20 . Theoretical studies by Grunze and others have shown that the ability for the water to form hydrogen bonds with the surface is of paramount importance to preventing fouling 19,21 . If the chains are too dense only a few molecules are associat ed with the surface, which leads to poor antifouling performance; the same issue is present with chains that are too short. Striking the proper balance is essential to performance for these materials.
46 The Wooley lab has achieved the balance through the creation of hyperbranched polymers with both fluoronated and PEG groups 97,98 . The most effective of these coatings, which was a combination of hyperbranched fluoropolymer with 45% PEG by weight, showed reduction of a number of fouling proteins. It also demonstrated the ability to inhibit the attachment of U. linza , decreasing the attachment density by approximately 85% relative to a glass slide control. The Jiang group has employed SBMA, which has both positivel y and negatively charged components within the monomer, as an antifouling material 99,100 . SBMA polymers had low attachment with U. linz a and N. incerta and also inhibited the movement of the N. incerta on the surface in laboratory assays. Additionally, settlement of B. amphitrite was completely prevented during the course of a 72 hour laboratory assay 101 which showed that SB MA is effective against a variety of organisms. The Walker lab took still another approach by creating block copolymers in which one block has a hydrophilic domain and another a hydrophobic domain 18,102 . These blo ck copolymer systems combine a block of hydrophobic polystyrene with hydrophilic vinylpy r rolidone. The concept is that the hydrophilic blocks will provide antifouling performance while the hydrophobic blocks will provide mechanical stability. The phase s eparated materials showed resistance to fouling by both proteins and U. linza . Interestingly, PS b PMMA materials, which are not expected to be antifouling on their own also showed resistance to attachment of U. linza . It is unclear exactly why this shoul d be the case but is an interesting result nonetheless. The Brennan lab has examined the potential use of hydrogels as antifouling materials with promising results 103 . These hydrogels were copolymerized PEG -
47 methacrylate compounds with different chemistries. The hydrogels were shown to be effective at prev enting attachment and facilitating release of U. linza, N. incerta and C. marina . When combined with the Sharklet microtopography attachment of U. linza was reduced up to 97% for the PEGDMA co GMA formulation. Finally, the Langer group performed a systema tic examination of a variety of different monomers in a combinatorial analysis of many different combinations of properties. They found that Lewis base containing monomers showed the highest resistance to protein fouling 104 . This approach has also been used more recently by other labs, showing similar results. All these ex amples point to the effectiveness of highly hydrophilic surfaces to control fouling. Surface Roughness Microscale roughness can be utilized to change the wettability of a surface and has been shown to predictably change the observed contact angle 105,106 . The change in contact angle is similar to the change observed between hydrophilic and hydrophobic materials. Textured surfaces with high contact angle and can have contact angles that approach 180 o on certain surfaces. The increase in contact angle in this case does not directly relate to the work of adhesion as it would for a smooth surface. However changes in roughness do have an effect on the adhesion on liquids and cells. The topography on the surface leads to t wo different hydrophobic wetting conditions. The first, Wenzel state wetting 105 , is a fully wetted surface that has a higher apparent contact angle than a s mooth surface ( Figure 2 8 , Equation 2 14 ). The roughness factor r , defined as the ration between the total surface area to the projected
48 surface area, is applied to the smooth contact angle, resulting in a higher apparent angle . ( 2 14 ) Wenzel initially developed Equation 2 14 as an empirical relationship between surface roughness and observed contact angle, however despite its empirical nature this relationship has been demo nstrated to accurately describe experimental data. Figure 2 8 . Schematic of the Wenzel wetting state in which the entire topography (tops and side walls) are wetted. A second , more hydrophobic condition first described by Cass ie and Baxter 106 is a composite interface. Cassie Ba xter wetting predicts a liquid is more stable if it does not wet the extra surface area of the side walls of the pattern ( Figure 2 9 ). Figure 2 9 . Schematic of the Cassie Baxter wetting state. The result is a composite air water surface interface that has a higher apparent contact angle than a smooth surface. The contact angle for a dr op in the composite s and the initial contact angle of the material ( Equation 2 15 ) .
49 ( 2 15 ) A drop will assume either the Wenzel or Cassie Baxter wetting condition on topographically modified surfaces with a high contact angle on smooth depending on whichever state is more thermodynamically stable. The critical contact angle for a smooth material to determine between the two wetting conditions is a balance between the contribution of the extra surface area from the feature side walls (denoted by the r term) and the stability of the drop in the composite state (denoted b y the s term) ( Equation 2 16 ) . ( 2 16 ) The design of the topography therefore dictates the wetting state of the material. The Wenzel and Cassie Baxter wetting conditions can have very different properties in terms of contact angle hysteresis and static contact angle, both of which have been shown to have an effect on biofouling 107 109 , and it is therefore important to take the expected wetting state into consideration when applying a topography to a surface . The most simplistic attempt to explain the effect of topography on cell attachment in the idea of attachment point theory. The attachment point theory postulates that organisms tend to attach to surfaces, and locations on surfaces, where they can form the most attachment points. The concept is that organisms will prefer to attach in locations where they will be protected from predators or removal from current and are therefore more likely to attach in areas where they can form many attachment points. This theory grew from the observations that fouling organisms tend to attach to locations where they can form the greatest number of contact points 110 . Callow 33
50 extended this concept to well defined microtopographies where it was also observed that U. l inza preferred to attach in locations where they could form the greatest number of contacts on topographies that increased attachment density, particularly in terms of contact area. Long 111 and Cooper 32 later showed that the concep t of attachment points extended to antifouling topographies. Ulva prefers to attach in the intersections between diamond patterns in the Sharklet topography, including both initially attached zoospores and aggregates of attached spores 32,111 . Figure 2 10 . Map of individually attached Ulva zoospores relative to the Sharklet unit cell Scardino et al . 113 showed that the number of attachment points was insufficient to predict a relative change in fo uling for the diatom Amphora sp., the green alga Ulva rigida , and the red alga Centroceras clavulatum ; in interestingly, it was effective for other organisms, the tubeworm Hydroides elegans and the bryozoan Bugula neritina .
51 These patterns were channel top ographies where the number of contact points were difficult to predict and often differed based on the orientation of the organism, thus limiting the ability for the attachment point theory to be used in this context. Several studies have examined the conc ept of entrapped air in the topography as a potential source of antifouling. The Lamb group generated hierarchical nanotopography which was very effective at preventing fouling 114 . Small angle x ray scattering studies show a difference between fully wetted and partially wetted surfaces, which the authors attribute to their topography. Their hypothesis is that the lack of wetting prevent s attachment however the published work on their surfaces is lacking in concrete evidence. Friedlander et al . 115 has attempted to address the issue of non wetting topographies for bacterial fouling 115 . Their study found that flagella were necessary to attach to a topography. The authors ability to wet the surface and therefore provide attachment points. The non wetting surface is extremely difficult to quantify during cell culture however, and remains an intriguing source for potential antifoulin g technologies, particularly those that utilize topography. The Brennan lab create d a predictive model for antifouling topographies based on the surface roughness of the topography. This model, called the Engineered Roughness Index (ERI) , combined aspects of the Wenzel and Cassie Baxter theories for wetting of a topography and applied them to organism attachment. The first iteration was based on four dissimilar topographies: channels, pillars, triangles and pillars, and Sharklet. The model combined the r oughness r s value for the topography as well as a degree of freedom term to describe the configuration.
52 Figure 2 11 . Plot of attachment density v. ERI I [Reprinted with permission from Schumacher, J. F. et al . Engineered antifouling microtopographies effect of feature size, geometry, and roughness on settlement of zoospores of the green alga Ulva. Biofouling 23 , 55 62 (2007)] The end result was a model that effectively predicted results of the published assay ( Equation 2 17 ) . ( 2 17 ) Further testing of the ERI model demonstrated that it failed for topographies that were designed to present a force gradient to attaching organisms 117 . Long et al . 117 addressed this issue by introducing a new term to the model, n , which represents the number of features in the model ( Equation 2 18 ) . This term replaced the degrees of freedom term in the first ERI model and serves the function of adding a factor to explain the contributions from the topographic configuration ( Figure 2 12 ) . ( 2 18 ) The second ERI model also introduced a normalized attachment density A/A o , where the attachment density on the topographically modified surface was normalized against the attachment density on a smooth reference surface. The attachment density for fouling organisms fluctuates sign ificantly depending on the season. Normalizing the
53 attachment density allowed several studies worth of data to be compared simultaneously on a single plot. Figure 2 12 . Second version of the ERI [Reprinted with permission from Long, C. J. et al . A model that predicts the attachme nt behavior of Ulva linza zoospores on surface topography. Biofouling 26 , 411 419 (2010) ] The second iteration of the ERI also introduced a natural log term into the attachment density component of the model. This was possibly motivated by the presently ongoing work of Cooper studying the attachment kinetics of the Ulva. The end result was effective at predicting the published data, as well as data from a new set of model surfaces designed to change both the n and s values. Magin et al . 28 ex tended the ERI model to a second organism, C. marina . This study examined the original topographies, the so called ERI series, for fouling with C. marina . It was found that the bacteria followed the same trend as the Ulva, albeit at a different slope, fo r both the stationary and log growth phases of the bacteria. The difference in slope was addressed by adding an organism specific term into the model ( Equation 2 19) nd velocity upon approaching the surface ( Equation 2 20) .
54 ( 2 19 ) ( 2 20 ) Th sed in the C. marina assay as well as to capture the differences between the organisms in terms of size . U. linza and C. marina attachment data the same slope relative to the ERI value for the different topographi es. Magin et al . 103 further extended the ERI model to materials other than PDMSe through the incorporation of a surface energy term. This study examined PEGDMA co GMA and PEGDMA co HEMA hydrogels for their antifouling efficiency, and found that the PEG based materials were more effective than PDMSe, and fur thermore the antifouling efficiency could be enhanced through the incorporation of surface topography. The attachment density on the topographically modified surfaces followed the trend of the ERI, however it was again at a different slope. The effect of surface energy on fouling was incorporated into the ERI model through a normalized surface energy term, which compared the surface energy of the topographically modified surface to that of the smooth reference. The end result is all materials and topograp hies tested with C. marina and U. linza appearing on the same line with the same slope. ( 2 21 ) The work with the ERI model shows several different components that are imp ortant regarding fouling on topographies. Specifically, the topography configuration, the organism size, and the surface material properties all play a critical role. However,
55 the slope of the line is still undefined. It is possible that the slope of th e first ERI model nd surface energy, however the slope of the relationship still does not have an explanation. Another limitation of the ERI model , was that it fail ed to predict other organisms such as N. incerta and B. amphitrite . Finally, the model was unable to predict increases in fouling . In all, the ERI model provides an excellent method to have a first prediction for how a new topography will impact fouling, however its empirical nature and limited applicabil ity leave room for improvement. In summary, the effects of topography on marine fouling are indisputable, while the mechanisms behind that fouling, and the ability to accurately predict how a new patter will behave after application are currently lacking, and merit further scientific exploration. Topographies that maximize the non wetting condition have been applied in numerous different applications to attempt to deter biofouling. Efimenko et al . 118 have used hierarchically roughened PDMSe surfaces to attempt to deter fouling in both field and laboratory assays. Th ese surfaces were prepared by plasma treating the surface of a stretched silicone and resulted in surface that was successful in reducing fouling. Aldred et al. used a similar type of topography to examine the preference of B. amphitrite for different cha nnel type topographies. These studies identified two different size scales, one approximately the same size as the cyprid foot and the other approximately the size of the cyprid body. Scardino and de Nys have extensively studied microfouling of diatoms a nd algae on patterns, successfully identifying some of the topography aspects involved in preventing barnacle attachment 112,113 . The Lamb group has re cently looked into superhydrophobic nanotopographies as a potential antifouling material with good results in prevent N. incerta attachment 114 .
56 A number of groups have studied biomimetic topographies for use as antifouling and nonwetting surfaces. Termite wings 119 , pitcher plants 64 , mollusk shells 120 , crab eyes 121 , dolphin skin 122 , and active biomimetic surfaces 67 have all been examined. The Brennan lab has extensively examined the effect of different topographies, in particular biomimetic topography based on shark skin, on the attachment of a variety of marine fouling organisms, including B. amphitrite, U. linza, C. marina and N. incerta 116,123,124 . The Sharklet microtopograph y was developed from this research ( Figure 2 13 ). This pattern mimics the denticles of a shark and has been shown to be effective against C. marina, U. linza and B. amphitrite as well as several pathogenic bacteri a species 28,116,117,124 127 , 97 Figure 2 13 . The Sharklet microtopography. The research on antifouling microtopographies by the Brennan Lab has demonstrated effects of feature size, s hape and configuration on attachment. A combined study with B. amphitrite and U. linza showed a relationship between the feature height, and therefore aspect ratio, and fouling for both species 124 .
57 Figure 2 14 . Hierarchal structure for the combined repellence of barnacles and algae. [ Reprinted with permission from Schumacher, J. F. et al . Spec ies specific engineered antifouling topographies: correlations between the settlement of algal zoospores and barnacle cyprids. Biofouling 23 , 307 317 (2007)] This study also sh owed that the size of the features must be tailored to the individual organism under study, as the large barnacle specific topographies did not prevent fouling by the Ulva . Additional studies by Callow et al. and Hoipkemeier Wilson et al. 30 have demonstrated that specific features sizes actually increase the density of attached U. linza . On these features the Ulva attach in between the features similar to the wicking wetting state for a liquid ( Figure 2 2 ). Long et al . 117 found the size and number of different features in a topography effect the fouling in a study of Sharklet like patterns. As the size of the diamond unit cell was increased, the density of U. linza attached the surface decreased. Schumacher et al . further showed the configuration of the topography influenced fouling beyond simply the number of features in a study of different feature lengths 128 . Taken together, it is plain that the effect of topography on cell attachment is complex and determining the exact effect of a given pattern on attachment is a difficult process.
58 Summ ary Marine biofouling is a dynamic process that has significant impact on a variety of industries. Effective coatings must resist colonization from organic molecules, unicellular organisms and multicellular organisms. As the industry transitions away fr om toxic coatings demand has risen for nontoxic antifouling technologies. The design of nontoxic antifouling coatings requires an understanding of the forces involved in attachment and adhesion of an organism. Colloidal interactions, mechanics of the adh esive bond, the interfacial energy and the roughness of the surface all must be considered when creating a new surface. The performance of a novel coating will depend on the characteristics of both the surface and the model organism under investigation. Certain characteristics such as surface modulus, hydration of the surface and suitable surface roughness appear to have the potential to create a universally effective surface for antifouling. The optimal conditions for modulus and hydration can be found from the underlying mechanics of the surface. Roughness, however, is less well understood in terms of how the surface structure fundamentally changes the bond between the cell and the surface and how that fundamental interaction can be exploited to create effective antifouling surfaces.
59 CHAPTER 3 SEA MODEL DEVELOPMENT Background Microtopographic surfaces have gained much attention recently as potential non toxic antifouling strategies to replace biocides in the marine environment 32,33,37,40,41,129 . The most recent literature has attempted to define a mechanism of antifouling. Configuration 114,117,128,130 , aspect ratio 124 , and feature size 125,131 are va riables that have been identified. However, the relative importance of each and the mechanisms through which the different topographical characteristics control attachment has thus far proved to be elusive. 1 The Attachment Point Theory was one of the earli est attempts to use topographic characteristics to predict organism attachment 33,110,113 . This theory is based on the empirical observation that organisms tend to attach in areas where they can maximize the number of contact points they can make with a surface, the logic being that these are the locations where the organism must expend the least amount of metabolic energy once attached. The theory is appealing for its simplicity, and has been shown to apply at lea st in part to a variety of organisms. Several attempts have been made to quantify the effect of attachment points 37,112,113 , 111 . Scardino et al . 113 showed that the number of attachment points was insufficient to predict a relative change in fouling for the diatom Amphora sp ., the green alga Ulva rigida , and the red alga Centroceras clavulatum ; interestingly, it was effective f or other organisms, the tubeworm Hydroides elegans and the bryozoan Bugula 1 Reprinted wit h permission from Decker, J. T. et al . Engineered antifouling microtopographies: an energetic model that predicts cell attachment. Langmuir 29 , 13023 13030 (2013)
60 neritina . Quantitative predictions based on Attachment Point Theory are complicated attachment points (e.g. Figure 3 1 ). The theory does not provide a way to quantify these topographies in a meaningful way, and can therefore only provide a useful observation for the development of other models. Figure 3 1 . The unit cell seen here offers three different sites with different numbers of attachment points. Features that offer 1, 2, and 3 attachment points at each location are highlighted in yellow. Previously, the Brennan Group attempted to quantitatively predict fouling through the Engineered Roughness Index (ERI) model 28,103,116,117 , which is based on the theories of wetting of a textured surface by Wenzel 105 and Cassie 106 . The ERI model relates the surface's Wenzel roughness r , the number of distinct features n , and the fractional area of feature tops s from the Cassie equation through an experimentally fitted slope m to the relative attachment density of organisms to a patterned surface compared to smooth ( N t /N s ) ( Equation 3 1) ( 3 1) The ERI model relates the relative attachment density on a topography to the physical attributes of the surface through the Wenzel and Cassie Baxter wetting theories commonly applied to topographical ly modified surfaces. Equation 1 has been shown to be effective in predicting the relative attachment density of zoospores of the green alga Ulva linza to topographies with feature spacing and width less than the
61 critical dimension of the spore (c. 5Âµm). The attachment density decreases logarithmically as the ERI value of the topography increas es in the fashion predicted by E quation 3 1. The first application of the ERI model was limited in scope, only examining zoospores of U. linza on topographies cast in polydimethyl siloxane elastomer (PDMSe). Magin et al . 28,103 extended the ERI model to the bacterium Cobetia marina as well as a second chemistry in hydroxyethyl methacrylate based hydrogels, by surface energy and the test s o quation 3 2, Figure 3 2 ). ( 3 2) Figure 3 2 . Plot of relative settlement against equation 2. [ Reprinted with permission from Magin, C. M. et al . Antifouling performance of cross linked hydrogels: refinement of an attachment model. Biomacromolecules 12 , 915 922 (2011] . The ERI predicts antifouling (i.e. inhibition of cell attachment) quite well (R 2 = 0.80), but fails to predict instances of fouling enhancement, i.e. increased attachment density of fouling organisms, as demonstrated in the recent paper by Xiao, et al . 132 . et al . 20 in an attempt to relate fouling to the size of the organism. However, the modified ERI still failed to predict
62 incr eased attachment compared to a plain, unpatterned surface. These shortcomings simply highlight the empirical nature of the ERI and that the role of surface topography as a determinant of fouling is not adequately addressed. In the present paper, we propose a new model that incorporates the concepts of the Attachment Point Theory with the predictive capabilities of the ERI model into a unified model. Our model will show that single cell and multi response to topography is largely control led the contact area available for attachment as defined by Cassie Baxter theory. Analysis Different locations on the same topography can have a different number of attachment points, and therefore must be treated differently using the Attachment Point The ory ( Figure 3 1 ). The topography must first be described in a way that will allow the investigation of individual locations in order to apply a quantitative model. This is accomplished by using a lattice model to describe different locations of the topography independently ( Figure 3 3 ). Lattice models are a commonly employed technique used to compare systems with two or more interacting components, and have found success i n describing polymer mixing behavior, polymer graft conformations, magnetic properties, and other physical phenomena 133 . Figure 3 3 . A lattice can be applied to the Sharklet AF unit cell in Figure 3 2 to accurately describe the differences between the sites. 1 2 3
63 The lattice model is effective because it allows for easy statistical analysis of a system based on discrete spatial relationships between the relevant system components. We applied this model to antifouling topographies by applying the lattice to the surface in a manner that allows the properties of a given site to be eval uated based upon the local conformation of the topography ( Figure 3 3 ). A smooth surface and a topographically modified surface were both examined in this way to elicit the inherent differences between these two types of surface. The Attachment Point Theory is ineffective because it provides no means to quantify the predicted amount of fouling. Therefore, the different sites must be described not only by their number of attachment points but also by the probabili ty of attachment at each site due to those attachment points ( Figure 3 3 ). Much of the biofouling literature indicates that certain surfaces, and locations on these surfaces, are more likely to accumulate fouling than others 16,27,33,111,134 . This effect cannot be explained as the result of statistical fluctuations. Rather, the literature indicates that fouling is most certainly not random and therefore must be controlled by the properties of the interface between the organism and the surface, namely the interfacial energy. The field of statistical mechanics provides a useful set of tools for describing the probability of an event in terms of energy. Each site has a certa in amount of interfacial area A that will result if an organism attaches at that site; this interfacial area is controlled entirely by the number of attachment points. The change in free energy G 123 to create a new interface in a three component system c an be related to A and the interfacial energy through E quation 3 3. ( 3 3)
64 The canonical ensemble was applied to calculate the probability (E quation 3 4) of a site being filled by making the assumption t hat filled sites have only one cell and the temperature and volume for all cells are constant. This probability, i.e . N t /N (the number of filled sites of a certain type ( t ) divided by the total number of filled sites) is dependent on the number of availabl e sites g t , the energy of each site E t , and both temperature T and the partition function Z of the lattice. ( 3 4) ( 3 5) We can then compare two different types of lattice sites using equation 3 6. ( 3 6) The use of E quation 3 6 permits the quantitative comparison of two surfaces for their AF potential. However, the comparison is complicated by the fact that the formation of interfaces is a function of both the surface properties and the organism. The organis ms in this case cannot be described by the kT term. The energy of the organism extends beyond kT because its movement is directed and no t defined by Brownian motion. W E o to replace kT (E quation 3 7). This is expressed as: ( 3 7) A logical choice for E o in E quation 3 7 is the attachment energy for whichever surface has been selected as a standard. For the studies examined in this paper, that surface is smo oth polydimethylsiloxane elastomer (PDMSe). We substituted E quation 3 -
65 3 into E quation 3 7 to yield an equation that related surface parameters (contact area A and surface energy ) to the relative settlement density ( 3 8) Equation 3 8 keeps the convention of using one of the test surfaces for the s rgy function that describes the relative attachment density of an organism between two surfaces. This normalized equation offers the benefit of not having to know the exact dimensions of an organism -rather, all that is necessary is the relative change be tween the surfaces. When the same surface chemistry is used, as is the case in the surfaces examined in this paper, E q uation 3 8 simplifies further to E quation 3 9. ( 3 9) The subscripts t and s in E quation 3 9 represent the value for the topographically modified surface and the smooth surface, respectively, and A ts is the difference between the interfaci al areas of the organism on the topography and smooth surfaces. Equation 9 drops the negative sign in front of the energy expression to avoid misinterpreting the effect of area on attachment, since area cannot possibly be negative as opposed to the interf acial energy change, which is likely to be negative for the majority of surfaces. An average area can be used when comparing two surfaces with more than one type of site ( Figure 3 3 ). The equations are th e same, but the effect of a heterogeneous surface can be taken into account. Some slight rearrangement leads t o the final form of the model (E quations 3 10 and 3 11).
66 ( 3 10) ( 3 11) Equation 3 11 represents a quantification of the Attachment Point Theory, which we will refer to as the Surface Energetic Attachment (SEA) model. This relation can easily b e applied to existing data for validation as well as used as a tool in the preparation of new surfaces to either enhance or inhibit attachment. The effectiveness of this model was investigated using four different organisms: zoospores of the green alga U lva linza ( U. linza ) , cells of the diatom Navicula incerta ( N. incerta ) , the marine bacterium Cobetia marina ( C. marina ), and cypris larvae of the barnacle Balanus amphitrite ( B. amphitrite ). These four organisms differ in size, shape and attachment mechan ism, which enables them to be independently assessed using the proposed model. Validation of the model with U. linza, N. incerta, C. marina and B. amphitrite provides a more complete picture of the relation between attachment area and relative attachment d ensity for fouling organisms. Experimental Methods Materials The base material for the test surfaces in this study was SILASTICÂ® T2 polydimethyl siloxane elastomer (PDMSe) from Dow CorningÂ®. SILASTICÂ® T2 is a platinum catalyzed silicone with good transpar ency and biocompatibility, and is able to maintain feature fidelity for the duration of the a ttachment assays. The PDMSe was fabricated by mixing 10 parts resin to one part curing agent (by weight), followed by five minutes of mixing. The PDMSe mixture was then degassed under vacuum for 30
67 minutes, poured into topographical molds and allowed to cure for 24 hours at ambient conditions. Pattern Selection A wide variety of topographies were used to test the validity of the proposed model. The majority of t he topographies are described elsewhere, and the attachment data for U. linza, C. marina and B. amphitrite attachment are also taken from previously published literature 116,117,124,128,131 . Most of these patterns are based on the Sharklet et al . 125 . This topography consists of ERI model, and have been included in the analysis for this study ( Table 3 1 , Figure 3 4 , Figure 3 5 , and Figure 3 6 ). Additional new patterns were fabricated to test the attachment of N. incerta ( Figure 3 7 , Figure 3 8 ). These patterns altered the size and spacing of the features in a systematic fashion in orde r to test the effect these parameters have on diatom attachment. The different feature dimensions effectively alter the A t value for diatoms attaching to these topographies, making the evaluated topographies an excellent test of the SEA model for N. incer ta . Figure 3 4 . against Ulva linza .
68 Figure 3 5 . "ERI Series" topographies, which systematically alter the r and s values of a topography to alter the ERI value ( E quation 3 1). Figure 3 6 . features (n) to change the surface's ERI value. Figure 3 7 . leaving the number of features and size of features constant.
69 Figure 3 8 topography (topography B) by altering the spacing between features and the width of features. Table 3 1 . Literature references for data used to validate the SEA model. Topography Source ERI Series Ulva Schumacher et al. 116 C. Marina Magin et al. 28 Gradient Series Schumacher et al. 128 Topography Source N series Long et al. 117 Wide Series Hoipkemeier Wilson et al. 131 Aspect Ratio Schumacher et al. 124 Inverse SK Long et al. 117 Channels and SK Magin et al. 103 Cell A ttachment D ata Navicula cells were cultured in F/2 medium contained in 250 ml conical flasks. After 3 days the cells were in log phase growth. Cells were washed 3 times in fresh
70 me dium before harvesting and diluting to give a suspension with a chlorophyll a content suspension at ~20 o C on the laboratory benches. Immediately, after addition of the diatom suspension, each dish was exposed to approximately 0.5 seconds of sonication to minimize entrapped air in the topographies. (In previous studies this has been shown not to damage cells or alter their behavior). After 2 hours the slides were gently washed in seawater to remove cells which had not properly attached. Samples were fixed in 2.5% glutaraldehyde, air dried and the density of cells attached to the surface was counted on each slide using an image analysis system attached to a fluorescence mi croscope. Counts were made for 30 fields of view (each 0.064 mm 2 ) on each slide. The attachment data for zoospores of the green alga U. linza, the bacterium C. marina, and the barnacle cyprid B. amphitrite were taken from previously published results 116,117,124,128,131 . The data from 116,117,128 have all been reported in previous papers from our group as having reduced attachment densities compared to smooth that can be p redicted by the ERI model (E quation 3 1). We have also included in our analysis the attachment data from Hoipkemeier Wilson et al . 131 showing enhanced attachment of spores of U. linza relative to smooth on pillar and channel topographies of similar size to the spores as we ll as the attachment data for U. linza and B. amphitrite on patterns with different aspect ratios from Schumacher et al. 124 . These data do not fit the ERI model and therefore provide good basis for analyzing the extent to which the SEA model can replace the ERI model for predicting both enhanced and reduced cell attachment density on a variety of topographies.
71 Cell Attachment Area C alculations Area calculations for U. linza attaching to topographies with spacing wider than the spore were made using the topography dimensions relative to an idealized version of the spore. Zoospores of U. linza are pyriform in shape while swimming and become spherical upon attachment. Zoospores are also pleomorphic since the deformable membrane is not surrounded by a more rigid cell wall. However, to simplify calculations, for the purposes of this paper, zoospores were idealized as rigid spheres 28 . The result of this simplification will be to underestimate the area of contact, however the simplification did not introduce appreciable error into the calculations. A spore in contact with a cur ved sidewall (i.e. a pillar) was assumed to have a contact area equal to half the value for its smooth counterpart. Spores were assumed to only attach in the spaces between topographic features due to the reported attachment pattern on these topographies 33 . The relative attachment area for organisms attaching to topographies with width all organisms other than U. linza ) was set t o the s value for the topography. Topographic dimensions were assumed to be constant throughout the patterned area and equal to the intended value stated in the reference. Measured dimensions deviate 116,117 . Two different methods of calculated the available sites for attachment were explored. The first was to assign every topography the number of sites the zoospores were able to occupy. For example, a 2x2 topography in which the zoospores were able to colonize the entire surface would have the same number of available sites as a
72 smooth surface. In this scenario, 5x5 channels where the Ulva only colonize the area between the channels will have a half the number of sites a smooth surface would. The other method for determining the number of available sites was to take into account the wetted surface available to the organism. The ide a behind this method is that the interfacial energy between the organism and the surface will only be the same for wetted areas of the topography. Therefore, fully wetted topographies would be the only ones with the number of sites equal to a smooth surfa ce. Topographies likely to not be fully wetted based at the time of the experiment were assigned a g s /g t value equal to s , and set to one for all other topographies. The stability of the non wetted state was calculated using the equation derived by Marmur for the minimum wenzel roughness value r min required for underwater superhydrophobicity 135 : ( 3 1 2 ) Topographies with a measured r min were considered to be in the Cassie wetting state during the experiment. The 0.15 factor was added to ensure that additional wettin g effects from the organisms (for example, forces from the Ulva probing the surface while searching for an attachment point) were accounted for when determining the wetting state for the topography. Simulation A Monte Carlo simulation was performed using MATLAB Â® software for points on a two ability to predict attach maps of U. linza on narrow topographies. Metropolis Monte Carlo sampling was used to determine the equilibrium distrib ution of points on a topographic unit cell. The unit cell lattice was defined based on the asymmetric unit cell
73 defined by Long et al . 111 ( Figure 3 9 2 with Fixed boundary conditions were used to define the edges of the unit cell. Figure 3 9 . Asymmetric unit cell of the Sharklet topography. The site energy was defined by t he attachment area, as seen in E quation 3 3. The area of att contact area. Probability was calculated by equation 9 with kT set to a reference value of o ne for a smooth surface as described previously. Equilibrium was determined by monitoring the ensemble energy at different simulation lengths. Simulations were performed for 10,000 points for each topography. Points were randomly distributed t o start t he simulation. The number of points was selected because it provided good uniformity throughout the unit cell at equilibrium while still allowing the simulation to run in a timely manner. The output from each simulation was subsequently turned into a his togram using the method described Long et al . 111 Briefly, this method involves assigning vector coordinates to each simulated cell relative to the origin of the unit cell (the lower left corner of Figure 3 9 ). These vector coordinates are then reflected and rotate to
74 superimpose a histogram of cell density for each location over a representation of a topographic unit cell. Results and Discussion Unit Ce ll Modeling The starting point selected for the evaluation of our SEA model was simulation of the attachment maps for zoospores of U. linza 111 . A simple Monte Carlo simulation gave a qualitative agreement between maps using the SEA model with those generated from the exper imentally derived data ( Figure 3 10 ). Figure 3 10 . Comparison between simulated and experimentally measured attachment maps for U. linza.
75 Areal attachment density predictions The SEA mo del was used to predict the areal attachment density for U. linza, N. incerta, C. marina and B. amphitrite on different topographies. The two methods of calculating the number of available attachment sites were explored separately. The simple, equivalent site method (not taking into account the wetting state) was used to effectively predict the attachment density of U. linza and N. incerta ( Figure 3 11 ). The change in attachment density between a smooth and pattern ed surface for N . incerta on narrow topographies (patterns with feature width and spacing smaller than the cell body) and U . linza on wide topographies (patterns with feature width and spacing larger than the spore body) was predicted by the model with hig h correlation coefficient (R 2 = 0.97). These same topographies do not fit the ERI model (R 2 = 0.12) ( Figure 3 1 1 ). The same parameters do not predict the attachment density for U. linza and C. marina to relativel 2 = 0.34). These are topographies that have been previously shown to fit the ERI model in terms of attachment density (R 2 = 0.84 for Ulva, R 2 = 0.78 for Cobetia) ( Fi gure 3 1 2 ). The results indicate that for topographies with well defined contact areas, i.e. topographies where every site is more or less identical, the SEA model can be used easily to calculate the attachment density without incorporating the wetting st ate of the topography. There appears to be some dependence on configuration as to whether or not the SEA model effectively predicts the attachment density. The topographies which are not effectively predicted ( Fig ure 3 1 4 ) all have sites with different numbers of attachment points, such as the case of U. linza on the Sharklet AF topography ( Figure 3 2 ). Interestingly, these topographies are accurately predicted by the ERI model ( Figure 3 1 4 ). The observation that the SEA model and the ERI model may act as
76 complimentary descriptions of the effect of topographies ( Figure 3 1 3 , Figure 3 1 4 ) raises the possibility of combining these two models into one equation that can be applied to all organisms and topographies. Figure 3 1 1. Correlation between relative permanent attachment density of zoospore s of U. linza and cells of the diatom N. incerta to topographies with one unique site. -1 -0.5 0 0.5 1 1.5 2 -1.00 0.00 1.00 2.00 3.00 ln(Attachment Relative to Smooth) A ts /A s Attachment Point Model Plot for Wide Topographies Ulva Wide Topographies Navicula N Series Navicula S Series Navicula ERI Series Navicula Pillar Series -1 -0.5 0 0.5 1 1.5 2 0 10 20 30 40 ln(Attachment Relative to Smooth) ERI Wide Topographies ERI Plot Ulva Wide Topographies Navicula N Series Navicula S Series Navicula -ERI Series Navicula Pillar Series R 2 = 0.97 R 2 = 0.12
77 Figure 3 1 2. Correlation between relative attachment densities of zoospores of U. linza and cells of C. marina to topographies with narr ow features. -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 -1 -0.8 -0.6 -0.4 -0.2 0 ln(A/As) A ts /A s Attachment Point Model Plot Ulva and C. Marina on 2x2 topographies Ulva ERI Series Ulva Gradient Series Ulva N Series Ulva Inverse SK Ulva Channels and SK C. Marina ERI Series y = 0.18x 0.31 RÂ² = 0.78 y = 0.07x 0.04 RÂ² = 0.84 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0 5 10 15 20 25 ln(N t /N s ) ERI Literature ERI Plot Ulva and C. Marina on 2x2 topographies Ulva ERI Series Ulva Gradient Series Ulva N Series Ulva Inverse SK Ulva Channels and SK C. Marina ERI Series R 2 = 0.34
78 The relative attachment density for spores of U. linza on topographies such as those seen in Figure 3 8 is difficult to predict from E quation 3 11 due to the aforementioned impact of these topographies on gregarious settlement. To compensate for this effect, as well as the non uniform attachment density over the topographic unit cell seen in Figure 3 8, we introduced the n from the ERI model (E quation 3 1). The n term is the only term in the ERI model that describes the configuration of the topography. The configuration of the topography is related to the number of contact points available for an organism, which is an important factor for attachment probability (Figure 3 8). Thus, n is added as a prefactor, along with substituting in p lace of A t in E quation 3 11, to account for the non uniform attachment behavior observed in Figure 3 8. The result is E quation 3 14. ( 3 14) Equation 3 14 can be applied to the topographies examined in Figures 2 and 7 to yield a single plot predicting both attachment enhancement and inhibition for spores of U. linza and attachment inhibition for cells of N. incerta and C. marina ( Figure 3 1 3 ). The n term was set to one for all N. incerta experiments since they show uniform attachment over the topography and initially at tach in random orientation as they do on a smooth surface, and thusly only effectively sense topographies the same regardless of configuration and number of features. The agreement between predicted relative attachment from the Attachment Point Model and e xperimentally measured values for relative attachment is extremely good (R 2 = 0.93), and the slope of 0.99 is very nearly the predicted value of 1. Figure 3 1 3
79 demonstrates the predictive power of this model for t hree different species of fouling organisms. Figure 3 1 3. Modified thermodynamic model to include contribution of number of features to the attachment density distribution. The addition of the n term into the SEA model is not th e ideal solution for a new model since it does not point to a mechanism through which antifouling topographies are effective. The second method for determining site concentration (taking into account the wetting state of the topography) showed a similar c orrelation to the one seen in Figure 3 1 3 with the advantage of no added terms to the model ( Figure 3 1 4) . The change in attachment density between a smooth and patterned s urface for N. incerta, C. marina and B. amphitrite on narrow topographies (topographies with spacing smaller than the cell body), zoospores of U. linza on both wide (spacing wider than the spore body) and narrow (spacing less than the spore body) topograph ies is predicted by the model with high correlation coefficient (R 2 = 0.83) and a slope of 1.05, near to the predicted value of 1 in E quation 3 11 ( Figure 3 1 4 ). -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 ln(N t /N s ) n* A/A s SEA Model with N Ulva Wide Series Ulva ERI Series Ulva Gradient Series Ulva N Series Ulva Inverse SK Ulva Channels and SK Navicula N Series Navicula S Series Navicula ERI Series Navicula Pillar Series C. Marina ERI Series R 2 = 0.93
80 Figure 3 1 4 . Comparison between the SEA model and experimentally measured attachment density data for U. linza, N. incerta, C. marina and B. amphitrite. The data can be compared to E quation 3 11. The results demonstrate that the SEA model predicts the attachment density for a number of different topographies and organisms. The model suggests the antifouling pockets between the features, i.e., non wetting state, which reduces the potential attachment sites (the ln(g t /g s ) term in the model) for the cell/organism. This is consistent with our previous evaluations of the wetting/dewetting properties of these patterns 138 , and corresponds with more recent observations made by other authors regarding antifouling topographies 114,115 . Conversely, the model suggests that th e best way to increase cell attachment, such as for applications in aquaculture, is to present a topography that maximizes attachment area. The model therefore predicts the observations that led to the Attachment Point Theory 112,113 . The attachment area changes based on the way organisms settled. Zoospores of U. linza , for example, attached to topographies with larger feature spacings by -3 -2 -1 0 1 2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 ln(N t /N s ) A ts /A s +ln(g t /g s ) Ulva Wide Series Ulva ERI Series Ulva Gradient Series Ulva N Series Ulva Positive and Inverse SK Ulva Channels and SK Ulva Aspect Ratio Series Navicula N Series Navicula S Series Navicula ERI Series Navicula Pillar Series C. Marina ERI Series B. Amphitrite Aspect Ratio Channels and SK R 2 = 0.83
81 inserting t hemselves between the features, fully wetting the topography. This increases a topography spore, which includes the sidewalls. Contrast this with the attachment area of the spore p lanar e. This allows us to simplify E quation 11 so that the local r value of a site can be expressed in terms of the size of the organism (assuming the number of attachment sites is constant). ( 3 12) Interestingly, E quation 3 12 predicts an increase in attachment as r increases, which is the trend observed by Xiao, et al . 132 , which incidentally also used topographies that were not accurately described by the ERI model. The case of cells of the diatom N. incerta attaching to topographies with narrow spacing is a sp ecial case of E quation 3 12. In this scenario, the cells are only in contact with the tops of the features. The r term is then equal to the fractional area of feature tops for the area covered by the cell ( E quation 3 13) ( 3 13) In both cases the term driving the relative attachment density in the SEA model i s also found in the ERI model (E quation 3 1): r for the case of zoospore attachment to of diatom attachment to narrow topographies (and subsequent decrease in attachment density). The predictive power of the ERI model for cer tain topographies is potentially derived from these similarities.
82 Interestingly, these two regimes correspond to the analogous regime for a liquid spreading on a topography. The spore, which settles between features, effectively 'wets' all of the availab le surface area. The attachment density of these spores is best Thus, the surfa S ummary The Surface Energetics Attachment (SEA) model, described in this manuscript, represents the first time one is able to predict both enhancement of cell attachment and inhibition of cell attachment to a microtopography. The SEA model is based upon statistical mechanics, in which the probability of attachment was related to the relative attachment energy at a speci fic site. The attachment energy for different topographies was estimated in a simple way by relating attachment energy to attachment area. The relative attachment area was also shown to be a good predictor of both total attachment density for four dispar ate organisms, zoospores of Ulva linza, the benthic diatom Navicula incerta, the marine bacterium Cobetia marina, and cypris larvae of the barnacle Balanus amphitrite. Interestingly, all four species follow the same relationship with attachment area, ind icating that this model may be applied to a wide variety of applications, including aquaculture and antifouling coating technologies. We believe that this approach offers the opportunity to further develop antifouling strategies that are nontoxic, environ mentally neutral, and more stable in a biological environment.
83 CHAPTER 4 RADIAL DISTRIBUTION OF ATTACHED ZOOSPORES AS A FUNCTION OF TOPOGRAPHY Background Marine biofouling remains a major financial burden on our world mostly through numerous industries, including shipping, aquaculture and power generation 12,14,139 . The financial impact requires that we continue to investigate relationships between the fouling species and materials used in the marine environment in an attempt to mitigate fouling. As part of our continuing effort to develop an understanding of the role of topography in preventing biofouling we present an analysis of the spatial relationships between topographical sites and the density distribution o f attached organisms, i.e., zoospores of the green algae U. linza . Our work, as well as that of other research groups, has clearly demonstrated the influence of topography on the settlement behavior of numerous organisms 140,141 . Systematic variations in pattern si ze, height and configuration has been proven to cause significant differences in attached organism adhesion strength 41,142 , attachment density 37,117 , and distribution 111,134 . Attachment density, of particular importance for the application of biofouling, can be enhanced 30,37 or inhibited 103,114,143 for a wide range of organisms. Organism density fluctuates as a function of surface topography both globally and locally depending upon the cha racteristics of both the organism and the pattern under investigation. The effect of topography on cell attachment density has been closely linked with the local configuration and dimensions of the pattern. Long et al . 111 and Cooper et al . 32 mapped preferential attachment sites for the zoospores of the U. linza within a single topographic unit cell. These mapping studies showed zoospores preferentially
84 attaching to locations where the number of attachment points could be maximized. This phenomenon has been observed for other organisms as well. Xiao et al . 144 observe d a preference for U. linza compared to edge sites on the same topography. This segregation phenomenon extends to larger organisms as well. Carl et al . 134 observed a preference for the mussel Mytilus galloprovincialis to segregate itself on different sections of an aquaculture net based on organism size and maturity. Additionally, Aldred et al . 37 observed in a choice assay the preference for cyprids of Balanus amphitrite to avoid topographies with dimensions similar to the cyprid body size and instead selectively colonize topogra phies with smaller dimensions. An understanding of the surface properties necessary to provide preferential sites for cell attachment, and the effect the spatial distribution of these sites may have on attachment, is of fundamental importance towards the d esign of new antifouling topographies. Subsequently, many studies have used physical models to in an attempt to predict the effect of surface topography on cell attachment 28,32,127,144 . We have introduced the Sur face Energetics Attachment (SEA) model as a method to relate the relative size and shape of an organism as it relates to the topography to attachment density 145 . The SEA model uses the relative change in area between a smooth surface and a topographically modified surface to predict the change in attachment density (E q uation 4 1) ( 4 1) The SEA model predicted effectively the inhibition of attachment density for four different organisms, i.e., Navicula incerta, Cobetia marina, Balanus amphitrite , and Ulva
85 linza. Additionally, a series of surface topographies, shown previously to increase the attachment density of U. linza, were fit with the model. Thus the SEA model was demonstrated effective at predicting both the inhibition and enhancement of cell atta chment to a series of microtopographies. The ability to predict the entire range of attachment density allows the SEA model to be used to explore the relationship between the configuration of the topographic unit cell and preferential attachment sites. Previously, the model was applied to the mapping studies for U. linza with good qualitative agreement with experimental results 123 . In both the experimental and modeled cases the local topographic characteristics, which increase fouling were identi fied. However, these studies were confined to the individual zoospores, and only mapped a single unit cell. The effect of adjacent unit cells on the spatial distribution of cells is still an unknown and should be explored further. It is our contention th at by developing a more global picture of the settlement density patterns, we may be able to further enhance the effect of material gradient attributes on antifouling technologies. The results indicating that different locations within the unit cell provid e preferential attachment sites lead to the observation that engineered topographies can have an effect on the gregarious attachment and growth of organisms. Many fouling organisms depend on the formation of biofilms to survive and thrive on a surface. O thers, such as U. linza , must attach gregariously to develop into mature, zoospore producing plants. Engineered microtopographies have been observed to both enhance 33 and inhibit 32,126 the ability for organisms to attach gregariously, either as a function of roughness scale or the configura tion of the pattern. For example, Cooper et
86 al. 32 showed that the Sharklet microtopography inhibits gregarious attachment of U. linza for a duration up to 240 minutes, while Callow et al . 33 showed channels and pillars with dimensions similar to that of the zoospore can enhance aggregation. As a result, topographies have been empirically observed to have a substantial effect on the fouling process beyond initial attachment of organisms. Currently, no quantitative measure of the effect of topographies on gregarious attachment behavior has been recorded in the literature. The radial distribution function (RDF) is often used in material science to describe the relative location s of atoms in a solid or liquid. The RDF is able to quantitatively describe the relative probabilities of different positions within a solid relative to a fixed point; that is, it provides a measurement of the relative distances between atoms. Additiona lly, the RDF is often used to determine the measure the effect of different treatments on cell spacing and orientation 83,88,89 , and can be used in cell biology to measure changes in cell grouping behavior 146 or rearrange ment of intracellular structures 147 . The RDF function is a powerful tool for measuring spatial effects in cell behavior caused by surfaces, and can be used effectively to quantify the effects of topography on the distribution of attached cells. Thi s study utilized the radial distribution function to quantitatively examine the spatial arrangement of zoospores of U. linza attached to engineered microtopographies. Experimental data for different topographies was examined to determine the effect of pat tern configuration of the spatial distribution. The zoospore radial distribution function mapping technique developed in this paper was then combined with the mapping technique developed by Long et al. 111 . Together, these three data sets should improve
87 our understanding of factors that can affect the spatial attachment behavior of U. linza on microtopographies. Materials and Methods U. linza Zoospore Attachment M ethodology The experimental data for this study were taken from the work of Cooper et al . 32 and Long et al . 117 The experimental details for attachment studies were described previously and so are only briefly summarized here. A series of five different topographies were used in these two studies ( Figure 4 1 ). For the U . linza assay, 10 ml of zoospore was released into a Quadriperm dish contacting the test surface. Zoospores were allowed to attach for the prescribed length of time, after which the surfaces were removed a nd the zoospores fixed with glutaraldehyde. Slides were then shipped back to the University of Florida for analysis. Figure 4 1 . N series topographies analyzed in this study and described previously 32,117 . Image processing Images of all surfaces were acquired using a Ze iss AxioCam microscope. Each surface and timepoint had a total of 36 images analyzed, with an average of 3000
88 spores per surface. Images were analyzed using MATLAB Â® , where zoospores were identified through fluorescence and had locations assigned to them ( Figure 4 2 ). The assigned points provided the basis for the zoospore centroid, which would be used to calculate the radial distribution function. Spores within a grouping were identified through a watershed tran from the analysis. The radial distribution function was then calculated for each surface using MATLAB Â® to tabulate the relative position of all observed zoospores. Figure 4 2 . Representative images for zoospores of U. linza on the n1 topography (left) and a smooth surface (right) Radial D istribution F unction The radial distribution function was mapped by a ratio of the number of zoospores dn in a ring with area d a a d istance r from the reference zoospore to the total number of counted zoospores N and the average density of zoospores in an image < > (E quation 4 2, Figure 4 3 ). Figure 4 3 . The pair distribution function measures the relative probability of a zoospore attaching a distance r away from the reference.
89 ( 4 2) Equation 4 2 sets the average density of zoospores at a given distance to a value of one. All values of g dr (r) that are above one indicate a relatively high probability, and values of below one indicate r elatively low probability compared to the average. Two different schemes of measuring zoospore distances were used for this study ( Figure 4 4 ). Measurement of individual zoospores gives a measure of the tendency for the zoospores to aggregate and can give insight into how the zoospore aggregates are shaped as well as their relative size. The RDF for groups of zoospores eliminates the contribution from zoospore zoospore c ontacts to the overall distribution. Since U. linza tend to settle gregariously the contribution to the RDF from zoospore zoospore contacts comprises the dominant contribution to the plot. Mapping of the zoospore groups eliminates this contribution and a llows us to probe further distances and identify any screening effects caused by the topography. Figure 4 4 . Zoospore mapping schematics for the radial distribution function.. Surfaces were compared for their r max, screening dis tance and in the case of the individual spore distribution first to second peak ratio values to explore the distribution of cells on the surface. The variation inherent in cell attachment experiments was
90 minimized through the large number of zoospores and zoospore groupings included in this analysis. Previous studies examining radial distribution function of bacteria on glass substrates also combined experiments for analysis, and these studies indicated the number of zoospores counted in this study (3000/s urface) is sufficient to draw meaningful conclusions from the measured screening distance and r max values 83,88 In g eneral, screening distance and r max values varied very little between similar variation between surfaces was examined by comparing the distributions between surfaces. Data for individual replicates from the kinetic study were unavailable for analysis. Reported values for screening distance and r max for both studies are the result of combining individual replicates into one calculated distribution. Differen ces between screening distances and r max values on different topographies were Zoospore Distribution Unit Cell M apping In order to fully explore the effect of local un it cell structure on zoospore distributions, and in particular differences observed for the n4 topography, the RDF was 117 . The unit cell is the same asymmetric cell used by Long 111 to map preferential sites for individual zoospores. Groups of zoospores were mapped for both r max and screening distance to determine the locations within the unit cell that match the global averages observ ed for the topography. These maps were distance from the average value on both the n4 topography and the smooth reference surface. Bright green squares represent distanc es much larger than average, bright red
91 squares represent distances much shorter than average, and dark areas represent average distances. Modeling Monte Carlo simulation was performed for points on a two dimensional lattice similar to that used in Chapter 3 ability to reproduce the experimental data observed in the images taken of zoospores on the topography. An energy value was assigned to each point in the lattice based on the number of contact point s at that point on the topography. A second energy function, based on a Lennard Jones potential, was used to simulate attrac tive forces between zoospores (E quation 4 3). Zoospores have been observed to attach gregariously, and although there is likely no t a physically attractive force between cells their interactions can be effectively modeled in this way. ( 4 3) Metropolis Monte Carlo sampling was used to determine the equilibrium distribution of spores on the surface. Pr obability was determined using E quation 4 4, where the energy wa s equal to the number of contact points defined by the lattice plus the sum of the spore spore interactions from E quation 4 3. The number of contact points is a simplification of the area of attachment and has been successfully applied previously to simul ation of Ulva spores 123 . The value of kT was set to one for all simulations. ( 4 4) The simulation was structured to mimic the experimental results. As such, the
92 normal distribution was used, with an average of 50 with a standard deviation of 20 (approximately equal to experimentally observed values). Additionally, the size of the Results and Discussion N S eries Distribution The effec t of topography configuration on Ulva spore spatial configuration was from Long et al . 117 . The distribution in Figur e 4 4 corresponds to the relative probabilities for different packing orders for the spores, and would likely change if a topography were to force spores to attach in a non close packed arrangement. The RDF for individual spores attached to these surfaces shows a slight difference from the smooth surface. A slight difference in r max and screening distance exists between the surfaces ( Table 4 1 ). However, these differences may be a result of the imaging techniques used and should not be considered statistically significant. A difference in the relative intensities between the first and second peaks can be seen for the different surfaces in Figure 4 4 . The first two peaks in the distribution function correspond to spores that are one and two spore body widths away, respectively. The n1 and n2 topographies have lower peak ratios than the other surfaces ( Table 4 1 ). This may indica te that these two topographies have a difference packing density than the other surfaces. Ulva spores tend to be pack hexagonally on a smooth surface ( Figure 4 3 ). However, the n1 and n2 topographies have configu rations that are essentially square packed, particularly when considering the intersections between features where spores have been found to preferentially attach 111 . Visual
93 inspection of attached spores supports the idea that the arrangement of preferential sites can lead to a change in packing order for the spores ( Figure 4 3 ) . Figure 4 5 . Radial distribution function for five different topographically modified surfaces as well as a smooth surface. The plots are offset by an a rbitrary amount to more clearly illustrate the shape of the curve. The horizontal lines denote a g(r) value of one A second contribution to the difference in peak ratios could be an increased t hat is, spores that would be part of an aggregate on a smooth surface but are merely close by on a patterned surface due to the physical barrier posed by the topography. The screening effect can be quantified by mapping the pair distribution function of s pore groups ( Figure 4 6 ). This distribution will eliminate contributions from adjacent spores, and increases the contrast between different surfaces. The lack of obstacles impeding aggregation leads to a random distribution of spore groups on a smooth surface, as seen in Figure 4 6 . However, the topographically modified surfaces show a large deviation from the random behavior on smooth. In particular, r max and screening distance both decrease dramatically for spore groups attached to topographies. Additionally, there is a trend for both screening distance and 0 20 40 60 80 100 120 0 5 10 15 20 25 30 g(r) r ( m) n5 n4 n3 n2 n1 Smooth
94 r max increased distance betwe the distance between likely nucleation sites for aggregates) in the cause the observed increase in r max and screening distance. Table 4 1 . Pair distribution functi on data for individual and groups of spores on different topographies Figure 4 6. Pair distribution function for groups of spores on the five n series topographies as well as a smooth surface. Distributions are stacked by an arbitrary amount for clari ty 0 5 10 15 20 25 30 0 5 10 15 20 25 30 g(r) r( m) n5 n4 n3 n2 Topography Screening distance groups ( m) Screen ing distance individual ( m) r max groups ( m) r max individual ( m) First/second peak ratio n1 6 2 11 5 1.84 n2 8 3 9 4 1.99 n3 9 3 13 4 4.08 n4 8 3 17 5 2.36 n5 8 3 23 4 3.27 Smooth 14 3 40 5 2.65
95 Statistical Comparison between N Series Surfaces The differences observed in the analysis for the combined distributions from all available surfaces appear to hold somewhat when the variation between individual test surfaces is taken into account. Anal ysis of the individual replicates found no significant differences between screening distance and r max for individual spores on any of the topographies ( Figure 4 7 ). Aggregates of zoospores varied significantly in terms of both screening distance and r max . The screening distance was found be significantly longer on smooth compared to topographically modified surfaces, with no significant differences between topographies. This may indicate that the topography has a proximity to other existing aggregates when compared to the smooth surface. For the r max measurements, no clear trend is apparent when comparing smooth and topographical ly modified surfaces, as well as topographically modified surfaces with different configurations. The r max distance on the n4 topography is significantly shorter than on the smooth, n1 and n5 surfaces. However, every topography other than n4 does not sho w any differences. The differences in r max do not correlate with observed attachment density, nor do they correlate any of the surface roughness aspects that have been measured. This may indicate that r max is unaffected by topography for zoospore aggrega tes, or possibly that r max is a random value for aggregates regardless of the surface. Further mapping of the r max values on the n4 topography (see Figure 4 8 ) indicates that locations of high and low r max values.
96 Figure 4 7 . Comparison between measures values of the radial distribution function. A) screening distance B) r max for different surfaces from Long et al . 117 Stars indicate statistically distinct groups. Kin etic Study Distributions The temporal evolution of the pair distribution function was also monitored for individual spores attaching to the n4 topography as well as smooth surface ( Figure 4 6 ). The r max value for each time point is approximately at the distance of one spore body (5 second peak ( Table 4 2 , Table 4 3 ). A trend towards lower peak ratios was observed as time increased, indicating that spore groups are likely growing during the experiment. Interestingly, there is a large difference between the peak ratios between smooth and topograp hy at the low time points, while the difference is minimized at long time points, * * * * * * ** *** *** *** *** *** * * * * * * ** ** **/*** **/*** *** ** A B
97 with the smooth surface having the lower ratio for the early time points and the topography having the lower time point at the upper time point. It is likely that spore aggr egates occur more quickly on the smooth surface simply based on the spore density at each time point. This would explain the lower peak ratio. Additionally, these results indicate that as Ulva spores reach an equilibrium spore density the probability for large spore aggregates is approximately the same between the smooth and topographically modified surfaces examined here. Figure 4 8. Radial distribution function for individual spores on the n4 topography at different settlement times. Distributions a re stacked by an arbitrary amount for clarity The distribution of spore groups was also examined for the kinetic study ( Figure 4 9 ). A trend towards lower screening distance was seen as the experiment time increas ed, while the r max value does not seem to follow a trend. The trend observed for screening distance indicates the screening effect observed in Figure 4 6 for different topographies evolves over time as more spores attach to the surface. The spore density is very low at the 30 and 45 minute timepoints due to the short experiment time. However, the similarity between the individual spore R DF at different timepoints 0 20 40 60 80 100 120 0 10 20 30 40 50 g(r) r ( m) 240 minutes 120 Minutes 60 Minutes 45 Minutes 30 Minutes
98 indicates the low spore densities are not the prim ary cause for the difference between observed spatial distributions. It is possible that as initial aggregates grow the ability for spores to attract other spores extends over a longer distance. As that distance increases, spores are attracted to more fa vorable sites near to an aggregate and are effectively screened by the topography from attaching adjacent to an existing spore. Table 4 2 . Ratio between the first and second peak in the pair distribution function for the n4 topogra phy and smooth measured at different settlement times Time First/second peak ratio Topography First/second peak ratio smooth 30 minutes 27.86 3.04 45 minutes 3.96 2.77 60 minutes 6.48 NA 120 minutes 3.15 NA 240 minutes 2.36 2.63
99 Figure 4 9. Pair distribution function for the spore aggregates on the n4 topography as a function of time. Distributions are stacked by an arbitrary amount for clarity Figure 4 1 0. Pair distribution function for individual spores on a smooth surface as a function of ti me. Distributions are stacked by an arbitrary amount for clarity 0 2 4 6 8 10 12 14 16 18 0 10 20 30 40 50 g(r) r ( m) 240 minutes 120 minutes 60 minutes 45 minutes 30 minutes -10 0 10 20 30 40 50 60 70 80 90 100 0 5 10 15 20 25 30 35 Smooth30 Smooth45 Smooth 240
100 Table 4 3 . Pair distribution function data for individual and groups of spores as a function of time on the n4 topography and smooth PDMSe Time Screening Distance Grou ps ( m) Screen distance individual ( m) r max groups( m) 30 minutes 16 4 16 45 minutes 13 3 19 60 minutes 8 4 27 120 minutes 7 4 13 240 minutes 7 3 15 Simulated Distributions Simulated results for the spore distribution shows good qualitative agre ement with experimental results, with some discrepancies. All examined configurations show a similar shape in the distribution. In particular, n2, n3, n4 and n5 were not measurably different from smooth in anyway ( Table 4 4 ). However, the n1 topography shows differences in the r max for groups of spores, as well as the first/second peak ratio. The difference in peak ratio is also seen for the experimental data the n1 and n2 topography has experimentally lower pea k peak ratios compared to the other tested surfaces. However, while the simulation was able to reproduce the peak ratio for a smooth surface effectively, the ratios for n2 and n1 both increased relative to smooth rather than decreased as observed experime ntally. However, the fact that a difference is observed is important for establishing effective strengths and weaknesses of our model (Figure 4 11 ) .
101 Figure 4 1 1. Simulated radial distribution function for 5 m zoospores on the n series topographies . Distributions stacked by an arbitrary amount for clarity An additional discrepancy between the simulation and experimental results deals with the difference between group distributions on smooth. Experimentally, the distribution for groups of spores on smooth is essentially random. The screening distance increases relative to the topographies, and the r max very large distance between groups is most probable. This is very different fr om what is observed in the simulation, where the results for the smooth surface are very similar to those of the topographies, with a low screening distance and r max values. The discrepancy between the experimental and simulation results for smooth indic ate an insufficient description of the spore spore interaction potential. The SEA model does not attempt to describe this interaction directly, instead focusing on the effect of surface topography on initial attachment. The spore distribution will be aff ected both by topography configuration (e.g. the location of the most preferential sites, and the affinity the spores have for those sites) as well as the preference for spores to attach gregariously. 0 20 40 60 80 100 120 0 5 10 15 20 25 30 g(r) r ( m) n5 n4 n3 n2 n1 smooth
102 Table 4 4 . Simulated results f or spore distributions on different surfaces Topography Screening Distance Individual ( m) Screening Distance Groups ( m) r max individual ( m) r max groups( m) first/second peak ratio n1 4 9 5 11 4.19 n2 4 9 5 9 2.98 n3 4 9 5 9 2.70 n4 4 9 5 9 2.60 n5 4 9 5 9 2.64 Smooth 4 9 5 9 2.65 Zoospore Distribution Unit Cell M apping Aggregates of zoospores were mapped for both r max and screening distance to determine the locations within the unit cell that match the global averages observed for the topogra phy ( Figure 4 12 ). This mapping technique allowed for simultaneous measurement of zoospore locations relative to the topography, similar to the previous mapping, as well as zoospore mapping relative to each other, which provided insight into the differing effects observed by the disparate locations on the topography. Clear differences between locations on the unit cell related screening distance, but not as much for r max in Figure 4 12 . When normalized against the global averages reported in Table 4 2 , r max shows a range of values both greater and less than the mean. R max appears to have a general trend of less than the mean on th e tops of features and greater than or equal to the mean between features, however this trend does not hold for all locations. Normalization against the smooth control shows all
103 locations of the unit cell (save the gap between the second and third longest features) have an r max less than smooth. Figure 4 1 2. Mapping of r max and screening distance for different locations within the n4 unit cell. Images are normalized against global averages for n4 (top) and smooth (bottom) to show differences between si tes. Colors are scaled based on intensity from red (relatively low value) to green (relatively high value) The screening distance map shows a dependence on location within the unit cell. The general trend within the unit cell is for the majority of loca tions to have screening distances equal or greater than the global average. The degree to which the screening distance deviates from the average is apparent from the intensity at each location, particularly at the end of the shortest feature and the cente r of the longest feature. These differences become more apparent when the sites are normalized against smooth, where these locations are shown to be equal to or greater than the smooth global average, with the other locations less than smooth. Average n4 Average smooth R max Screening Distance
104 The fact th at the local site geometry appears to play a role in screening distance, but not r max , should be discussed. The screening distance for groups of zoospores on a indicates the distribution is random beyond the screening distance. Taken together, it appe ars as though a single group of zoospores does not have much influence on any threshold, whereas the r max for n4 falls beyond that threshold. The n4 unit cell is much distance from any location in the unit cell. The observations in this study indicate that topograp hy has more of an influence on the screening distance based on the local site geometry, and these differences lead to a difference in measured r max , however that difference is likely not relevant for antifouling. Since the site geometry appears to play a r ole in determining the screening distance for zoospores at a given site it is useful to examine the locations in which the measured screening distance deviates from the rest of the unit cell. Both of the locations indicated as having long screening distan ces have experimental analogs that have been tested. The end of the shortest feature resembles the n1 topography, while the center of the longest features resembles the channels topography. It is interesting to note that these two topographies have simil ar attachment densities for U. linza , and that both of those attachment densities are similar to a smooth surface. However, the full n1 topography shows both screening distance and r max much lower than smooth, indicating that the geometry of that unit cel l still screens zoospores from each other.
105 Further work is necessary to understand the relationship between local site geometry effects on zoospore settlement. Conclusion The spatial arrangement of U. linza zoospores on a surface was examined through the use of the radial distribution function. The measurements taken in this study indicated the spatial distribution of attached zoospores is affected by surface topography, particularly in terms of the spacing and distribution of aggregates of zoospores. Wh ile no significant differences were observed in terms of the location of individual zoospores, the screening distance and r max values for aggregates of zoospores were found to vary with by topography and, in the case of r max , configuration of the topograph y. The topography appears to have a screening effect on the zoospores based on the screening distance measurements, and the configuration of preferential surface sites can lead to different spatial arrangements for the zoospores. The screening effect of t he topography develops over time, and the effect of measuring the spatial distribution before equilibrium settlement density is more pronounced on the topography. Mapping of the RDF as a function of aggregate location within the unit cell showed variation s in both screening distance and r max as a function of local geometry. The differences in r max appeared to be randomly distributed, while screening distance increased for locations with local configurations similar to topographies with experimental zoospo re attachment density similar to smooth surfaces. This observation matched the fact that the screening distance on a smooth surface is longer than that of the topographies, indicating that ability for the topography to influence the distance between zoosp ore aggregates may contribute to antifouling. Combined, it is apparent
106 that the distribution of U. linza zoospores on a surface can be controlled through the use of surface topography in ways that may be relevant for antifouling.
107 CHAPTER 5 EXAMINATION OF THE WORK OF ADHESION ON ENGINEERED MICROTOPOGRAPHIES B ackgroun d The design of nontoxic antifouling surfaces requires an understanding of the relationship between surface properties and cell attachment. Measurable surface properties, such as surface en ergy and surface roughness, are frequently used as indicators for antifouling efficacy 1 7 . Surface energy, for example, has been empirically correlated with attachment density for a variety of organisms 1,4 . The work of adhesion has been used to d irectly tie measured surface energy to cell attachment 2,8,9 . Work of adhesion is related to the surface energy of the two adhering bodies ( 1 and 2 ) and the interfacial energy between the two bodies ( 12 ) ( Equation 5 1 ) . ( 5 1 ) The studies that have previously used work of adhesion to predict cell attachment s imply used the term as an indication for attachment. The value provided a measurement for the change in energy upon attachment for an organism, thus providing a means to determine the thermodynamic feasibility of attachment. Attachment has been observed with unfavorable values for work of adhesion, however, and measurements of the work of adhesion for fouling organisms alone has not been sufficient to predict biofouling 9 . Surface roughness has also been directly related to attachment density through the ERI model. The limitations of this relationship have been discussed previously 6,10 ,11 (see Chapter 2). The ERI model provides an advantage over the work of adhesion in
108 that the fouling density can be directly predicted through surface measurements. The work of adhesion, on the other hand, provides a first principles based approach to predicting fouling that should, in principle, work universally with all antifouling surfaces to a degree not possible with the ERI model. T he Surface Energetics Attachment ( SEA ) model bridges the gap between the direct predictions of the ERI model and the indirect predictions gained from measuring the work of adhesion. The SEA model uses a lattice to directly relate the work of adhesion W to the relative cell attachment density ( N t /N s ) (see Chapter 3) (E quation 5 2). ( 5 2 ) The area term was factored out of the work of adhesion to form the SEA model relationship between attachment area and antifouling potential for engineered topographies. This area relationship provided excellent agr eement with experiment al data, as evidenced by an R 2 value on the order of 0.83 . The predictions from the SEA model in Chapter 3 were made with the assumption that the difference in work of adhesion between a cell attaching to a smooth surface and cell att aching to a patterned surface is solely a function of the change in attachment area. The relative change in work of adhesion was not measured for each topography and therefore may actually deviate more than the area relationship suggests. Measurement of the work of adhesion of water on the topography would examine this assumption. The work of adhesion for water on a smooth surface can be related to the contact angle through the Young DuprÃ© equation 12 . The work of adhesion here is a function of
109 the liquid vapor surface energy lv (equal to 72 mN/m for water) and the measured contact angle ( Equation 5 3 ) . ( 5 3 ) The contact angle in Equation 5 contact angle assumes a perfectly smooth surface and thus indicates the balance between the liquid vapor, surface vapor, and liquid surface interfacial ener gies. The work of adhesion in E quation 5 3 can therefore only be evaluated for contac t angles measured on a smooth surface. Surface roughness can have a significant effect on the wetting of liquid on the surface and subsequently the measured contact angle . Wenzel 13 formulated the first ured contact angle on a rough surface. The Wenzel equation relates surface roughness (defined as the ratio between the total area and planar area r ) directly to the measured contact angle * through the ( Equation 5 4) . ( 5 4 ) The assumption in Equation 5 4 is that the surface is fully wetted. This is not the case for many topographies, particularly engineered top ographies with tall features. These topographies will have wetting only on the tops of the features, with airgaps in between the features. Cassie and Baxter 14 formulated a relationship between the Young contact angle, the observed contact angle and the fractional area for the feature tops s to help describe this wetting state ( Equation 5 5) . ( 5 5 )
110 The measured contact angle on a rough surface can be related to the work of 4 and 5 5 ca n be substituted into Equation 5 3 to relate the observed contact angle to the work of adhesion ( Equations 5 6 and 5 7). ( 5 6 ) ( 5 7 ) In E quat ions 5 6 and 5 7 the work of adhesion for the liquid is a function both of the observed contact angle on the topography as well as the dimensions and configuration of the pattern. Mathematically, these two relationships are equivalent to the Young DuprÃ© e quation. Any change to the topography will result in a change in the observed contact angle; for example, it is impossible to hold all variables constant in E quation 5 6 and only change the r value of the surface since the * value will change as well. As a result, the work of adhesion should be the same for a smooth surface and a patterned surface of the same chemistry as measured by Equations 5 3 and 5 6 assuming that E quation 5 5 accurately relates the observed contact angle and the angles. If the surface is tilted the drop will remain pinned for a time before slipping from the surface. The leading edge will see the con tact angle increase while the trailing edge will see the contact angle decrease. These are referred to as the advancing and receding contact angles, respectively. The difference between these measurements is the contact angle hysteresis. The advancing a nd receding angles refer to the maximum
111 and minimum values for the contact angle measured before the drop begins to move and can be used as more complete measurements of the wetting behavior of a surface. Choi et al . 15 related the local s a t the leading and trailing edge of the drop to the observed advancing and receding contact angles. Their relationship substituted the global s value used in Equation 5 5 for the local fractional area of feature tops sl for the drop contact line at the l eading or trailing edge of the drop ( Equation 5 8). ( 5 8 ) Equation 5 6 can be used to relate the Cassie Baxter relationship, and therefore the work of adhesion, to the observed advancing and receding c ontact angle. The sl value can also be used to predict the contact angle for the drop spreading in different directions on an anisotropic topography. Topographies with features that are not similar in all directions will have different contact angles de pending on the perspective of the measurement 16 . Channels, for example, have elongated drops and can have significantly different contact angle measurements depending on the viewing orientation. Equation 5 8 extends E quation 5 5 to explain these observations and can be used to elucidate directional differences in the work of adhesion on a patterned surface. The contact angle measurements relate to the work of adhesion av eraged over a large area on the topography due to the scale of the measurement. A 5 l water drop contacts over 2000 different features on the Sharklet topography, which makes the measurement excellent for detecting average properties across the pattern. However, local interactions have been shown to play an important role in the attachm ent of fouling organisms. Ulva linza , for example, have been shown to preferentially attach to the
112 intersections between the diamond unit cell on the n series topographies (see Chapter 3). The SEA model was shown to predict this preferentially settlemen t behavior. The contact angle measurements are not sensitive to the local changes in the work of adhesion and therefore a secondary technique should be used to probe the local differences within the topographic unit cell. The atomic force microscopy ( AFM ) can be used to measure specific interactions between a probe and a surface. The AFM uses a tip on a cantilever to probe surface interactions. The tip is brought down to, and subsequently pulled from, the surface. The deflection in the cantilever is meas ured and from that deflection the surface forces are calculated. The tip can be functionalized with different chemistries or fabricated to different geometries. For example, the tip can be a 5 m sphere, similar to how the SEA model treats the Ulva spore. AFM has been previously implemented in the study of bacterial adhesion by several authors, with relevant forces mapped for antifouling 17 19 . In this c ase we can measure the force between a probe that is similar to the Ulva spore and the surface at different locations within a topography . These forces can be used to examine the magnitude of forces as the sphere approaches the surface and, relevant for th e work here, the adhesive force required to remove the probe from the surface which is related to the work of adhesion. The work of adhesion between the AFM probe and the surface can be calculated using the Johnson Kendall Roberts (JKR) model for contact b etween elastic bodies. The JKR model relates the work of adhesion W to the adhesive force F adh and the radius of the probe R ( Equation 5 7 ). The model assumes adhesive contact between the probe and the surface and is valid for surfaces with low elastic m oduli 18,20 . A large
113 work of adhesion indicates a stable bond i.e. a large force is required to separate the bonded su rfaces. The AFM can be used to measure the F adh term in E quation 5 9 at different locations on the surface, and from that data the work of adhesion can be calculated. The work of adhesion calculated in E quation 5 9 is the same term that is related to the contact angle in Equation 5 3 . ( 5 9 ) This chapter will use two complimentary techniques to characterize the work of adhesion to engineered microtopographies. The first, contact angle, will examine the average work of adhesion between a liquid (DI water) and the patterned surface. The work of adhesion will be analyzed for direction dependence and the contact angle measurements will be evaluated for how well they agree with the established relationships for contact angle and topography . This method will ex amine the average predictions from the SEA model and grant a deeper understanding of the average work of adhesion on a topographically modified surface. The second technique, AFM, will measure the work of adhesion between a 5 m borosilicate glass sphere and a PDMSe patterned film at different locations within a single unit cell (the n4 unit cell). The AFM will probe the specific differences between the sites on the unit cell and provide a better understanding of the work of a dhesion on site configuration. These measurements will help validate the SEA model prediction that the work of adhesion will be different depending on the local geometry of the topography.
114 Materials and Methods Pattern S election Four different series of p atterns were designed to test the SEA model for Ulva attachment density. The logic behind the design of these topographies is described in detail in Chapter 6. Three of the series were selected for contact angle analysis. Contact angles were measured on the c hannel s eries , n series and a ngle series (Figures 5 1 through 5 3). These patterns varied the width of the channels (channel series), the number of unique features in the Sharklet unit cell (n series) and the ratio between feature lengths, and theref ore the angle of the diagonal, for the Sharklet unit cell (angle series). Figure 5 1. Channel series topographies used contact angle measurements
115 Figure 5 2. N series topographies used contact angle measurements Figure 5 3. Angle series topographi es used contact angle measurements The patterns used for the contact angle study are asymmetric and therefore have different measured contact angles depending on the orientation of the pattern. Subsequently, the patterns were examined in two different or ientations, one with the spreading direction parallel to the long axis of the features and one with the spreading direction perpendicular to the long axis of the features ( Figure 5 4) .
116 Figure 5 4. The Sharkle t topography indicating the different directions for measurement of contact angles and adhesion data. AFM mea surements were taken on the +2.6 SK2x2_n4 topography. This topography offer ed a variety of si tes for Ulva attachment and has been shown to have p referential sites for attachment . All tested patterns were replicated in PDMSe (Xiameter T2 from Dow Corning). Details on mold preparation and pattern replication can be found in Chapter 6. Contact Angle M easurements Contact angles were measured using a c ustom built video goniometer. This system has an automatic pump system and wa s equipped with a n Edmund Optics 1312 . An LED light source was used to illuminate the drops. The stage was co nnected to a motor to have a dynamic tilting stage. DI w ater was used as the probe liquid. Drops with 10 o at a rate of 5 o /second. Continuous video w as taken of the drop during this time . Five measurements were taken for each patterned area in each direction. Paralle l Perpendicular
117 Video data was analyzed using a freely available contact angle analysis program developed by the EMSG research group at the University of Edinburgh . This program thresholds the image of the drop and has the user identify the edges where the contact angle measurements should be calculated. A parabola was fit to the edge and from that equation a contact angle is calculated for both the advancing and receding edges of the drop. This process wa s repeated for every frame of the video, allow ing a single video to be analyzed in the span of about 20 seconds. Comparison with I mageJ measurements of contact angles showed d ifferences on the order of Â±1 o which validated the accuracy of this measurement program . Contact Line Images The contact line of liquid drops were visualized on engineered topographies using fluorescently dyed DI water. Rhodamine B (1 wt%) was dissolved in DI water. A 5 l drop was placed on a topography and visualized on an inverted microscope. The fluorescence of the drop was used to visualize the contact line at low (4x) and high (40x) magnification. AFM M easurements AFM was used to map the surface interactions betwee n the Sharklet in DI water (NanoandMore USA) . The sphere was cleaned with ethanol before use. The spring constant of the cantilever was calibrated using the thermal noise method 21 . A force volume map of the topography was acquired. Force distance curves were measurements per line and 64 lines comprising the entire image. A 10 nm threshold was used for the indentation into the surface. Each pixel on th e image is the result of
118 one force curve. Attractive and adhesive forces were identified and mapped for the different locations within the unit cell. Statistical Analysis Contact angle measurements were compared to each other using ANOVA and the Tukey tes t. Significance was determined using value of 0.05. The standard error was calculated for work of adhesion values calculated from the contact angle measurements. A two tailed t test was used to compare differences between the contact angles measured in the parallel and perpendicular orien tations. Results and Discussion Contact Angle Measurements The contact angle measurements are summarized in Figures 5 5 through 5 7. The channel series (Figure 5 5) showed differences between the perpendicular and parallel directions in terms of the thre e contact angle measurements and the contact angle hysteresis. The difference was observed for all of the measurements except the receding contact angle on the 2x2 m channels. All other measurements were higher in the perpendicular direction A statistically significant trend was observed for decreasing static, advancing and receding contact angles as the channel width increased. The values ranged from 119 o to 98 o for static, 130 o to 107 o for advancing and 102 o to 85 o for receding. The trend for decreasing angles with increasing channel width did not exist in the perpendicular direction. These results indicated the channel width affected the contact angle more in the direction of the channels compared with the direction perpendicular to the channels. This makes sense from the perspective of the sl , which would depend on the channel width in the parallel direction and would not in the perpendicular direction.
119 Figure 5 5. Summarized results for contact angle measurements on the eight channels series topographies. Bars represent the standard de viation of the measurements The angle series topographies also showed differences between the parallel and perpendicular directions (Figure 5 6). The perpendicular direction was, in general, higher in terms of static and advancing contact angles (for exam ple, 137 o perpendicular and 130 o parallel for the a1 topography static contact angle). The receding contact angle had five topographies that showed equivalence between the two directions (a1, a5, a6, a7 and a8) and three topographies where the perpendicul ar contact angles was measured to be larger than the parallel contact angle (a2, a3 and a4).
120 A statistically significant trend was observed for lower static contact angles in the parallel direction as the angle of the diagonal increased (from a1 at the hig h end with 130 o to a6 at the low end with 120 o , p = .0003). No trend was observed in the perpendicular direction. A similar trend, for lower contact angles as the angle of the diagonal increased, was observed for the advancing contact angle. However this trend was not statistically significant. The trend for decreasing contact angle as the feature length increases in the parallel direction is similar to the trend that was observed for the channel topographies. However, the reasoning for the trend does not appear to be the same. As the angle of the diagonal increases and the individual features become longer the sl changes most in the perpendicular direction. The drop is likely pinned more by the small angle topographies compared with the larger angled topographies. These topographies have a higher density of intersections between diamond unit cells in the para llel direction. These intersections could act as pinning defects for the spreading drop, leading to a larger observed contact angle in the parallel direction for the low angle patterns. The n series topographies showed the least difference between the pa rallel and perpendicular directions in terms of magnitude (Figure 5 7). The trend for the perpendicular angles to be larger than the parallel angles observed for the angle and channel series was also observed for the n series. The receding contact angle w as found to be larger for some of the n series topographies, however only n7 showed a statistically significant difference in this regard (p = 0.03) . No trends were observed related to contact angle and the n value of the topography.
121 Figure 5 6. Summar ized results for contact angle measurements on the eight angle series topographies. Bars represent the standard deviation of the measurements Many similarities exist between the contact angles measured in these three series of patterns. The results are s imilar to what Long et al. 22 observed on the gradient series topographies (see Chapt differences were only found after examining the slip angle of the drop on the surfaces. The slip angle was not measured by goniometer used for this study. It is possible that differences in the we tting behavior of the angle, n and channel series topographies would become clear if these measurements were obtained. Future wetting studies on Angle Series
122 these patterns should examine this aspect and relate the results to the differences between the patterns. Fi gure 5 7. Summarized results for contact angle measurements on the eight angle series topographies. Bars represent the standard deviation of the measurements Comparison of Contact Angle Data to s The work of adhesion can only be evaluated based on the contact angle measured contact angle on the topography are followed. The predicted values of the contact angle were compared to the measured contact angles for the differen t patterns ( Figure 5 8) . The s l term was used to predict the advancing and receding contact N Series
123 angle as described by Choi et al . 15 The sl term accurately predicted some of the contact angle, in particular receding angles in the perpendicular direction. Additionally, the advancing angles for the channel series in the parallel direction were also predicted well (within 10 o of the predicte d value). Topographies with a sl value of 0 had measured advancing contact angles 20 o and 40 o lower than the predicted value of 180 o . This is possibly due to the fact that an advancing angle of 180 o would be almost impossible to physically achieve with out the drop reaching an adjacent feature. Deviations were seen for receding angles on the topographies in the parallel direction. The measured angle was between 20 o and 50 o lower than the predicted value of 126 o for these measurements (Figure 5 8). Fi gure 5 8. Predicted contact angles compared to experimentally measured contact angles for the 27 patterns in this study. The deviations, between the measured and predicted receding contact angles in the parallel direction, may be explained by the shape of the contact line. It has been 0 20 40 60 80 100 120 140 160 180 200 0 0.2 0.4 0.6 0.8 1 * (Degrees) sl CH Par Adv CH Perp Adv CH Par Rec CH Perp Rec Ang Par Adv Ang Perp Adv Ang Par Rec Ang Perp Rec N Par Adv N Perp Adv N Par Rec N Perp Rec Theory
124 suggested that the contact angle of a drop , on a particular topography , is controlled by the triple line. The triple line is the point where the drop makes contact with the topography 23 . The topographies analyzed for this study caused the contact line to assume an irregular shape which indicated the need to take the local structure of the topography into account during analysis ( Figure 5 9 and Figure 5 10). Figure 5 9. Irregular contact line shape on the +2.6SK2x2_n9 topography. Figure 5 1 0. Drop on the +2.6CH8x2 topography. The tortuous contact line structure caused a larger than predicted s value for the drop, particularly on the advancing and receding edges of the drop in the parallel direction. The local sl value for the edges of the angle and n series topography in the parallel direction was estimated to be 0.3 less than the predicted value based on the
125 measured contact angle (a change from a local sl value of 0.5 to 0.8). The sl value was calculated using a simple straight line relationship along the diamond of the unit cell (Figure 5 11A). Images of the contact line showed that it follows a much longer path (Figure 5 9). A fractal analysis of the unit cell, seen in Figure 5 11B, identified the increased path length for the contact line of the drop of the topography . This analysis considered each feature individually rather than e xamining the entire diamond as one structure. The examination of every feature led to a longer path, and subsequently larger local s , for the drop in the parallel direction of the topography. Figure 5 1 1. The contact line can be drawn in two different ways to estimate the local s value for the topography. A) The method used to calculate Figure 5 8 used a simple straight line re lationship that underestimated the contact between the drop and surface. B) Capturing the tortuosity of the contact line led to a larger measured s value that more close predicted the measured contact angles. The calculated contact angles for the paralle l direction of the angle and n series topographies fit much more closely (Figure 5 12). The receding angles for the n and channel series are still overestimated by the model, however the other adjusted values fit more closely. The difference in receding angle on the n series could be explained by nano roughness of silicone. The smooth PDMSe had an advancing angle of 115 o and a receding angle of 81 o . The Choi model becomes the Cassie Baxter model with a sl A) B)
126 value of zero. The Cassie Baxter model predicts no hysteresis on a smooth surface. Nano roughness in the smooth surface could lead to a decreased receding angle relative to the model predictions. Nanoroughness would also lead to increased advancing angles relative to the model, as observed for the angle and channel series, in Figure 5 1 2. Predicted contact angles compared to experimentally measured contact angles for the 27 patterns in th is study. The s values were calculated using the method shown in Figure 5 11 B. However, the fact remains that a range of receding contact angles were observed for the n series in the parallel direction. The sl value of these surfaces did not predict a difference i n terms of contact angle. This fact indicates that the contact angle in the receding direct ion was not described by E quation 5 8 and thus should not be included in subsequent work of adhesion calculations. It is unclear as to the exact reason why thes e topographies do not fit with the Choi model, although a likely 0 20 40 60 80 100 120 140 160 180 200 0 0.2 0.4 0.6 0.8 1 1.2 * (Degrees) sl CH Par Adv CH Perp Adv CH Par Rec CH Perp Rec Ang Par Adv Ang Perp Adv Ang Par Rec Ang Perp Rec N Par Adv N Perp Adv N Par Rec N Perp Rec Theory Static
127 explanation is an incomplete description of the contact line geometry for these patterns. Appendix D, which has a series of contact line images, shows that the geometry can noticeably vary b etween patterns, lending credence to this hypothesis. Work of Adhesion Calculations from Contact Angle Data The work of adhesion was calculated based on the measured contact angles and the local s value of the topographies. The average of the four differ ent directions was used to examine the work of adhesion of the drop, with the exception of the n series in which the receding contact angle in the parallel direction was omitted. In general the work of adhesion on the topography was found to be no more th an 20% different than the work of adhesion on the smooth surface. The largest deviation from this trend was seen for the higher angle series (a5 through a8), which was the result of higher than expected contact angles in the parallel direction. However, t he t test found these values to be statistically equivalent to smooth (p = 0.47 for a8, which had the largest deviation from the smooth value). The work of adhesion was not statistically different from the value on smooth for the topographies examined here as calculated by a Tukey test ( =0.05) . This observation agrees with the assumption of the SEA model, where work of adhesion was assumed to be a function of area. Since the area effects were accounted for in the calculation the hypothesis was that the work of adhesion should be the sa me on the topographies as on the smooth surf ace. The results in Figure 5 13 indicated that this is the case.
128 Figure 5 1 3. Normalized work of adhesion W t /W s for DI water on different topographies based on the measured advancing and receding contact an gles. AFM M easurements The AFM force measurements detected differences in terms of adhesion at different locations wi thin the topography (Figure 5 14 ) . The lowest adhesion (about 70 nN) was observed at the corners of th e features where the probe made the least contact with the features. Adhesion was also low on the edges of the features. Adh esion was the highest (about 180 nN) in the areas between the features. Figure 5 1 4. Map of the maximum adhesion forces measured by the AFM on the Sharklet micro topography in water. The scale bar is in units of nN 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 2x2 4x2 6x2 8x2 10x2 12x2 14x2 16x2 a1 a2 a3 a4 a5 a6 a7 a8 n2 n3 n4 n5 n6 n7 n8 n9 W t /W s
129 L ocalized areas of large attraction forces between the probe and the surface (12 18 nN) were observed at the intersection s between Sharklet diamonds ( Figure 5 1 5 ) . A step down in attractive force is observed between feat ures (6 14 nN). The lowest attractive force is observed on the tops of the features (1 5 nN). Figure 5 1 5. Map of the maximum attractive force measured by the AFM on the Sharklet microtopography in water. The scale bar is in units of nN The AFM measurements revealed a site dependence of the force of adhesion on the probe at different locations within the topographic unit cell. The observed differences between sites appear ed to be a function of the contact area between the probe and the surface. The adhesion in these cases was less than the observed adhesion on the smooth surface. Additionally, the differences in attractive forces (the forces relevant for attachment ) show a dependence on location, specifically that locations with a large number of contact points have a measurable higher attractive force in terms of the AFM measurements. Th e dissimilarity between the forces at different locations on the topography wa s attributed to the geometry of the surface since the ion concentrations, probe size/chemistry and other relevant parameters are all constant as defined by the experimental parameters. These measurements confirmed
130 the predictions of th e SEA model in terms of the effects of local geometry on probe interaction. The patterns examined in Chapter 3 and subsequently in this chapter were cast in polydimethyl siloxane elastomer (PDMSe), which has a modulus of 1 3 MPa. The work of adhesion for the smooth PDMSe surf ace calculated using the JKR model (E quation 5 9) was 22 mJ/m 2 , or approximately the surface energy of the PDMSe. We can use that result to validate the use of the JKR model for the work of adhesion on these surfaces . Tabor 24 developed a nondimensional parameter to determine the rel ative applicability of the JKR and Derjaguin Muller Toporov ( DMT ) models for adhesion between a sphere and a surface. The DMT model is an alternative to the JKR model that describes contact between a relatively small sphere and a surface. The Tab or parame t is related to the radius of the sphere R , the work of adhesion W , the mechanical properties of the tips and surface K and the distance at which the adhesive forces are interacting (equal to the distance at which the tip snaps to the surface or appro ximately 10 nm in this case). ( 5 10 ) In Equation 10 the K modulus of the tip and the sample ( Equation 5 11 ) . The ti p is borosilicate glass, which has a mo 25 . The surface is PDMSe , which has a modulus of 1.25 ( 5 11 )
131 t is 5.7 which indicated the contact between the probe and the surface could be modeled using the JKR model 26 . The value for the force of adhesion on the different locations on the topography were used t o map the work of adhesion using the JKR model at the different locations. The map of the different work of adhesion values was then used to create a map of work of adhesion relative to the smooth control surface ( Figure 5 13 ). The map shows that all sites on the Sharklet microtopography have lower work of adhesion compar ed to the smooth surface with a normalized average for the image of 0.87. Locations between the features were the most similar to the smooth su rface and locations on the tops of the features showed the largest change from the smooth surface . The measured values of W adh show ed alm ost no difference from the between feature sites and smooth (ratio of ~1.001) while the tops of the features were slig htly lower (ratio ~0.97) and the edges of the features were much lower (ratio ~0.5). These values resulted in the feature tops showing strong inhibitory tendencies towards adhesion while the between feature sites showed approximately the same value compar ed to smooth. These results are consistent with the SEA model calculations that assume d similar work of adhesion for the patterned area and the smooth . The area of attachment was determined to be the major factor that variation in the attachment densitie s . The adhesion of the AFM probe to the surface appears to be minimized at the edges of the features. This indicates that the feature edges (in this case, the areas where the probe makes the least amount of contact with the surface) were the location s whe re the fouling release properties were the highest. It would be extremely difficult,
132 however, for cell attachment on the feature edges considering a much more favorable location wa s present immediately adjacent to that site . I n this case, the area between the features where adhesion wa s much higher and the spore c ould form two or three contact points with the surface. Therefore, the best fouling release topography should be comprised entirely of edges, either very sharp ridges with narrow spacing or pilla rs that are narrow and spaced very close together. Figure 5 1 6. Work of adhesion at different locations within the Sharklet (+2.6SK2x2) microtopography. Red sites indicate sites with a lower work of adhesion than the smooth surface and green sites in dicate sites with a higher work of adhesion. S ummary Topographically modified silicone surfaces increased the static and advancing contact angles and reduced the receding water contact angle relative to a smooth surface. The contact line geometry wa s altered by the shape of the topographies, which
133 led to a need to closely examine the contact line in order to properly assign the sl value for the topography. Once this value was identified the work of adhesion could be properly calculated, which was found to be lower, but statistically similar, to a smooth surface. The n series, in the receding parallel configuration, still did no t match the model predictions after the change was made and were omitted from work of adhesion calculations The observation that the work of adhesion for liquid was similar on the topographies agreed with the assumptions used for the SEA model. AFM measur ements showed site dependent adhesion at different locations within the Sharklet microtopography. The work of adhesion at these different sites indicated the adhesive energy was lower than on the smooth surface. The results of the AFM experiment indicated that the ideal fouling release topography would minimize the area of interaction between the surface and the cell (as predicted by the SEA model) and would also have features that are much smaller than the size of the cell body.
134 CHAPTER 6 HIGH THROU GHPUT METHODS OF EXAMING THE ATTACHMENT OF ULVA LINZA TO ENGINEERED MICROTOPOGRAPHIES Background Strategies to prevent biofouling by Ulva linza include the use of biocidal agents, antifouling topographies, and non fouling chemistries. Numerous techniques have been utilized to direct the design of these surfaces, in particular those methods that are nontoxic to the algae 31,148 150 . These antifouling strategies are in general based on published literature and build off of previous results for Ulva as well as other organisms. Systematic studies of different pattern configurations and dimensions have identified pattern characteristics that will enhance and inhibit fouling 16,30, 33,116,124 . In general it has been found that topography with width and spacing less than that of the organism is ideal for antifouling. Conversely, patterns with feature dimensions on the order of the Ulva spore will increase fouling density. Configur ation also plays a role, although it is less clear the pattern configuration properties that lead to antifouling. The SEA model was developed to provide a means to rationally design surfaces in terms of ideal topography. The SEA model relates the area ava ilable to the organism for attachment, both in terms of wetted area of the pattern and wetted area by the organism ( Equation 6 1 ) . ( 6 1 ) Previously (see Chapter 3 ) the SEA model was shown to accurately predict the fouling density of Ulva as well as other organisms on antifouling topographies. These model predictions agree well with the experimentall y measured attachment densities for
135 all of the patterns examined. However, more testing on a wider range of patterns is necessary to fully explore all aspects of the model. The SEA model should be validated and tested on a range of patterns designed to bo th fit within the expected model parameters as well as push the limits of what the model predicts to be effective and ineffective antifouling surfaces. The current spore attachment assay does not provide an efficient is extremely time consuming . New anti fouling coatings are currently screened individually ( Figure 6 1 ). M icroscope slides coated with the topography or chemistry under investigation, allowed to equilibrate in artificial sea water for 24 hours, and ar e subsequently submerged in a spore solution. Spores are allowed to attach for a predetermined period of time after which the slides are removed from the solution, gently rinsed to remove lightly adhered spores. Spores are then fixed with glut a raldehyde and counted. Spores are identified through fluorescent imaging the chlorophyll in the spores fluoresces when excited which allows the spores to be easily identified and tabulated. Figure 6 1 . Traditional slide schematic for t esting engineered microtopographies. T he Brennan lab has only examined about 15 different patterns in the past 10 years of antifouling study as a result of the current pattern screening procedures . A high throughput technique would alleviate some of the inefficiencies inherent in the current design of antifouling topographies and allow for rapid screening of the SEA model predictions and development of new antifouling topographies. High throughput
136 techniques have been widely used in the pharmaceutical ind ustry and have recently been used effectively to screen biomaterials 104,151 157 . Anti biofouling materials were first screened in a high throughput manner by the Langer group in an attempt to identify chemistries that would be effective at preventing protein adsorption , and have since been expanded to the development of novel coatings for biomedical devices 104,157 . These studies found that L ewis base containing monomers were the most effective at preventing attachment of proteins. The Webster group at North Dakota State University has done extensive work on combinatorial methods for the design o f antifouling chemistries for marine applications 153,155,158 160 . Their technique involved combining silicone and urethane coatings to create tough, low surface energy coatings that will be effective against a ran ge of organisms. These experiments have yielded promising results that have led to the development of commercially viable antifouling coatings. Topographies are screened less often in a high throughput manner, possibly due to the difficulty in scaling up traditional photolithographic techniques. A recent example is the use of a topographic array, called the TopoChip, to screen patterns for directed stem cell differentiation 161 . These authors used a set of feature dimensions and configurations to create a library of patterns that could be screened for the desired stem cell lineage. This study was successful in creating the library of patt erns and data analysis strategy that has led to a commercial product based on this technology. H igh throughput techniques generally rely on combinatorial methods to create the surfaces. Combinatorial methods involve identifying several different factors ( for example, feature shape, dimension and spacing) and then using random combinations of these factors to generate new surfaces. Combinatorial technique s can be used to
137 create nearly infinite different surfaces and are therefore good for high throughput s tudies. However, the randomness of the technique does not allow for precise control over the different patter ns in a way that will allow for rational examination of different parameters for a given purpose. Since we are currently interested in examining the SEA model, the predictions of the SEA model and the ability of the model to generate new surfaces it makes more sense for us to use the model to rationally generate several series of surfaces rather than creating a library of thousands of different pat terns that do not separate out different aspects of the topography. Another concern in the design of high throughput experiments is the statistics involved in measuring the cell response. The combinatorial methods used to screen topographies, such as thos e used to develop the Topochip , combine as many patterns as possible into a small area in hopes that th e different patterned areas do not affect the results on the adjacent patterns . These patterns are randomly arranged and extremely small. The Ulva have been shown previously to have an effect on adjacent sites in terms of attachment 32 , particularly in terms of immediately adjacent sites. These effects appear to extend at least four time the spore body size beyo nd the location of the individual spore. Previous studies have shown anomalous results of patterned areas based on the background chemistry 143 . However, studies from our group have indicated that smooth areas adjacent to a topograph y have attachment densities similar to a full smooth slide, and that patterned areas that are 1 inch by 1 inch have attachment densities similar to full smooth slides, indicating that it is appropriate to examine multiple surfaces on a single slide. Additi onally, there is a real concern regarding the number of spores that are available to be imaged in a given area. The
138 spore density on the Sharklet topography is, in general, about 100 spores/mm 2 , or an 85% reduction from the smooth surface. Techniques suc h as the mapping require a large number of spores to produce accurate results, and density measurements may be artificially low based on the arrangement of the patterned areas and the size if the area of attachment is too low (see the SEA model for a bette r description of how the available attachment area may influence attachment density). This study will develop a high throughput method for screening antifouling topographies for effectiveness against the fouling algae Ulva linza . This study will examine 3 1 different topographic configurations rationally designed from the SEA model predictions divided into four different categories. The results of the attachment on these patterns will be compared to larger patterns to ensure that no anomalous attachment is occurring. The spore density and distribution of the spores relative to each other and the pattern will be mapped to determine the relative effects of different pattern configurations and dimensions. Finally, these results will be compared to the predic tions of the SEA model to determine the efficacy of the model and identify areas in which the predictive power of the SEA model may be improved. Materials and Methods Pattern Design All patterns were mapped using Layout Editor, a freely available tool for drawing and formatting patterns for photolithography. The patterns for this study were designed based on specific criteria laid out by the SEA model 123 . The model predicts a relationship between relative organism size and attachment density. To e xplore this relationship further and determine the limits of the predictive power of the model we used a series of channel topographies ( Figure 6 2 ) . Previously, channels have been
139 shown to have little to no effec t on attachment density of Ulva 116 . However, the SEA model predicts a decrease in attachme 50%. The channel series expanded the range of channel topographies that have been tested with the Ulva to attempt to pinpoint the channel width at which the SEA model fails to predict fouling. Figure 6 2 . Channel series topographies used for the 32 pattern array et al. 117 RI model by altering the Sharklet unit cell to include different numbers of distinct features in the diamond unit cell ( Figure 6 3 ) . Figure 6 3 . N series topographies used for the 32 pat tern array
140 The study previously showed excellent agreement with the ERI model in terms of changing the n value the design of the patterns does not significantly alter the r value and only slightly changes the s value of the topographies. The SEA mode l does from the ERI by predicting diminishing returns as the n value inc reases; this is the opposite of what is expected from the ERI model , which predicts a linear dec rease in attachment density with increasing n values. Testing the n series allowed us to probe the effect of pattern configuration for Sharklet like patterns more fully and identify interesting aspects of spore distribution as they relate to increased spa cing of preferential sites. A third series, referred to in this study as the angle series, systematically altered the ratio between feature lengths in a four feature diamond unit cell. The result in a or the series of patterns ( Figure 6 4 ) . The change in the ratio of feature lengths altered the distance between the predicted preferential sites in the topography. It is unclear if the spatial distribution of the se sites has an effect on fouling. The SEA model predicts that these sites will attract s and the spatial distribution of the sites should be unimportant. However, previous studies examining the radial distribution of spores on topographies indicated that the arrangement of these sites is important at least on the scale of local spore distributions. Furthermore, it has been suggested that one of the reasons the n1 topography is ineffective is the spacing of the preferential sites spores are allowed to attach in the preferential sites without disrupting the spore spore contact necessary for high spore
141 densities. The angle series systematically probed this relationship in a way that yielded insight into the mechanism of antifouling for topographies. Figure 6 4 . Angle series topographies for the 32 pattern array The final series tested in this study is a group of mixed asymmetric and geometric shape s. In this group there are two subgroups ( Figure 5 5) . The first is a group of geometric shapes with different sizes and spacings. These patterns consist of triangles, squares, pentagons and hexagons and uses th ese different configurations of feature shapes to test the effect of changing the feature configuration without dramatically altering the SEA model relevant properties of the pattern. Additionally, the pentagon features are spaced wider than the spore bod y, which should increase attachment as predicted by the SEA model. The other set of patterns are asymmetric variations on the Sharklet pattern. These patterns systematically fill in preferential sites or create asymmetric new sites which will have an eff ect on the fouling density and distribution of the spores. The end result many novel patterns that have not yet been explored that have a high potential for antifouling efficacy based on the model predictions.
142 Figure 6 5. Asymmetric and geometric patt erns for the 32 pattern array Patterns are named in such a way as to indicate the height, width, spacing and configuration of the pattern. The general scheme for these designations is: H eight_Category_WidthxSpacing_SpecificDesignation Topography height is noted in microns with a + or designation denoting whether the features are protruding or recessed relative to the plane of the slide, respectively. The category designation refers to the general configuration of the pattern i.e. SK for Sharklet like pa tterns, CH for channels, etc. Width and spacing are reported in microns. Specific designation labels the exact topography within the series i.e. n2 or a2 if it is the second topography in the n or angle series, respectively ( Table 6 1 ). Slide Design The patterns were arranged symmetrically on a microscope slide for assay ( Figure 6 6 ) . The traditional set up for assaying surfaces with Ulva has been to provide a 3 inch x 1 inch microscope slide. The same format was used for this study, with the only modification being the number of different patterns on the slide. Each series of patterns included 8 different patterns, for a total of 32 different patterns. Each patterned a rea was 5mm x 5 mm and spaced 1 mm apart. These dimensions were selected as
143 the minimum reasonable size for accurate measurement of Ulva attachment density and distribution. Table 6 1 . Topography designations for this study. The height, width and spacing refer to the April 2013 assay List No. Channel Width List No. Asymmetric patterns 1 + 2.6 CH2x2 9 + 2.6 1 6 Contact Points 2 + 2.6 CH4x2 10 + 2.6 SK2x2 Asymmetric 1 3 + 2.6 CH6x2 11 + 2.6 SK2x2 Asymmetric 2 4 + 2.6 CH8x2 12 + 2.6 SK2x2 Asymmetric 3 5 + 2.6 CH10x2 13 +2.6 Triangles 6 + 2.6 CH12x2 14 +2.6 Squares 7 + 2.6 CH14x2 15 +2.6 Pentagons 8 + 2.6 CH16x2 16 +2.6 Hexagons n Series Angle of Diagonal 17 + 2.6 SK2x2_n2 25 + 2.6SK2x2_a1 18 + 2.6 SK2x2_n3 26 + 2.6 SK2x2_ a2 19 + 2.6 SK2x2_n4 27 + 2.6 SK2x2_ a3 20 + 2.6 SK2x2_n5 28 + 2.6 SK2x2_ a4 21 + 2.6 SK2x2_n6 29 + 2.6 SK2x2_ a5 22 + 2.6 SK2x2_n7 30 + 2.6 SK2x2_ a6 23 + 2.6 SK2x2_n8 31 + 2.6 SK2x2_ a7 24 + 2.6 SK2x2_n9 32 + 2.6 SK2x2_ a8 Figure 6 6 . Schematic for the high throughpu t Ulva experiments. 32 5x5 mm patterned areas are spaced 1 mm apart on a microscope slide
144 T he April 2013 assay used two different pattern arrays . Two versions of the 32 pattern array, one with recessed features and one with protruding features, were teste d ( Figure 6 7 ) . These two patterns differed only in the configuration of the features relative to the plane of the slide. Figure 6 7 . Two different pattern configurations used for the April 2013 assay. The April 2014 tested two different pattern configurations. One had the four different pattern series grouped in the four quadrants of the slide and the other had the patterns randomly configured. The polyurethane molding process led t o two different spacing and width measurements on the two sets of slides. The configured slides were 2x2 m with a height of 2 m and the random slides were 1.8x2 m with a height of 2 m. Mold Fabrication Silicon wafer molds were used to create the microtopography. Single side polished silicon wafers (prime grade, 100 orientation) were treated with hexameth yldisilazane (HMDS) under heat for 5 minutes. Photoresist (Shipley S1813) o C for two minutes. Wafers were
145 exposed using the Heidelberg Laser Writer. Exposure took approximately two hours for one 32 patt ern array. Post exposure wafers were developed in 300 MIF photoresist developer for one minute. After rinsing with DI water the wafer was baked at 125 o C for Photoresist was stripped using oxygen plasma. Wafers were treated for eight minutes with HMDS post etch. Repeated attempts to test randomized pattern arrangements led to extremely low attachment to the silicone. This was determined to be related to the two step mol ding method employed for the April 2013 assay (see Appendix B for details). As such, the slides for the April 2014 assay were cast from polyurethane molds using a one step process. Polyurethane resin (Estane 58212 from Lubrizol) was dried at 104 o C for 2 hours. Approximately 2g of resin was heated to 170 o C and pressed in a Carver hot press. Aluminum plates lined with Teflon sheets were used as platens for the hot press. Polyurethane resin was pressed for 5 minutes at 1000 PSI to form sheets for molding. Silicon wafer molds were used as the master for embossing the pattern onto the polyurethane. The polyurethane sheet and silicon wafer mold were heated to 170 o using the same aluminum platens as before. Once heated, the polyurethane was embossed by apply ing a small amount of pressure (5 PSI) for 30 seconds followed by removal from the platens. Pattern fidelity on the urethane mold was examine by optical microscopy prior to casting. Bioassay Sample Preparation Xiameter T2 poly (dimethylsiloxane) elasto mer (PDMSe) from Dow Corning was used as the substrate surface for this study. Xiameter T2 is a platinum catalyzed, addition cured resin with added silica particles, and is chemically equivalent to the
146 Silastic T2 used in previous Brennan lab studies 32 . Silicone resin was mixed a 1 0. 1 base to curing agent ratio and mixed for five minutes. The silicone was subsequently degassed for 1 6 20 minutes. Silicone was poured onto the wafer mold and allowed to cure for 24 hours. Glass microscope slides were prepared for coating by applying a tie coat of allyltrimethoxy silane (ATS). The pH of 95% ethanol was adjusted to five and a stoichiometric amount of ATS was added to the solution. The solution was stirred for five m inutes. Glass slides were passed over a flame 2 3 times before the ATS solution was applied to the surface. The ATS was allowed to react for four minutes. The slides were subsequently rinsed and baked for 10 minutes at 110 o C. Patterned silicone films w ere then applied to silane treated glass slides and bonded with a fresh batch of uncured silicone. The silicone was then allowed to cure for 24 hours. The samples cast on the polyurethane molds were cast using a one step process. Silicone and glass micro scope slides were prepared as described previously. Silicone was poured onto the urethane mold and allowed to cure between the mold and the glass slide for 24 hours. Scanning Electron Microscopy Feature fidelity and feature height was verified using scann ing electron microscopy (SEM). Silicone samples were sputter coated with Au Pd. An FEI XL 40 field emission SEM was used for this study. The microscope was operated at 10 kV and the samples imaged in secondary electron mode. Ulva Bioassay The Ulva bioas say was performed using standard techniques by the Callow and Clare labs at the University of Birmingham and Newcastle University, respectively . The
147 April 2013 assay was performed at the University of Birmingham and the April 2014 assay was performed at N ewcastle University. filtered artificial seawater for 24 hours prior to testing. Zoospores were obtained from mature plants of U. linza by the standard method. A suspension of zoospores (10 ml; 1x10 6 spores ml 1 ) wa s added to individual compartments of quadriperm dishes, in the dark. After 45 minutes in darkness at c. 20 o C, the slides were washed by passing 10 times through a beaker of seawater to remove unsettled (i.e. swimming) spores . Slides were fixed using 2.5% glutaraldehyde in seawater. The density of zoospores attached to the surface was counted on each patterned square on 6 replicate slides using an image analysis system attached to a fluorescence microscope. Spores were visualized by autofluorescence of chlo rophyll. Counts were made for 10 fields of view (each 0.15 mm 2 ) on each patterned square on each slide. Image Processing Brightfield and fluorescent images (10x magnification, 1 mm2 area) were taken for all locations on a patterned area (5 images per patte rn replicate). Spore locations were identified using the fluorescent images. The assigned points provided the basis for the zoospore centroid, which would be used to calculate the radial distribution function. Spores that were part of an aggregate were i dentified using a watershed transform. locations were identified using the brightfield image. Applying a threshold to these images left only the patterned features as objects. The smallest features were identified and then used as the basis for mapping the locations of the spores relative to the topography. See Appendix A for a more detailed description of the program used to analyze the image data .
148 Image Data Analysis The ma pping technique used in Chapter 4 was applied to the data to determine the relative positions of the spores on the topography and relative to other attached spores. This analysis was performed for the n series and angle series topographies for the configu red array from April 2014. Briefly, vector coordinates were calculated for spore centroids relative to the unit cell and used to map the position of the spore relative to the topography. The radial distribution function was then applied to calculate the distribution of attached zoospores relative to the different positions on the unit cell. complete description of the method used can be found in Chapter 4 and Appendix A . Statistical Analysis of the Effect of Pattern L ocation The effect of pattern placement on the slide was examined by comparing the density of a reference area compared with the density of the surrounding patterns. This analysis was performed on the r andom array from the April 2014 assay. The average density of the surrounding patterns was compared with the density of reference pattern. The Pearson correlation coefficient was calculated for these two values to determine if any correlation was present . Correlation was determined by a t test on the data relative to the correlation coefficient ( ). Identical patterns on the same slide were also compared for both assays. The spore density values for these areas was compared using a t test ( ) . Results and Discussion Assay 1 April 2013 Assay The two configurations tested in the April 2013 assay were used to examine the edge effects of the patterns. The area surrounding a protruding pattern has been
149 observed to accumulate more zoospores than the surrounding areas, possibly effecting attachment on the patterns. The recessed pattern array was developed to eliminate that effect and provide a comparison to identify the ideal pattern configuration for testing. The assay from April 2013 showed at tachment of U. linza to all examined topographies in both the protruding and recessed configurations. The protruding configuration maintained better feature fidelity throughout the experiment compared to the recessed features ( Figure 6 8 ). This is possibly due to over exposure of the features on the recessed slide. Protruding +2.6 SK2x2_n4 pattern Recessed +2.6 SK2x2_n4 pattern Figure 6 8 . P o st assay images of protruding (l eft) and recessed (right) features. Images courtesy of John Finlay There was contrast between the different patterns on the slide in the sense that there were different number of spores attached to the different patterns i.e. the pattern configuration app eared to have an effect on attachment ( Figure 6 9 ). The fact that there were different spore densities on different patterned areas of the slide indicates that the assay can effectively show differences between pa tterns. Additionally, the similarities between two identical patterns (n4 and a1) indicate that the attachment density results are reproducible at different locations on the slide.
150 Figure 6 9 . Attachment density of U. linza t o topographies on the 32 pattern array slide. Error bars correspond to the standard error of the measurements. The channel topographies were shown to increase attachment density relative to the smooth control (the control in this case is the smooth should er surrounding the slide). It is not clear why this should be the case. The SEA model predicts the channels should decrease the attachment density of the spores for all of the tested dimensions. The increase in attachment density peaked for the 8x2 m configuration. It is possible that the spores are attaching in large numbers between the features up to this point, either because the spores themselves are smaller or the features are more easily moved out of the way by an attaching spore. Alternately, the spores may be able to attach to the tops of the features, which could be their preferred attachment position. Either explanation does not fully address the observation that the channel topographies increase the attachment density. The n series topogra attachment density for the n values tested in this study (n2 n9). Previously the Brennan group has observed an inverse correlation between the n value of a topography and 0 200 400 600 800 1000 1200 2x2 4x2 6x2 8x2 10x2 12x2 14x2 16x2 Con As1 As2 As3 T Sq P H n2 n3 n4 n5 n6 n7 n8 n9 a1 a2 a3 a4 a5 a6 a7 a8 Channels Asymmetric patterns n-series Angle of diagonal C Spore density (spores/mm 2 ) Protruding Recessed
151 attachment density of Ulva. Those results were not reproduced here, indicating that the n value does not play a direct role in the attachment density. Previously observed changes in attachment density with the n value were adequately explained by the predicted wetting state of the topogra phy as predicted by the SEA model in Chapter 3. The results of this assay therefore align with the predictions of the SEA model. The angle series showed a correlation between increasing angle and attachment density, particularly on the protruding topograp hies. The correlation corresponds to increasing s as the angle increases, therefore matching the predictions of the SEA model. The recessed features have a similar trend, however the relationship is not as well defined. There is a systematic increase in attachment density from a1 a3 followed by a jum p in attachment density at a4 to what appears to be an average value for the remaining angle series topographies. The difference can be attributed to the difference in feature fidelity between the protruding and recessed features. The asymmetric patterns showed differences between the protruding and recessed features, particularly for the triangles and squares topographies. The squares have been tested previously under the designation of pillars and have shown to be ineffective at preventing fouling 116 . Here, we show that the squares are effective when in the protruding arrangement with high f idelity and ineffective when recessed with poor fidelity. The triangles topography has been tested previously as well as a combination of triangles and pillars. That topography was shown to reduce fouling by about 60% compared to a smooth control. Here, the triangles by themselves show the same behavior as the squares, with the protruding features limiting attachment and the
152 recessed features enhancing it. These two results point to the importance of feature fidelity in preventing attachment of fouling organisms Assay 2 April 2014 Assay A second assay was performed to examine the effect of the pattern arrangement on the attachment density to different topographies. It was hypothesized that the proximity to an antifouling topography would have an effec t on the observed fouling density for the surrounding patterns. Thus, two arrangements were tested, one with randomized patterns and another with patterns in the same configuration as the April 2013 assay. These two patterns surfaces acted as separate te sts since the feature dimensions were not exactly reproduced between the two samples. The result is one sample (the nonrandom configuration) with dimensions that are 2x2 and equal to what has been shown to prevent attachment and another sample with slight ly wider dimensions (1.5x2.5) that will allow the Ulva to fit between the features. Additionally, the features for both sets of topographies were only 2 m tall, which will lead to the nonwetting state being unstable for all of the topographies except the channels on the nonrandom slide. The expectation is that the wider features will increase the attachment density relative to the control while the narrow dimensions will prevent fouling, although to a lesser degree than what was observed previously. The experimental results agree with these predictions.
153 Figure 6 10 . Ulva density results from the April 2014 assay. Error bars are Â± the standard error of the measurement The unexpected results from the April 2013 assay were not reproduced, namely the increase in attachment density on the channels and the low attachment density on the squares. Instead, these results are more similar to what is expected based on the SEA model in terms of attachment. It is possible that the featu re height played a role in the April 2013 study regarding the unexpected observations. Effect of Pattern Location on Spore Density Maps relative to the average attachment density were generated for the four different surface examined in this study ( Figure 6 11 ). These maps clearly show the differences in attachment density between the different topographies on the slide. The configured arrays (B, C and D in Figure 6 11 ) clearly show the relationships between the topography series and the attachment density. The arrays from April 2013 show a 0 100 200 300 400 500 600 2x2 4x2 6x2 8x2 10x2 12x2 14x2 16x2 Con As1 As2 As3 T Sq P H n2 n3 n4 n5 n6 n7 n8 n9 a1 a2 a3 a4 a5 a6 a7 a8 Channels Asymmetric patterns n-series Angle of diagonal C Spore Density (spores/mm 2 ) Random Configured
154 difference between the sides of the slide, specifically the side with the angle and n series on the left and the side with the chann el and asymmetric series on the right. Similar grouping is seen on the configured slide from April 2014, however only for one series (the channel series in this case). Figure 6 11 . Each different 32 pattern array has a differen t distribution of spore densities. A) Randomly configured array from April 2014 B) Configured Array from April 2014 C) Recessed array from April 2013 D) Protruding array from April 2013 The question remains as to why the large blocks of relatively high an d low density exist on the configured arrays. It is possible that the different topographic areas cause surrounding areas to have spore attachment densities similar to themselves i.e. effective antifouling topographies cause surrounding topographies to be artificially effective. The random array was used to address this concern The randomly configured array from April 2014 does not show obvious areas of relatively high or low density, particularly when compared to the other three arrays. This indicates that the blocks observed for the other arrays are the result of the groups of similar patterns and not necessarily the result of the spore density of the adjacent patterns.
155 The relationship between spore attachment density to a reference topography and the adjacent topographies was further explored by calculating the Pearson product moment correlation coefficient between the two variables. This coefficient determines if two variables are linearly correlated and can be used to test the statistical significa nce of any observed correlation. The Pearson coefficient is the R value commonly calculated to determine the fit of a linear regression to a data set, however in this instance it is usually reported as R 2 to describe the variance of the data points from th e regression values ( Figure 6 12 ). R values range between 1 and 1, the extremes of which indicate that the two variables are perfectly correlated. Therefore a value of 0 indicates no correlation between the two variables. Figure 6 12 . The correlation between the attachment density to one topography and the attachment density to the surrounding topographies on a randomly configured sample is very low (R 2 = 0.0028). The R value for the relationship between the attachment density on one topography and the attachment density on adjacent topographies on a randomly configured sample is 0.052. The significance of the relationship can be tested using a RÂ² = 0.0028 0 50 100 150 200 250 300 350 400 0 100 200 300 400 500 600 Average Surrounding Topography Density (spores/mm 2 ) Topography Denstiy (spores/mm 2 )
156 T test by relating the R value and the degrees of freedom df to the t statistic ( Equation 6 2 ) ( 6 2 ) The t test indicated that the value of R is not statistically different from a random distribut ion with a 95% confidence interval ( p = 0.79). The statistics concluded that there is no relationship between the spore density on one topography and the spore densities on the adjacent topographies. A second concern related to the validity of the measure d attachment densities stems from differences for different locations within the slide. It is possible that the measured attachment density is a function of both the pattern configuration as well as the patterns location on the slide. This issue was addr essed by including two replicates of the Sharklet pattern in the arrays. Both the n4 pattern from the n series and the a1 pattern from the angle series are the traditional Sharklet pattern used as a standard in previous assays. These two patterned areas were located on opposite sides of the slide for both the configured and random arrays. Comparison of the attachment density between the two Sharklet patterns on the four different arrays showed no statistical difference between the similar patterns ( Figure 6 13 ) . This indicates that comparisons between topographies that are not near each other on the slide are appropriate. The tests detailed above established the independence of the different topographies in terms of attachment density. Previous studies by the Callow and Brennan groups have shown that the spores attach gregariously, which indicated that the spores communicate with each other during attachment 31 33 . The c ommunication between
157 spores is shown by the gregarious attachment behavior if one spore attaches to the surface, the adjacent sites become more probable for attachment. The cooperative attachment of the zoospores led to concerns that communication betwe en spores on one topography and spores on an adjacent topography will cause different attachment density than what would be observed with the pattern by itself. The fact that the patterns appear to be independent raises the question of how long of a dista nce the spores can effect adjacent areas. The spore distribution data from Chapter 4 indicated the influence of a single spore on attachment density ranged from approximately 5 m (the size of one spore body) and about 20 m. The distribution function fo r spores on the smooth control for this study showed the same result ( Figure 6 14 ). Figure 6 13 . Comparison between Sharklet patterns at different locations within the 32 pattern arra y shows no statistical differences within the slide ( = 0.05). The influence of one patterned area over another would therefore be expected to extend approximately 20 m from the edge of the patterned area. This result points to a minimum spacing and density between patterns where the results would be exp ected to be independent of the surroundings of approximately 20 m. The patterns in this study 0 50 100 150 200 250 300 350 400 450 Random 2014 Configured 2014 Recessed 2013 Protruding 2013 Spore Density (spores/mm2) Comparison between similar patterns on single slide
158 are spaced 1000 m apart, far beyond the expected minimum distance for independence. Figure 6 14 . The radial distribution funct ion for zoospores of Ulva linza on smooth PDMSe surface from Chapter 4 showed average attachment density about 20 m from the reference spore. Comparison to SEA Model Prediction The SEA model was used to predict the attachment density for the 128 differe nt patterned areas tested in this study ( Figure 6 16 ). The same assumptions regarding wetting state were used as stated in Chapter 3. The Ulva were assumed to be able to fit fully between the features for the ran dom array from the April 2014 assay, thus gaining extra contact area from the side walls. Optical microscopy images of the patterns corroborates this assumption ( Figure 6 15 ). The relationship between the SEA mod el predictions and the experimental data showed similar results to those observed for previously published work in Chapter 3 ( Figure 6 16 ). The correlation coefficient and slope agree with the model predictions (R 2 = 0.85). The four different patterned areas have values that span the range from -5 0 5 10 15 20 25 30 0 5 10 15 20 25 30 g(r) (arbitrary units) r ( m )
159 effective antifouling to effective enhancement of attachment. These patterns are therefore able to explore the range of predictions from the SEA model and provide evidence for its validation. Figure 6 15 . Spores (dark spots in image) fit between the channels on the 4x2 channels topography for the randomly configured array Figure 6 16 . Plot of the relationship between ar ea of attachment, available attachment sites and observed attachment density for Ulva linza to the topographies in this study There is a clustering of spores around the A ts /A s value of 0.4, which corresponds to the s value of the Sharklet derivatives tested in this study. The results look to be somewhat inconsistent, with a large number of patterns reducing fouling far more than would be expected. It is possible that this is evidence of the patterns not ln(N t /N s ) = 1.06[ A ts /A s + ln(g t /g s ) ]+ 0.1 RÂ² = 0.85 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 -2 -1.5 -1 -0.5 0 0.5 1 ln(N t /N s ) A ts /A s + ln(g t /g s ) April 2014 Random April 2014 Configured April 2013 Protruding April 2013 Recessed
160 being fully wetted for these experiments. The assumption is that all patterns with an r value less the critical r value are wetted during the experiment. It is possible that some of these patterns are not in fact full y wetted during the experiment, however whether they are partially or fully dewetted is impossible to say. The opposite may be true as well. Some of the patterns where the reduction in fouling density is lower than expected may be wetted despite the assu mptions stating that they should be. It is also possible that the assumption of the spore body always being 5 m is inadequate. Smaller spores could potentially fit between the features. There is a distribution of spore sizes in reality, and it is possi ble that the accuracy of the model could be improved if this distribution is taken into account. Spore Distribution Maps The n and angle series topographies from the configured April 2014 assay were mapped for the spore locations as well as the local scre ening distance ( Figure 6 17 ). Performing this measurement provided a measure against published data to validate the assay as well as further insight into the effect of the site locations on preferential attachment . Spore maps relative to the topography agree with previous maps for the n series. The angle series shows the same preference for intersection sites seen in the n series, despite the change in distance between the sites as well as the decreased density o f these sites.
161 Figure 6 17 . Mapped spore locations (left column) and screening distances (right column) for the n and angle series (up to a6) for the configured array from the April 2014 assay.
162 Pattern Spore Location Map Scr eening Distance Map N2 N3 N4
163 N5 N6 N7 N8
164 N9 A2 A3
165 A4 A5 A6 Mapping of the screening distance did not show any obvious correlations between site location and screening distance. For the n series there appears to be a trend toward longer screening distances at the interior of the unit cell (near the center of
166 the longest feature). This same trend was observed in Chapter 4, where the Sharklet unit cell showed the longest screening distances at the center of the longest feature for groups of spores. The angle series does not show any obvious trends in screening distance. The spore maps coupled with the distribution maps indicate that the observed long screening distances may be the result of the low spore density surrounding these sites. The spores attach to the intersections at a much higher density than the interior of the unit cell diamond. The higher density at the intersections may be the cause of the short screening distances. The sc reening distances reported here are for groups of spores, while the spore maps are for individual spores. The differences between the spore map for groups and individual spores is neglible. The spores attaching at the interior of the diamond may be more aggregated than the spores at the intersections. The majority of the spores on the topographies are attached as single spores (for comparison, the number of spore groups on the a2 topography is 192 spores/mm 2 with an areal spore density of 224 spores/mm 2 ) . The change in distributions therefore is likely caused by local site preferences and not a change in the spore aggregation probability. S ummary A high throughput array for testing new antifouling topographies was developed. This array tested 32 pat terns simultaneously and was tested with four different pattern dimensions for a total of 128 patterns. Analysis of the effects of spore density on surrounding topographies showed no correlation between adjacent topographies. Additionally, similar patter ns at different locations within the assay yielded statistically between topographies. The SEA model predictions for the different patterns showed
167 good agreement with t he experimental data (R 2 = 0.85). Mapping of the preferential sites for the n and angle series topographies showed similar behavior in terms of preferential sites at the intersections between diamonds. Mapping the screening distance within the unit cell showed relatively shorter screening distances at the preferential sites. The lack of spore aggregates on these topogarphies indicated the observed differences in screening distance are the result of the preferential sites for individual spore attachment a nd not the result of increased probability for aggregation.
168 CHAPTER 7 BARNACLE ATTACHMENT AND PROBING BEHAVIOR ON TOPOGRAPHICALLY MODIFIED SURFACES Background The relationship between marine biofouling and topographic dimensions has been described in terms of the area of attachment through the Surface Energetics Attachment (SEA) model. P revious chapters (Chapter 3 and Chapter 6) demonstrated the utility of the SEA model relationship to predict attachment of microfouling organisms, specifically U. linz a, N. incerta and C. marina. Microfouling comprises an important component of the marine biofilm. A second group , macrofouling organisms , make up another component of the marine biofilm. These organisms are multicellular and include organisms such as ba rnacles, tubeworms and mussels. Macrofouling leads to more significant increases in drag and fuel consumption than microfouling and as such is important to target with an antifouling coating 1 . The SEA model provides a means to p redict the ability of a topography to prevent attachment, however Chapter 3 included three data points for macrofouling organisms, specifically attachment of Balanus amphitrite to channels and Sharklet topography. The relationship between attachment of th is organism and the area of attachment must be more fully explored to determine the ability of the SEA model to predict attachment of large fouling organisms. Balanus amphitrite ( B. amphitrite ) has been studied extensively as a model organism for barnacle fouling 2 8 . The cypris larvae of B. amphitrite probes a surface before secreting a proteinaceous adhesive to form a permanent attachment. The cyprid has a foot like appendage that is used to probe the surface in search of an ideal attachment location. This searching behavior includes swimming above the surface as
169 6 . The foot is comprised of microscopic appendages similar to those of gecko feet, however it is not believed that these appendages have the sam e function as the setae of the gecko 6 . The attachment and adhesion processes of B. amphitrite have been shown to be affected by surface topography in terms of attachment density 2,5,7,9 . The effect of topography on the attachment of barnacle cyprids is of interest to us as we examine the mechanisms that lead to antifouling by engineered microtopographies. Several works chronicle the empirical effect of topography on the attachment of barnacles 4,10,11 . Schumacher et al . 10 investigated the effect of Sharklet and channel topographies on the attachment of B. amphitrite . This study examined two pattern sizes gned to be similar to the size of the cyprid foot. The study also examined the effect of feature aspect ratio on cyprid attachment . The study found that all patterns inhibited the attachment of the barnacle cyprid, with the larger feature dimensions bein g more effective. Additionally, larger feature aspect ratios were found to be more effective than smaller aspect ratios, with a trend very similar to what was observed for Ulva spores 10 ( Figure 7 1 A and 7 1B ). The effect of topography on the attachment of B. amphitrite was further examined by Aldred et al. 4 . This study attempted to optimize the dimensions of a ridges like topography to promote antifouling and fo uling release of the barnacle cyprid. The features for this study were sinusoidal ridges that were laser machined into a mold, and therefore did not have straight sidewalls as the features in the Schumacher study did. The study by Aldred showed a prefere nce for cyprids to attach to locations where the strongest bond could form 4 . In this case, antifouling topographies were found to have
170 A) B) dimensions on the order of 40 60 500 Figure 7 1 . A) Both channels and Sharklet topogr aphy inhibited the attachment of B. amphitrite. Stars are statistically significant groups B ) B . amphitrite and U. linza have a similar relationship between topographical aspect ratio and attachment density reduction. [Re printed with permission from Schumacher, J. F. et al . Spec ies specific engineered antifouling topographies: correlations between the settlement of algal zoospores and barnacle cyprids. Biofouling 23 , 307 317 (2007)].
171 Clearly, topography has an effect on the attachment of the barnacle cyprid. The literature suggests that this effect is a function of the organism size. The relevant s ize for the inhibition of the barnacle cyprid appears to be both on the order of the searching foot appendage and the cyprid body. The SEA model is well equipped to describe the relationship between topography and organism on the basis of the relative size of the two. The lattice for the SEA model can be drawn both on the basis of the foot or the body, but the results are the same, in which dimensions slightly less than the size in question will lead to the greatest degree of attachment reduction. The data from Schumacher 10 was included in the first analysis of the efficacy of the SEA model. This data included both the effect of aspect ratio on the wetting behavior and also the effect o f the different topography sizes on the organism. The organism foot. The antifouling effect is increased as the height of the features, and therefore the stability of the non wetting state, increases. The data from Aldred is difficult to put into the model because the topographies are not well characterized. It would seem as though the general trends from the SEA model are confirmed, however the exact predictions are difficult due to the shape of the features. It is unclear how much area is being contacted by the cyprid during attachment on this surfaces, or what the wetting state would be due to the shape of the side walls. This leads to the conclusion that including similar features in a new study would be beneficial. The observed antifouling effects of the topographies for B. amphitrite may be a result of attachment area as predicted by the SEA model or may be caused by some
172 other mechanism such as disruption of the normal searching and attachment behavior of the cyprid. It is not clear from the literature how the topographic configuration affects the searching behavior of the cyprid and it is also unknown how the topography effects on searching behavior are related to the SEA model predictions. This study will systematically examine these aspects of the topography and their effect on cyprid attachment and searching behavior in an att empt to reconcile these discrepancies. Materials and Method Pattern Design The patterns for this study were selected based on the SEA model predictions about the effectiveness of the relative size of the pattern as well as an attempt to examine the effect s of configuration on barnacle attachment. Previous studies (see Schumacher et al. 10 ) showed that a difference exists between the Sharklet and the channels topographies in terms of bar nacle attachment. The SEA model suggests that the differences between these two patterns should be negligible and is worth exploring further. against Ulva were tested for this stu dy. The Ulva follow a pattern where attachment density decreases as the n value increases, and it is beneficial to examine how the barnacles change their attachment behavior in response to changes in n value and therefore configuration. Three different s izes are to be used with similar r values for each: 5x5, 20x20 and 200x200. The two larger features are on the order of the barnacle foot and body, respectively. We expected, based on the SEA model and previous reports that these two topography sizes wou ld inhibit attachment. Conversely, the 5x5 features are much smaller than the foot and on the order of the sensory organs
173 that comprise the foot . These appendages are 1 2 m in diameter. It is unlikely that the 5x5 m features will inhibit attachment, and may in fact enhance it by adding extra surface area in terms of accessible roughness by the sensory organs . The smallest features act as a control in the sense that they provide a surface that we expect to enhance fouling rather than inhibit it as we usually try to do. Table 7 1 . Topographies used for the first Balanus assay Small Dimensions Middle Dimensions Large Dimensions +7.8SK5x5_n2 +26SK20 x20_n2 +100SK200x200_n2 +7.8SK5x5_n4 +26SK20x20_n4 +100SK200x200_n4 +7.8SK5x5_n6 +26SK20x20_n6 +100SK200x200_n6 +7.8SK5x5_n8 +26SK20x20_n8 +100SK200x200_n8 Mold Fabrication Silicon wafer molds were made using the procedure described in Chapter 6 . Bri efly, fresh silicon wafers are spin coated with photoresist and exposed using a Heidelberg laser writer. After development the wafers are baked for 15 minutes and etched to the desired height. Wafers are treated with HMDS post etch to provide a nonstick surface for the silicone. Bioassay Sample Preparation Xiameter T2 poly (dimethylsiloxane) elastomer (PDMSe) from Dow Corning was used as the substrate surface for this study. These surfaces were cast in the sa me method described in Chapter 6 . Samples were assayed as free standing silicone films.
174 Barnacle A ssay Attachment density assays of B. amphitrite as well as video microscopy experiments were performed at Newcastle University by Dr. Nick Aldred using standard techniques 10 . Cypris larvae of B. amphitrite were provided by the Ritschoff group at Duke University. Prior to assay PDMSe surfaces were conditioned in artificial sea water for 24 hours. Topographies were fully wetted by rubbing a metal spatula over the surface. A drop of artificial sea water (1 mL volume) was placed on a 1 inch square artificial sea water containing 20 three day old cyprids . The drops and test surfaces were contained in a petri dish and cyprids allowed to attach in the dark for 48 or 72 hours at 28 o C. Cyprids were then counted to determine th e percentage that had attached to the surface . Attachment was reported as the perc entage of cyprids from the initial 20 that had expelled their permanent adhesive. Barnacle Video A ssay Two Basler high definition monochrome cameras were positioned at 90Âº to each other in the horizontal plane. Illumination was provided by high output IR emitting LEDs. The cameras were inclined by 20Âº in order to look down on the test surfaces. The movies of exploring cyprids were tracked automatically using commercial software (Simi Reality Motion Systems Gmbh). Three dimensional tracks were reconstru cted from the data sets for each camera and the reconstructed tracks and associated velocity profiles were used for analysis. The cyprid tracks were analyzed for step frequency, step distance and path root mean squared (RMS) radius of gyration. The cypri d was said to have stepped when the velocity exceeded 0.7 mm/s. Step frequency was calculated as the average number of
175 steps per second. Step distance was calculated as the average distance the cyprid traveled between steps. The RMS radius of gyration was calculated as an average of the distance between each measurement along the path to the average position of the cyprid ( Equation 7 1 ) . This was performed by comparing the cyprid position r i to the average posi tion r mean averaged for the total number of positions measured N . ( 7 1 ) Statistical Analysis The barnacle attachment was reported as an average percent attachment of the cyprids in the assay on each replicate surface. The standa rd error was calculated for each topography. The radius of gyration, step frequency and step length are also reported as an average for the available measurements with the calculated standard error. Statistical significance was determined using ANOVA and the Tukey test. Results and Discussion Attachment Assay 1 Cyprids were found to attach to all n series topographies ( Figure 7 2 ). The 5x5 topographies were found to increase attachment density relative to smooth . Conversely, both the 20x20 and 200x200 patterns were found to decreases attachment density relative to a smooth control. A significant difference in attachment was observed between 48 and 72 hour time points. The cyprids appeared to have difficulty at taching within the shorter timeframe of the 48 hours experiment and were able to make more significant attachments after a longer attachment time. The n value of the topography did not appear to have a significant effect on attachment. The 200x200 and
176 5x 5 topographies did not show any differences between n values, while the 20x20 had random fluctuations between n values and did not show any particular trend. Figure 7 2 . Attachment data from the first barnacle cyprid attachment experiment from July 2013. Bars represent the standard error T he patterns for this study were all fully wetted. Wetting was performed mechanically by rubbing a metal spatula over the non wetted surface to force water into the areas between features. The rubbing was not believed to cause changes in feature fidelity, but the microscale of the features was not evaluated post assay and may have been effected by the wetting technique. Pre assay wetting was necessary to promote normal attachment to the topogr aphies. Barnacle cyprids were observed to be unable to attach to partially wetted topographies, and in some cases were observed to become stuck to the features and unable to free themselves or generate normal attachments (Nick Aldred, personal communicati on) . As a result, no useful information was obtained 0 10 20 30 40 50 60 70 % Settlement 48 h 72 h
177 from non wetted samples and therefore the wetted samples were necessary to probe the differences between feature size and configuration. Post assay contact angle analysis was performed to evaluate chan ges in the wetting properties of the film during the assay. The post assay contact angles showed normal values for the water contact angle on both smooth and patterned areas ( Figure 7 3 ). The contact angle data indicated the test surfaces did not change during the experiment and that differences in attachment density with time were not explained by changes in the surface composition. Figure 7 3 . Contact angle of test surfaces post assa y. Samples were gently rinsed with DI water prior to testing. Error bars are the standard deviation. Attachment Assay 2 A second assay was performed on the same n series topographies . In general, the results of the first assay were repeated ( Figure 7 4 ). Low attachment density was observed after 48 hours, with a n increase after 72 hours of attachment time . The increase in attachment density was most pronounced for the smooth surface. The results are overall mo re scattered than with the first assay, but the trends between 100 110 120 130 140 150 160 Contact Angle (degrees)
178 increasing attachment in terms of the 5x5 topography and decreasing attachment in terms of the 20x20 and the 200x200 hold true. Figure 7 4 . Attachment data from the second barnacle cyprid attachment experiment from August 2013. Bars represent the standard error Comparison to the SEA model The SEA model predicts the trends of the barnacle cyprid attachment for the first assay. The variation present in the second ass ay leads to some deviation from the model predictions at the 72 hour time point, however the general trends predicted by the model still hold true ( Figure 7 5 ). -5.00 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00 % Settlement 48 h 72 h y = 1.04x 1.96 RÂ² = 0.36 y = 1.01x + 0.02 RÂ² = 0.74 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 ln(N t /N s ) ts /A s Assay 1 72 Hours Assay 2 72 Hours Assay 1 48 Hours Assay 2 48 Hours
179 Figure 7 5. SEA model predictions for the 4 timep oints measured in the two B. amphitrite attachment assays. The top trend line is for Assay 1 and 48 hours Assay 2 and the bottom trendline is for 72 hours Assay 2 The two lines in Figure 7 5 represent the best fit for data falling close to the SEA model p redictions (top line) and data shifted from the SEA model predictions (bottom line). The two trendlines are parallel to each other with a slope value close to the SEA model prediction of one. Video microscopy Barnacle cyprid searching behavior was succe ssfully measured on the three n8 topographies as well as smooth PDM S e. The cyprid was observed to significantly alter the searching behavior when confronted with the topography. The measured step size on the antifouling topography were significantly diff erent on the topography compared to smooth (Figure 7 6) . The step frequency was found to be statistically equivalent for the four test surfaces (Figure 7 7). Figure 7 6. Average step distance for the barnacle cyprid during a track. Bars indicate s tandard error of available measurements. Horizontal lines indicate statistically similar groups (Tukey test, =0.05) 0 0.05 0.1 0.15 0.2 0.25 0.3 smooth 5x5 20x20 200x200 Step Distance (mm)
180 Figure 7 7. Step frequency of the barnacle cyprids on the different topographies. Bars indicate standard error of the measurements. The larger features (20x20 m and 200x200 m) appeared to cause the cyprid to remain in a single location without adequately probing the surface. It is unclear if the cyprid was unable to probe the surface further or simply unwilling to move from a singl e location. Cyprid steps were at regular intervals on the smooth surface based on the motion maps ( Figure 7 11 ). Conversely, maps of the motion of the cyprid showed that the steps were irregular and disjointed on the inhibitory ( Figure 7 9 , Figure 7 10 ). The cyprid followed an irregular stepping pattern on the topography ( Figure 7 1 1 ). Plots of the cyprid velocity with time show a spike in velocity for every step the cyprid takes on the surface. The smooth surface shows regularly spaced steps approximately 3 seconds apart. These steps resulted in approximately straight line motion although some turns were observed. The 5x5 patte rn also showed steps 3 seconds apart, however these steps appeared slightly more irregular than on the smooth. The cyprid showed irregular stepping with an increased average time between steps on the 20x20 and 200x200 patterns. There appeared to be period s of time in 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 smooth 5x5 20x20 200x200 Steps/Second
181 which the cyprid would completely cease motion on the topography followed by instances of rapid motion. This is different than the regular stepping motion on the smooth PDMSe which indicated the topography was having an effect on the cyprid s earching behavior. Figure 7 8. Representative tracks of surface exploration by cyprids on a smooth T2 surface. Figure 7 9. Representative tracks of surface exploration by cyprids on the +7.3SK5x5_n8 surface.
182 Figure 7 1 0. Representative tracks of surface exploration by cyprids on a +26SK20x20_n8 surface. Figure 7 1 1. Representative tracks of surface exploration by cyprids on a +100SK200x200_n8 surface.
183 Figure 7 1 2. Representative velocity profiles of cyprids explorin g a) Smooth T2 b) 5 x 5 c) 20 x 20 and d) 200 x 200 SK_n8 surfaces. The difference between the search paths for the cyprids on the different topographies was quantified by examining the RMS radius of gyration of the path ( Figure 7 1 3 ). The radius of gyration provided a measure of the overall shape of the cyprid path. The cyprid tracks on the antifouling topographies (the 20x20_n8 and 200x200_n8) showed a statistically significant decrease in radius of gyration of the cyprid path. The 5x5_n8 topography was not statistical ly different from smooth. The change in radius of gyration quantitatively corroborated the observed paths of the cyprids on the different surfaces and indicated the feature dimensions had an impa ct on the searching behavior of the cyprid. The three different topographies did not show
184 statistical differences. The two antifouling topographies had radius of gyration values equal to approximately half of the 5x5_n8 values , however the large standard deviation in these measurements caused the results to be statistically similar. The differences between the two appear to be important, however, given the similarity between the 5x5_n8 and smooth , coupled with the attachment density data. The shorter ra dius of gyration for this study were correlated with decreased attachment density which may be related to the antifouling mechanism for the topographies. Figure 7 1 3. Root mean square (RMS) radius of gyration for cyprid tracks. The error bars are the standard error of the available measurements. Bars show statistically similar groups (Tukey test, =0.05) Discussion Three sets of topographic feature sizes were used for this study. The first was , the second was set to target the be fully probed by the cyprid foot and the patterns is shown in Figure 7 2 and Figure 7 4 . There appe ared to be no significant 0 0.5 1 1.5 2 2.5 Smooth 5x5 20x20 200x200 RMS Radius of Gyration (mm)
185 difference between the two inhibitory topography sizes in terms of attachment. The searching behavior of the cyprid also appeared to be affected by the two inhibitory topographies in the same manner. The change in searching beha vior on the 20x20 m and 200x200 m topographies may be due to the difficulty the cyprid has in orienting itself on larger features compared to smaller features. The conclusion here is that sizes of topography that target both the cyprid foot and the cyprid body may be use d effectively for antifouling for the barnacle cyprid. Pattern configuration has long thought to be an important component for antifouling 12 . In this study, the configuration of the pattern, at least in terms of the n series, did not have a sign ificant impact on the attachment density of the barnacle cyprid. There are several possible explanations for the lack of difference in attachment density between feature configurations. The effect of configuration on the attachment of Ulva linza has been attributed , in part, to the ability of preferential attachment sites on the topography to disrupt the gregarious settlement of the zoospores. This disruption was shown by Cooper et al . 13 and was also indicated b y the calculated radial distribution functions for attached zoospores on the topographies (Chapter s 4 and 5) . The data presented here does not indicate preferential attachment sites for the barnacle cyprid . The search paths of the cyprids are not mapped r elative to the topography and it was not possible from this data to identify if the cyprids have preferential attachment points on these topographies. However, it seems unlike that they would have preferential attachment sites. The 5x5 m and 20x20 m to pographies are too small for one site to be preferential based on the size of the cyprid body and the 200x200 m topography is too awkward for the cyprid to establish preference based on the tracking
186 data. It is more likely that small scale roughness, on the order of the sensory organs in the cyprid foot, would lead to preferential attachment sites. The conclusion therefore is that pattern configuration has an effect on cyprid attachment only in terms of the available area for attachment. T he pattern conf iguration does not seem to affect the attachment density of the barnacle cyprids . Pattern size, however, appear ed to have an influence on the way in which the cyprids probe the surface. The pre sence of the topography disrupted the step size and step freq uency for the cyprid. This effect was only seen for the topographies that inhibit ed attachment, which suggest ed that that the disruption of the normal searching behavior leads to the inhibition of attachment. In particular it appear ed as though certain t opographies prevent ed the cyprid from probing the surface . The data from the video microscopy experiments indicated that some of the anomalous searching behavior ( Figure 7 5 ) was possibly the result of the antifou ling nature of the topography and not the result of the physical pattern itself. The change in the step size and radius of gyration indicated that the cyprid explores a smaller area on the antifouling topographies compared to the smooth surface during the assay. The correlation between the radius of gyration and the attachment density on the topographies indicated the two may be related, however it has yet to be established if the one causes the other. The wetting state of the topographies and the relatio nship between the wetting state and the attachment of the barnacle cyprids is an important distinction that must be discussed further. The SEA model predicts that the best surface for antifouling is a perfectly non wetted surface, or a surface that has g t value of zero. The topographies
187 used in this study will naturally be non wetted, save for the feature tops. The topographies were purposefully designed to be similar to the topographies tested with the Ulva in terms of the wetting state so the results c ould be directly compared. However, running the experiment with nonwetted topographies proved to be impossible due to the fact that the barnacle cyprids appeared to become trapped to the topography. It is not immediately clear why this would be the case. The fact that t he non wetted topography resisted the normal attachment function of the barnacle cyprids supported the conclusions of the SEA model in which the non wetted surface is the best for antifouling. However, the mechanism by which the SEA model predicts non wetting will prevent fouling, the reduction in attachment sites, did not appear to precisely apply to the situation with the cyprids. The cyprid was still able to find the s urface, and was still able to attempt to find a location to attach bu t , rather than just never finding a place to attach, the barnacle instead beca me stuck and cannot free itself to search further. It would appear as though there is some other force preven ting the cyprid from searching. It is also possible that the non we tted surface is toxic to the cyprids and the observations of stuck cyprids are actually cyprids that have died as a result of contacting the surface , however no toxicity was reported . The SEA model effectively predicted the attachment density for three of the four attachment measurements (slope 1.02, R 2 = 0.74). The fourth measurement, Assay 2 after 72 hours, showed lower attachment than what was predicted by the SEA model. The plot of A ts /A s v. ln(N t /N s ) for this timepoint showed a similar slope to what is predicted by the SEA mode (1.04 measured compared to 1.00 predicted), albeit with a low R 2 value of 0.36. The similarity in slope indicated the model was still able to
188 describe the differences between the topographies for this timepoint. The deviation from the model came with the y intercept of the trendline ( 1.96 measured compared to 0 predicted). The difference between the model prediction and the experimentally measured response appeared to be related to the attachment kinetics of the cyprid between the 48 timepoint (which fits the model) and the 72 hour timepoint (which does not fit the model) the attachment density on smooth increases 10 fold while the attachment density on t he topographies do not appreciably increase (Figure 7 4). The reason for the difference between the topographically modified and smooth surfaces is not immediately clear. The difference may be related to the searching behavior of the cyprid on the topogr aphies. The alteration in the probing behavior led to the cyprid searching for longer in order to identify a suitable attachment site. The difference between the two assays could be attributed then to the biological diversity of the cyprids, and the hypo thesis would be that after 96 hours of assay the results would once again fit the model. The results of this study point to several different design techniques that may be used in the future to prevent the attachment of B. amphitrite . The SEA model can clearly be applied on both the scale of the cyprid body and the scale of the cyprid foot to prevent attachment. It is possible that hierarchical topography that combines both length scales will be the most effective at preventing attachment. The hierarch ical topography brings the possibility that a single topography can be designed to inhibit both the large barnacle cyprid and the small Ulva zoospore. Schumacher previously attempted to apply this method to the creation of a universal antifouling surface, with limited success 10 . In that study the Ulva were undeterred by the hierarchical
189 topography, which was a 200x200 channels with 2x2 Sharklet superimposed over the top. The SEA model predicts that the Ulva will preferentially attach along the edges of the large pattern in a much higher density than if that pattern did n o t exist, thus mitigating the effectiveness of the Ulva specific Sharklet pattern. It is possible that adding topogr aphy along the side walls of the features, or possibly using some other hierarchical scheme to prevent the Ulva from colonizing between the larger features may work well. S ummary Microtopography has an effect on the attachment and searching behavior of B. amphitrite . Antifouling topographies we re found to cause a decrease in the step size and alter the normal search path of the cyprid as measured by the radius of gyration of the search path . The SEA model predicts the attachment density for three of t he four time points and topographies examined. The fourth, a 72 hour timepoint, did not have any increased attachment for the topographies while the smo oth increased 10 fold. This le d to the SEA model providing an inaccurate y intercept. It is possible these surfaces remain antifouling for longer than the time examined for this assay, leading to different attachment kinetics for the cyprid.
190 CHAPTER 8 CONCLUSIONS AND FUTURE WORK Conclusions The work of adhesion was shown to relate to the attachment dens ity of fouling organisms through the Surface Energetics Attachment (SEA) model. The SEA model related the local work of adhesion for a cell to the geometry at an attachment site. The model was related to the topographic configuration to predict the local attachment pattern of U. linza and the areal attachment density of U. linza , N. incerta , C. marina, and B. amphitrite . Experimentally available attachment density results from these four organisms showed good agreement with the model predictions (R 2 = 0. 83). Additionally, the model effectively predicted the local attachment density of U. linza on topographically modified surfaces. These experiments suggested that control of the wetting state of the topography and the fractional area of the feature tops were the most important factors in controlling fouling to topographies. The effect of topography on cell attachment was further explored by calculating the radial distribution function (RDF) for U. linza attached to topographies. The RDF was calculated fo r spores on the n series topographies, from a kinetic study and also were simulated using the SEA model. The n series showed similar results depending on topographic configuration. Kinetic study results showed dependence on the time of the measurement on the calculated RDF. This effect was attributed to the increase in spore density leading to larger spore groupings and therefore longer range interactions. Simulations reproduced a similar radial distribution function, however the noise inherent to the ex perimental measurements led to some discrepancy between simulated results and experimental results. The distribution of zoospores was additionally explored by
191 combining the unit cell mapping technique and the RDF mapping technique. These results showed s horter screening distance for spores attached to the intersecting sites on the Sharklet diamond compared to spores attached near the interior of the diamond. These sites correspond to areas of high attachment density and low attachment density, respective ly. The work of adhesion was quantified on the topographies in terms of contact angle measurements and AFM adhesion measurements. Contact angle results on the n series, angle series and channel series topographies showed results that depended more on the size of the features than the configuration of the features. The channel series showed a trend towards lower contact angles for wider channels as the measurements were made perpendicular to the channels. The other series and directions were affected by t he drop pinning at the intersecting and channel gap sites of the topography. The advancing and receding contact angles were compared to the s l term to predict contact angle with some agreement with the theory when the contact line geometry was taken into account. The work of adhesion was found to be statistically similar between the smooth and topographically modified surfaces. AFM measurem ents with a 5 m colloidal probe on a 2x2 m Sharklet topography showed site dependence for attractive forces and adhesive forces. Attractive forces showed larger attraction at the interstitial sites. Adhesion was highest between the features and lowest at the edges of the features. A map of the work of adhesion on the topography compared to the work of adhesion on smooth was also calculated. The work of adhesion on the topography was found to be approximately
192 90% of the work of adhesion on smooth on av erage. A site dependence, similar to what is predicted by the SEA model, was also observed. Further experimentation on the attachment of U. linza to engineered topographies was also performed. These experiments were carried out a new, high throughput sam ple geometry that simultaneously examined 32 patterned areas. The patterned areas were broken into four groups to test different aspects of the SEA model. These were n series, angle series, channel series and geometric/asymmetric series. The 32 pattern array was found to discriminate between attachment densities to different locations on the slide. Additionally, the location of the pattern, and the arrangement of the patterns on the slide, were found to have no effect on the observed attachment density. The SEA model effectively predicted the attachment density for the 128 different patterned areas examined across the two experiments using the new pattern geometry (R 2 = 0.84). Unit cell mapping for spore locations and RDF measured screening distance o n the n and angle series topographies found similarities between the different patterns. Spores preferred the intersecting sites between diamonds regardless of the density of intersecting sites. The screening distance was also observed to be, in general, shorter at the intersecting sites. The conclusion is that the change in density is related to the density of spores at these different sites. Bioassays were also run with the macrofouling species B. amphitrite . Three sets of n series topographies were p repared with different topography dimensions. The smallest set was designed to be below the critical dimensions to inhibit attachment. The middle set was designed to target the cyprid foot and the largest was designed to target the cyprid body. Measurem ents were taken at 48 and 72 hours culture for two different
193 assays for a total of four different measurements. The SEA model was found to predict the attachment density for 3 of the 4 assays. The 72 hour time point in the second assay showed lower attach ment than what was predicted with the SEA model. The attachment behavior of the barnacle cyprid was further explored through the use of video microscopy. Three dimensional tracks were obtained for cyprids on the n8 topography for the three different size s as well as a smooth surface. The radius of gyration for the cyprid track, step frequency and step distance were measured for each track. The radius of gyration and step distance were statistically lower on the two topographies that were observed to inh ibit attachment and similar to smooth on the topography that enhanced attachment. Step frequency was unchanged on the different surfaces. The conclusion is that the low attachment density for barnacle cyprids on topography is related to both the area of attachment and also the ability for the organism to probe the surface. The SEA model predictions are therefore only valid at very long time points after the cyprid has exhausted its ability to probe the surface. Future Work The SEA model predicts the two most important characteristics for an antifouling topography are the area available for attachment and the wetting state of the topography. The most effective antifouling topography is predicted to have a s value that approaches 0 and remain in the non wetted Cassie state. The best antifouling topography, therefore, should have the most stable nonwetted state. Two ways of achieving this goal are to create features with large undercuts and to create featu res with nanoscale roughness. Features with large undercuts will increase the r value for the topography and therefore increase the underwater stability of the superhydrophobic state. The current method for etching the silicon wafer molds produces under cuts of
194 approximately 100 nm on the side walls of the features. These undercuts could be increased to 150 nm by changing the etching process or could be manufactured to be as large as 1 m if the method for creating the topographies was altered (Figure 8 1). Figure 8 1. Topography in PDMSe with a 500 nm undercut. Nanoscale roughness can also be used to increase the r value of a surface and therefore the stability of the Cassie wettin g state. The scale of this roughness may have a detrimental effect for fouling, however, as it is likely at a scale that can be exploited by fouling organisms to increase attachment. A good solution would be to combine the nanoscale topography with micro scale topography that the SEA model predicts will prevent attachment. Combining two or more roughness scales will be necessary to create a universally effective antifouling surface based on the SEA model predictions. A large scale topography (200 m) will target the barnacle cyprids while remaining larger than the 20 30 m distance that the Ulva spores can signal each other. A 2x2 m
195 topography could then be superimposed on the larger topography to target the Ulva, and finally a nanoscale topograp hy applied over everything to insure the stability of the nonwetting state as the various organisms probe the surface. The smaller topography should be present on the sidewalls as well as the top surfaces of the larger topography. The stability of the non wetted state could be further explored through extension of the AFM measurements. The AFM measurements in this dissertation were performed on fully wetted topographies. The difference in work of adhesion for wetted and nonwetted topographies can be probe d with the AFM and would give a better understanding of the importance of the wetting state of the topography for antifouling. Additionally, the energy necessary to fully wet the topography could be probed by the AFM as well by measuring a series of force curves over the same well defined location on a nonwetted topography. As the force is applied the area will wet and both the amount of force required to wet the topography as well as the change in adhesion at that site could be quantified. These measure ments would give a quantifiable observation of the difference in work of adhesion and the relative effect of different topographies depending on the scale of the structure and the size of the organism.
196 APPENDIX A MATLAB Â® IMAGE ANALYSIS A MATLAB Â® code wa s written to facilitate the analysis of the Ulva spore images. The goal of the program is to take an image of spores, identify the locations of topographic features, individual spores and spore groupings and run the mapping and RDF analysis in a streamlin ed manner. As such a number of steps go into properly analyzing the images that are discussed here. First, images must be acquired for each topography in question. The most accurate method is to take two pictures, a brightfield image where the topography is identified followed by a fluorescent image where the spores can be identified (Figure A 1). Figure A 1. Bright field and fluorescent images that could be analyzed with the MATLAB Â® program. The left image shows U. linza attached to the a5 to pography and the right image shows the spores imaged with fluorescent imaging. It is possible to separate the spores from the topography in a single brightfield image by controlling the threshold. However, experience tells us that a great number of ima ges will have topographic areas that are similar to the spores in terms of shading and will thus be inaccurately picked up by the image analysis software as spore
197 locations. The result is having to manually go through every image to discard or alter the a nalysis for those that are not properly characterized by the software which somewhat defeats the purpose of using MATLAB Â® in the first place. The names of the brightfield and fluorescent images should be numbered sequentially in order to be read by the MA TLAB Â® program, for example Brightfield (1), Brightfield (2), etc. Once the images are taken they are placed in a folder to be referenced by MATLAB Â® . The location of the folder is written into the code. Depending on the topography in question the parameters for the topography will also need to be input. The mapping algorithm developed by Long depends on the calculation of the number of unit cells b etween reference points. The number of pixels per unit cell must be added as an input into variables div1 and div2 in the program (Lines 107 and 108). There is a table of values for the 40x magnification on the Zeiss microscope in MAE 306 included in the comments of the program. Additionally, the height and width (in microns) of the unit cell also needs to be input into program into variable height and width (Lines 84 and 85). The program reads the relevant number of figures into a variable called n1 . A for loop is run with this n1 value as its limit so as to cover all of the measured images. Initially, two images are read into the program, one brightfield into variable bw with the corresponding fluorescent image read into variable Spores . The brightfi eld image is analyzed first to identify the locations of the spores. This is accomplished by adjusting the contrast of the image (this step is not always necessary but has been useful for images taken in 306) and then converted to a binary image.
198 The bina ry image of the topography includes obvious distinctions between the different features. The size of the different features, as well as their locations, is easily calculated by MATLAB using the regionprops command. Regionprops identifies relevant charact eristics for the white regions of a binary image, i.e. Area, Centroid, Orientation, etc. The images are filtered for orientation and area to remove artifacts that are present in the thresholded image. The remaining features are calculated as a function of the area. A plot of this data clearly shows different regions of area. It is necessary for the mapping to calculate reference features for both the vector coordinates as well as the individual spores. As a result all of the features of a certain type must be identified. Long 1 and Cooper 2 used the longest feature in the topography as their reference which resulted in maps with the longest feature at the 0,0 location. However, the shortest feature is the easiest to identify using this analysis, which is the reason why it acts as the 0, 0 position for the maps generated from the MATLAB Â® program. The range of smallest features is identified and the centroid for these features are stored in variable FeatureCenters . The vector coordin ates for this topography are then calculated using these feature centers. Two reference features are identified for v1 and v2 . Other features within the relevant plane are identified for both references. These two features form the basis for the calcula tion of the v1 and v2 values for the topography. After the vector coordinates are calculated the spore locations are identified. This is accomplished through the same procedure used to calculate the feature positions. Spore locations at the edges are eli minated to prevent inaccurate counting
199 through the imclearborder command and identified spore objects are filtered for area to prevent noise from being measured as a spore location. Individual spore locations are found using a watershed transform to separ ate adjacent circular objects. Groups of spores are first dilated to prevent misidentification of spore groups as individuals followed by the centroid identification sans watershed transform. Centroid positions are transformed from pixel coordinates to ve ctor coordinates using the small reference features previously identified to calculate the feature positions. From these vector coordinates the spore map is calculated using the program written by Chris Long. The spatial distribution for these spores is also calculated. A bin size is identified usually equal to the size of 1 micron for ease of analysis. The distances for each spore spore pair in each image are calculated and converted to g(r) values using the previously defined equations. Angular distr ibution values are also calculated for potential analysis at a later date. The spore location within the unit cell as well as the relevant spore locations in terms of distance and angle are saved to a multidimensional matrix where each slice represents an angle and distance and each value represents a location within the unit cell. The result is a precise map of not only spore locations but also the locations of other spores relative to that location. The data is output in several forms for analysis. A s ample analysis image is shown with the v1 and v2 vectors drawn along with the locations of the counted spores and their reference features. The spore map histogram is generated as in previous studies. Additionally, each relevant slice of the g(r) map is also displayed and saved to the MATLAB folder. These images are color coded to identify areas of relatively high
200 and low density at a given distance, which is denoted by the file name (provided the bin size is set to 1 micron). The angular information is too cumbersome to display in this manner, however it can be easily located during review of the mapped data.
201 APPENDIX B ANALYSIS OF DIFFERENCES BETWEEN WINGS AND SHOULDERS ON SILICONE RETAIN SAMPLES P roblem Statement Attachment experiments with Ulva linza pe rformed in July and September of 2013 demonstrated a significant difference between smooth surfaces cast against a glass plate or a silicon wafer. These studies showed up to 97% reduction in Ulva attachment density on smooth PDMSe cast against a silicon wa fer compared to smooth PDMSe cast against a glass plate ( Figure B 1 ). Figure B 1 . Ulva attachment density as a function of casting surface on PDMSe. A and B are different images of the same slide showing the border between the two different smooth surfaces. C shows a fluorescent micrograph of the two surfaces. Figure courtesy of John Finlay. The differences in fouling density were not expected and the two surfaces have been previously observed to be the same in terms of U. linza attachment density
202 (including in the April 2013 assay in Chapter 6). This appendix summarizes the attempts to characterize and explain the differences between the two surfaces. Analysis of Retain Samples Cont act angle/Surface energy The most obvious potential difference between the two surfaces is a change in surface chemistry, which results in a change in surface energy as measured by contact angle. Contact angle measurements were taken using a custom built video goniometer and automated drop shape analysis software. Two liquids were used (DI water and methylene iodide) with drop volume of 5 l. Dynamic contact angle was measured by tilting the sample. No statistically significant difference between static, advancing and receding contact angles was observed between control PDMSe smooth slides and retain slides from the July 2013 Ulva experiment ( Figure B 2 , Figure B 3 , Table B 1 ) Table B 1 . Average contact angle measurements on control and retain surfaces. Sample/liquid Static Advancing Receding Control/DI water 109.4 o 116.6 o 92.8 o Control/Methylene iodide 72 .2 o 74.5 o 65.9 o Wafer cast/DI water 108.4 o 116 o 91.2 o Wafer cast/Methylene iodide 72.4 o 75 o 65.1 o Figure B 2 . Contact angle measurements from the two samples. Bars are the standard deviation of the measurements. 50 70 90 110 130 Static Advancing Receding Contact Angle (degrees) Control/DI Water Retain/DI water Control/Methylene Iodid Retain/Methylene Iodide
203 Figure B 3 . Contact angle hysteresis on the two samples The two different liquids allowed the surface energy for each PDMSe sample to be calculated using the Owens Wendt method. The polar and dispersive components of surface energy were ca lculated using the average static, advancing and receding contact angles from each sample. In all cases the calculated values were similar between the control and retain surfaces, as would be expected from the contact angle measurements ( Table B 2 ). Table B 2 . Surface energy calculated from contact angle data from the control and retain samples. Polar SE (mN/m) Dispersive SE (mN/m) Control Static 0.14 21.56 Control advancing 0.04 15.91 Control receding 3.04 22.58 Retain static 0.22 21.27 Retain advancing 0.01 17.63 Retain receding 3.49 22.78 ATR FTIR Chemical analysis of the control and retain surfaces was conducted using Attenuated Total Reflectance Fourier Transform Infrared Spect roscopy (ATR FTIR). A 0 5 10 15 20 25 30 Hysteresis Control/DI Water Retain/DI water Control/Methylene iodide Retain/Methylene iodide
204 germanium crystal was used to collect the absorbance signal from five different locations on each surface. An average value was then found for the absorbance at each wave number ( Figure B 4 ) . No difference was observed between the two surfaces. Figure B 4 . ATR FTIR spectra for control and retain PDMSe samples. Mechanical Testing Several potential points of difference between control and retain samples were identi fied. Specifically, differences in the way the casting surface is treated with hexamethyl disilazane (HMDS), the method the films are mounted to the glass slides, and the casting surface itself. As such, six different test surfaces were identified ( Table B 3 ). Table B 3 . Samples identified for additional testing Cure Surface HMDS deposition before cast? Second layer of PDMSe? Glass Yes Yes Glass Yes No Glass No Yes Glass No No Silicon Yes Yes Silicon No No Tensile tests were performed on the four glass cast silicone materials identified in Table B 3 . Tests were conducted using ASTM D 1822 type L dogbone specimens at a rate of 50.8 mm/minute. Strain was measured using a laser extensometer, and load -0.05 0 0.05 0.1 0.15 0.2 0.25 600 1100 1600 2100 2600 3100 3600 Average Control Average Retain
205 through a 22 lb load cell. No statistical difference in modulus was observed for these samples ( Figure B 5 ) Figure B 5. High strain (upper) and low str ain (lower) modulus values for glass cast PDMSe films AFM AFM images of a patterned retain slide from the July 2013 assay showed a clear difference in surface morphology between the two surfaces, with an RMS roughness for the glass cast side at 10 nm and R MS roughness for the wafer cast side at 3 nm (Figures B 6 through B 9). A retain sample from July 2010 which did not have attachment differences between the two smooth surfaces showed little difference in terms of roughness from the July 2013 samples (Fig ure B 10). Table B 4. RMS roughness values for different test surfaces. Sa mple RMS Roughness (nm) 2013 Pattern 2.75 2010 Pattern 2.12 2013 Control 7.15 2010 Shoulder 5.42 2013 Shoulder 8.23 0 0.5 1 1.5 2 2.5 3 1 Layer No HMDS 1 Layer HMDS 2 Layers No HMDS 2 Layers HMDS Modulus (MPa) Modulus of glass cast films Upper Modulus Lower Modulus
206 Figure B 6. AFM images from the glass cast shoulders of retain slides from July 2013 Figure B 7. AFM images from the wafer cast center of retain slides from July 2013
207 Figure B 8. AFM height images from a smooth control retain from Ju ly 2013 Figure B 9. AFM height images from the smooth shoulders of retain sample from July 2010
208 Figure B 1 0. AFM height images from the wafer cast center of July 2010 retain samples The difference in roughness values between the surfaces (patterned surface form 2013 and 2010 in particular) were not enough to expl ain the difference in fouling density. Captive Air Bubble Measurements Captive air bubble experiments were performed to determine the effect of soaking the sample on the measured wettability. The contact angle between the wafer statistically different immediately after placement in the water (Figure B 11). The difference between the wing and shoulder regions went away after 24 hours. The air bubble contact angle increased on the wing region during this time and remained constant on the shoulder region. This data indicated the surface was changing during the test. However, subsequent sessile drop measurements in air on the same sample showed no difference between samples in terms of contact angle. Additionally, removin g the sample from the water after 24 hours and subsequently
209 replacing it and running more captive bubble measurements showed no change. While there appears to be a difference between the surfaces in the captive bubble measurement, the difference disappear s during the course of the conditioning process used before the measurements. Figure B 1 1. Measured captive air bubble contact angles on the wings and shoulders from a retain sample from July 2013. Bars indicate the standard deviation of five measurements on the same sample. Acrylate Grafting An optical difference between the wings and shoulders of silicone slides was observed after grafting using the technique modified from Chen et al . 3 . Shoulder regions appeared cloudy placed in water post grafting, while wing regions remained clear (Figu re B 12). These differences were not observed on retain samples from July 2010. Both the wing and the shoulder area were determined to have acrylate graft on the surface. The difference in optical clarity was determined to be independent of HMDS treatment of the mold, silicone base/curing agent ratio, silicone batch, silicone 60 65 70 75 80 85 90 95 100 Wing Immediate Shoulder Immediate Wing + 24 hours Shoulder + 24 hours Contact Angle (degrees)
210 degassing time, silicone stirring time, film storage and film curing time. No reproducible method was found to eliminate the difference in optical clarity between the shoulders and the wings. Figure B 1 2. Acrylate grafted silicone slide with a two layer wing region (analogous to topography mounting step) and one layer shoulder region . . The optical clarity is dif ferent between the wing and shoulder regions of the slide (indicated with arrows). Photo courtesy of Cary Kuliasha S ummary No chemical or surface energy differences were detected between the shoulder and wing regions of the retain slides. Mechanical testing of similar films showed no difference in terms of material modul us. AFM measurements showed a difference in roughness for these two surfaces, however the difference in roughness was also present on older retain samples that did not have a difference in terms of Ulva attachment. Captive bubble measurements showed a di fference between wings and shoulders immediately after submersion. The difference between the two surfaces disappeared after 24 hours of conditioning in DI water. The difference in the captive air bubble experiment indicated the surface was rearranging i tself to a more hydrophilic arrangement during the time of the experiment. Acrylate grating showed a difference in optical clarity between the wing and shoulder regions of the slide. The cause of the difference between the two surfaces was not identified by any of the analytical techniques used here. The only method found to eliminate the differences was to move
211 to a one step molding process which led to the use of the TPU molds for the April 2014 Ulva assay in Chapter 6.
212 APPENDIX C MADIN DARBY CANINE K IDNEY CELL ADHESION TO TOPOGRAPHIES B ackground T he most relevant measurement of cell adhesion stems from the actual removal of organisms from the surface. The body of the dissertation characterized cell attachment in terms of the work of adhesion. Preli minary experiments were performed using eukaryotic cells (Madin Darby Canine Kidney Cells) to determine the effects of engineered microtopography on cell adhesion strength. A rotating viscometer was used to apply shear to the cells . The shear stress for a rotating fluid is a function of the separation between the rotating disk and the surface h , the dynamic viscosity of the fluid the angular velocity , and the distance from the center of rotation r ( Equati on C 1 ) . ( C 1 ) The viscometer applied shear stress in both the parallel and perpendicular directions. The goal was to measure the difference in adhesion strength in the parallel and perpendicular direction of the topography. The visco meter also evenly varied the shear stress with radius, which allowed for the measurement of many shear stresses simultaneously. Materials and Methods Cell adhesion experiments were performed on +1, 2x2 channels and Sharklet. The channels and the Sharkle t tested here provides a contrast between an antifouling and benign topography that may yield insight into how the specific Sharklet pattern creates a difference in cell attachment by means of measuring the adhesion.
213 Cell culture was performed by Steven Ze hnder in the Angeli ni lab at the University of Florida. PDMSe surfaces were conditioned with collagen for 24 hours prior to cell seeding. Topographies were cut into 1 inch squares fixed into a petri dish with optical glue. Madin Darby Canine Kidney (MDC K) cells were allowed to attach to a surface for four hours prior to adhesion testing. Cells were dyed to facilitate counting post assay. A dextran PBS solution was created for the adhesion testing. Dextran was added to a 10 wt% amount to increase the v iscosity of the solution to 0.01 Pa*s. The angular velocity was set at 267 rad/s and the gap between cells and disk to 0.75 mm. The maximum shear stress achieved with these parameters was 25 Pa which proved to be enough to remove 95% of the cells attache d to the topographies. Post assay cells were imaged for density and compared to the average density at the center of the disk. Measurements were taken in parallel and perpendicular to the topography to compare the effect of pattern configuration on attach ment strength. Results All measured patterns lowered the a dhesion strength of the MDCK cells relative to the smooth control. Variations in cell density existed around the center of the patterned area (Figure C 1) . At approximately 10 Pa of applied stress the cells began to be removed. Normalized cell density decreased linearly on the patterned surfaces to the maximum measured shear stress of 23 Pa. No cell removal was observed on the smooth surface. There appears to be a directional dependence on the r emoval of the cells, with the perpendicular direction favoring release over the parallel direction at high shear stresses.
214 Figure C 1. Adhesion strength of MDCK cells to the Sharklet microtopography. The normalized cell density refers to the number of cells in the center of the sample where the applied shear stress was 0 Pa. Conclusion The shear stresses examined in this study were not sufficient to remove cells from the smooth PDMSe surface. As such it was not possible to draw conclusions from this w ork. Further studies are necessary to determine the magnitude to which engineered microtopography reduces adhesion strength for MDCK cells.
215 APPENDIX D CONTACT LINE IMAGES Contact line images were acquired to support the sl analysis used in Chapter 5. Here, a sample of these images is shown for reference when examining the contact angle results in that chapter. Figure D 1. Contact line geometry on the +2.6SK2x2_n3 topography. This image shows the parallel edge of the topography
216 Figure D 2. Co ntact line geometry on the +2.6SK2x2_n4 topography. This image shows the parallel edge of the topography Figure D 3. Contact line geometry on the +2.6SK2x2_n5 topography. This image shows the parallel edge of the topography
217 Figure D 4. Contact lin e geometry on the +2.6SK2x2_n7 topography. This image shows the drop shape perturbed by the pattern Figure D 5. Contact line geometry on the +2.6SK2x2_n9 topography. This image shows the drop shape perturbed by the pattern
218 Figure D 6. Contact line geometry on the +2.6CH2x2 topography. This image shows the drop following the topography into a smooth curve at the parallel edge Figure D 7. Contact line geometry on the +2.6CH12x2 topography. This image shows an elongated drop following the topog raphy
219 REFERENCES 1 Savarese, J. in Proceedings of the 14th Biennial Coastal Zone Conference, New Orleans, Louisiana. Retrieved November. 2007. 2 Schultz, M. P., Bendick, J. A., Holm, E. R. & Hertel, W. M. Economic impact of biofou ling on a naval surface ship. Biofouling 27 , 87 98 (2011). 3 Eckman, J., Savidge, W. & Gross, T. Relationship between duration of cyprid attachment and drag forces associated with detachment ofBalanus amphitrite cyprids. Marine Biology 107 , 111 118 (1990). 4 Schultz, M. P. Effects of coating roughness and biofouling on ship resistance and powering. Biofouling 23 , 331 341 (2007). 5 Gollasch, S. The importance of ship hull fouling as a vector of species introductions into the North Sea. (2002). 6 Chapman, J. W., Breitenstein, R. A. & Carlton, J. T. Port by port accumulations and dispersal of hull fouling invertebrates between the Mediterranean Sea, the Atlantic Ocean and the Pacific Ocean. Aquatic Invasions 8 (2013). 7 Godwin, L. S. Hull fouling of maritime v essels as a pathway for marine species invasions to the Hawaiian Islands. Biofouling 19 , 123 131 (2003). 8 Johnson, L. E., Ricciardi, A. & Carlton, J. T. Overland dispersal of aquatic invasive species: a risk assessment of transient recreational boating. E cological Applications 11 , 1789 1799 (2001). 9 Griffiths, R. W., Schloesser, D. W., Leach, J. H. & Kovalak, W. P. Distribution and dispersal of the zebra mussel (Dreissena polymorpha) in the Great Lakes region. Canadian Journal of Fisheries and Aquatic Sci ences 48 , 1381 1388 (1991). 10 Administration, N. O. a. A. Basic Questions about Aquaculuture , 2014). 11 Adams, C. M., Shumway, S. E., Whitlatch, R. B. & Getchis, T. Biofouling in marine molluscan shellfish aquaculture: a survey assessing the business and economic implications of mitigation. Journal of the World Aquaculture Society 42 , 242 252 (2011). 12 Fitridge, I., Dempster, T., Guenther, J. & de Nys, R. The impact and control of biofouling in marine aquaculture: a review. Biofouling 28 , 649 669 (2012). 13 Rajagopal, S., Sasikumar, N., Azariah, J. & Nair, K. V. K. Some observations on biofouling in the cooling water conduits of a coastal power plant. Biofouling 3 , 311 324 (1991).
220 14 Walker, M. E. et al . Economic impact of condenser fouling in existing the rmoelectric power plants. Energy (2012). 15 Nair, K. Marine biofouling and its control with particular reference to condenser cooling circuits of power plants An overview. Journal of the Indian Institute of Science 79 , 497 (2013). 16 Magin, C. M., Cooper, S. P. & Brennan, A. B. Non toxic antifouling strategies. Materials Today 13 , 36 44 (2010). 17 Finlay, J. A. et al . Barnacle settlement and the adhesion of protein and diatom microfouling to xerogel films with varying surface energy and water wettability. B iofouling 26 , 657 666 (2010). 18 Weinman, C. J. et al . Protein adsorption resistance of anti biofouling block copolymers containing amphiphilic side chains. Soft Matter 6 , 3237 3243 (2010). 19 Schilp, S. et al . Physicochemical properties of (ethylene glyco l) containing self assembled monolayers relevant for protein and algal cell resistance. Langmuir 25 , 10077 (2009). 20 Chen, S., Zheng, J., Li, L. & Jiang, S. Strong resistance of phosphorylcholine self assembled monolayers to protein adsorption: insights i nto nonfouling properties of zwitterionic materials. Journal of the American Chemical Society 127 , 14473 14478 (2005). 21 Wang, R., Kreuzer, H. & Grunze, M. Molecular conformation and solvation of oligo (ethylene glycol) terminated self assembled monolayer s and their resistance to protein adsorption. The Journal of Physical Chemistry B 101 , 9767 9773 (1997). 22 Harder, T. & Qian, P. Y. Waterborne compounds from the green seaweed Ulva reticulata as inhibitive cues for larval attachment and metamorphosis in t he polychaete Hydroides elegans. Biofouling 16 , 205 214 (2000). 23 Harder, T., Lam, C. & Qian, P. Y. Induction of larval settlement in the polychaete Hydroides elegans by marine biofilms: an investigation of monospecific diatom films as settlement cues. Ma rine Ecology Progress Series 229 , 105 112 (2002). 24 Maki, J., Yule, A., Rittschof, D. & Mitchell, R. The effect of bacterial films on the temporary adhesion and permanent fixation of cypris larvae, Balanus amphitrite Darwin. Biofouling 8 , 121 131 (1994). 25 Unabia, C. & Hadfield, M. Role of bacteria in larval settlement and metamorphosis of the polychaete Hydroides elegans. Marine Biology 133 , 55 64 (1999).
221 26 Callow, J. A. & Callow, M. E. Trends in the development of environmentally friendly fouling resis tant marine coatings. Nature communications 2 , 244 (2011). 27 Ista, L. K. et al . Effect of substratum surface chemistry and surface energy on attachment of marine bacteria and algal spores. Applied and environmental microbiology 70 , 4151 4157 (2004). 28 Ma gin, C. M. et al . Engineered antifouling microtopographies: the role of Reynolds number in a model that predicts attachment of zoospores of Ulva and cells of Cobetia marina. Biofouling 26 , 719 727 (2010). 29 Shea, C., Lovelace, L. & Smith Somerville, H. De leya marina as a model organism for studies of bacterial colonization and biofilm formation. Journal of industrial microbiology 15 , 290 296 (1995). 30 Hoipkemeier Wilson, L. et al . Antifouling potential of lubricious, micro engineered, PDMS elastomers agai nst zoospores of the green fouling alga Ulva (Enteromorpha). Biofouling 20 , 53 63 (2004). 31 Callow, M. E., Callow, J. A., Pickett Heaps, J. D. & Wetherbee, R. Primary adhesion of enteromopha (chlorophyta, ulvales) propagules: quantitative settlement studi es and video microscopy . Journal of Phycology 33 , 938 947 (1997). 32 Cooper, S. P. et al . Engineered antifouling microtopographies: kinetic analysis of the attachment of zoospores of the green alga Ulva to silicone elastomers. Biofouling 27 , 881 892 (2011) . 33 Callow, M. E. et al . Microtopographic cues for settlement of zoospores of the green fouling alga Enteromorpha. Biofouling 18 , 229 236 (2002). 34 Callow, J. A. & Callow, M. E. in Biological adhesives 63 78 (Springer, 2006). 35 Callow, J., Callow, M ., Ista, L., Lopez, G. & Chaudhury, M. The influence of surface energy on the wetting behaviour of the spore adhesive of the marine alga Ulva linza (synonym Enteromorpha linza). Journal of The Royal Society Interface 2 , 319 325 (2005). 36 Wustman, B. A., G retz, M. R. & Hoagland, K. D. Extracellular matrix assembly in diatoms (Bacillariophyceae)(I. A model of adhesives based on chemical characterization and localization of polysaccharides from the marine diatom Achnanthes longipes and other diatoms). Plant P hysiology 113 , 1059 1069 (1997). 37 Aldred, N., Scardino, A., Cavaco, A., De Nys, R. & Clare, A. S. Attachment strength is a key factor in the selection of surfaces by barnacle cyprids (Balanus amphitrite) during settlement. Biofouling 26 , 287 299 (2010).
222 38 Chaw, K. C., Dickinson, G. H., Ang, K. Y., Deng, J. & Birch, W. R. Surface exploration of Amphibalanus amphitrite cyprids on microtextured surfaces. Biofouling 27 , 413 422 (2011). 39 Nedved, B. T. & Hadfield, M. G. in Marine and industrial biofouling 203 217 (Springer, 2009). 40 Bloecher, N., de Nys, R., Poole, A. J. & Guenther, J. The fouling hydroid Ectopleura larynx: a lack of effect of next generation antifouling technologies. Biofouling 29 , 237 246 (2013). 41 Carl, C. et al . Enhancing the settle ment and attachment strength of pediveligers of Mytilus galloprovincialis bychanging surface wettability and microtopography. Biofouling 28 , 175 186 (2012). 42 Maruzzo, D., Conlan, S., Aldred, N., Clare, A. S. & HÃ¸eg, J. T. Video observation of surface exp loration in cyprids of Balanus amphitrite: the movements of antennular sensory setae. Biofouling 27 , 225 239 (2011). 43 Maruzzo, D., Aldred, N., Clare, A. S. & HÃ¸eg, J. T. Metamorphosis in the cirripede crustacean Balanus amphitrite. PloS one 7 , e37408 (20 12). 44 Shikuma, N. J. et al . Marine Tubeworm Metamorphosis Induced by Arrays of Bacterial Phage Tail Like Structures. Science 343 , 529 533 (2014). 45 Horiguchi, T., Shiraishi, H., Shimizu, M. & Morita, M. Imposex in sea snails, caused by organotin (tribut yltin and triphenyltin) pollution in Japan: a survey. Applied Organometallic Chemistry 11 , 451 455 (1997). 46 Bryan, G., Gibbs, P., Hummerstone, L. & Burt, G. The decline of the gastropod Nucella lapillus around south west England: evidence for the effect of tributyltin from antifouling paints. Journal of the Marine Biological Association of the United Kingdom 66 , 611 640 (1986). 47 Matthiessen, P. & Gibbs, P. E. Critical appraisal of the evidence for tributyltin mediated endocrine disruption in mollusks. Environmental toxicology and chemistry 17 , 37 43 (1998). 48 Alzieu, C. Impact of tributyltin on marine invertebrates. Ecotoxicology 9 , 71 76 (2000). 49 Gipperth, L. The legal design of the international and Europ ean Union ban on tributyltin antifouling paint: Direct and indirect effects. Journal of environmental management 90 , S86 S95 (2009). 50 Azeredo, J., Visser, J. & Oliveira, R. Exopolymers in bacterial adhesion: interpretation in terms of DLVO and XDLVO theo ries. Colloids and Surfaces B: Biointerfaces 14 , 141 148 (1999).
223 51 Hermansson, M. The DLVO theory in microbial adhesion. Colloids and Surfaces B: Biointerfaces 14 , 105 119 (1999). 52 Bostr \ "o, m. M., Williams, D. R. M. & Ninham, B. W. Specific ion ef fects: why DLVO theory fails for biology and colloid systems. Physical review letters 87 , 168103 (2001). 53 Hwang, G., Ahn, I. S., Mhin, B. J. & Kim, J. Y. Adhesion of nano sized particles to the surface of bacteria: Mechanistic study with the extended DLV O theory. Colloids and Surfaces B: Biointerfaces 97 , 138 144 (2012). 54 Harimawan, A., Zhong, S., Lim, C. T. & Ting, Y. P. Adhesion of< i> B. subtilis spores and vegetative cells onto stainless steel DLVO theories and AFM spectroscopy. Journal of collo id and interface science 405 , 233 241 (2013). 55 Leckband, D. & Israelachvili, J. Intermolecular forces in biology. Quarterly reviews of biophysics 34 , 105 267 (2001). 56 Redman, J. A., Walker, S. L. & Elimelech, M. Bacterial adhesion and transport in poro us media: Role of the secondary energy minimum. Environmental science \ & technology 38 , 1777 1785 (2004). 57 Watermann, B., Berger, H. D., SÃ¶nnichsen, H. & Willemsen, P. Performance and effectiveness of non stick coatings in seawater. Biofouling 11 , 101 118 (1997). 58 Brady Jr, R. F. & Singer, I. L. Mechanical factors favoring release from fouling release coatings. Biofouling 15 , 73 81 (2000). 59 Griffith, A. VI. The Phenomena of Rupture and F low in Solids. Phil. Trans. Roy. Soc.(Lon.) A 221 , 163 198 (1920). 60 Irwin, G. R. Analysis of stresses and strains near the end of a crack traversing a plate. J. appl. Mech. (1957). 61 Kendall, K. The adhesion and surface energy of elastic solids. Journal of Physics D: Applied Physics 4 , 1186 (2002). 62 Brady Jr, R. F. A fracture mechanical analysis of fouling release from nontoxic antifouling coatings. Progress in organic coatings 43 , 188 192 (2001). 63 Chaudhury, M. K., Finlay, J. A., Chung, J. Y., Callow, M. E. & Cal low, J. A. The influence of elastic modulus and thickness on the release of the soft fouling green alga Ulva linza (syn. Enteromorpha linza) from poly (dimethylsiloxane)(PDMS) model networks. Biofouling 21 , 41 48 (2005). 64 Wong, T. S. et al . Bioinspired s elf repairing slippery surfaces with pressure stable omniphobicity. Nature 477 , 443 447 (2011).
224 65 Epstein, A. K., Wong, T. S., Belisle, R. A., Boggs, E. M. & Aizenberg, J. Liquid infused structured surfaces with exceptional anti biofouling performance. Pr oceedings of the National Academy of Sciences 109 , 13182 13187 (2012). 66 Xiao, L. et al . Slippery Liquid Infused Porous Surfaces Showing Marine Antibiofouling Properties. ACS applied materials & interfaces 5 , 10074 10080 (2013). 67 Shivapooja, P. et al . B ioinspired surfaces with dynamic topography for active control of biofouling. Advanced Materials 25 , 1430 1434 (2013). 68 Levering, V., Wang, Q., Shivapooja, P., Zhao, X. & LÃ³pez, G. P. Soft Robotic Concepts in Catheter Design: an On Demand Fouling Release Urinary Catheter. Advanced healthcare materials (2014). 69 Shum, H., Tripathi, A., Yeomans, J. M. & Balazs, A. C. Active Ciliated Surfaces Expel Model Swimmers. Langmuir 29 , 12770 12776 (2013). 70 Balazs, A. C., Bhattacharya, A., Tr ipathi, A. & Shum, H. Designing Bio inspired Artificial Cilia to Regulate Particle Surface Interactions. The Journal of Physical Chemistry Letters (2014). 71 Tripathi, A., Shum, H. & Balazs, A. C. Fluid driven motion of passive cilia enables the layer to e xpel sticky particles. Soft Matter (2014). 72 Owens, D. K. & Wendt, R. Estimation of the surface free energy of polymers. Journal of applied polymer science 13 , 1741 1747 (1969). 73 Van Oss, C. J., Chaudhury, M. K. & Good, R. J. Interfacial Lifshitz van de r Waals and polar interactions in macroscopic systems. Chemical Reviews 88 , 927 941 (1988). 74 Baier, R. E. The role of surface energy in thrombogenesis. Bulletin of the New York Academy of Medicine 48 , 257 (1972). 75 Absolom, D. R. et al . Surface thermody namics of bacterial adhesion. Applied and Environmental Microbiology 46 , 90 97 (1983). 76 Baier, R. E. Surface behaviour of biomaterials: the theta surface for biocompatibility. Journal of Materials Science: Materials in Medicine 17 , 1057 1062 (2006). 77 B usscher, H. et al . Measurement of the surface free energy of bacterial cell surfaces and its relevance for adhesion. Applied and Environmental Microbiology 48 , 980 983 (1984).
225 78 Van Loosdrecht, M. C., Lyklema, J., Norde, W., Schraa, G. & Zehnder, A. J. Th e role of bacterial cell wall hydrophobicity in adhesion. Applied and environmental microbiology 53 , 1893 1897 (1987). 79 Bellon Fontaine, M. N., Rault, J. & Van Oss, C. J. Microbial adhesion to solvents: a novel method to determine the electron donor/elec tron acceptor or Lewis acid base properties of microbial cells. Colloids and Surfaces B: Biointerfaces 7 , 47 53 (1996). 80 Dexter, S., Sullivan, J., Williams, J. & Watson, S. Influence of substrate wettability on the attachment of marine bacteria to variou s surfaces. Applied microbiology 30 , 298 308 (1975). 81 Gerson, D. F. & Scheer, D. Cell surface energy, contact angles and phase partition III. Adhesion of bacterial cells to hydrophobic surfaces. Biochimica et Biophysica Acta (BBA) Biomembranes 602 , 506 5 10 (1980). 82 Marszalek, D. S., Gerchakov, S. M. & Udey, L. R. Influence of substrate composition on marine microfouling. Applied and environmental microbiology 38 , 987 995 (1979). 83 Bos, R., Mei, H. C. & Busscher, H. J. Physico chemistry of initial micro bial adhesive interactions its mechanisms and methods for study. FEMS microbiology reviews 23 , 179 230 (1999). 84 Bakker, D. P. et al . Bacterial deposition to fluoridated and non fluoridated polyurethane coatings with different elastic modulus and surface tension in a parallel plate and a stagnation point flow chamber. Colloids and Surfaces B: Biointerfaces 32 , 179 190 (2003). 85 van Kooten, T. G., Spijker, H. T. & Busscher, H. J. Plasma treated polystyrene surfaces: model surfaces for studying cell biomate rial interactions. Biomaterials 25 , 1735 1747 (2004). 86 Boks, N. P., Norde, W., van der Mei, H. C. & Busscher, H. J. Forces involved in bacterial adhesion to hydrophilic and hydrophobic surfaces. Microbiology 154 , 3122 3133 (2008). 87 Bakker, D. P., Postm us, B. R., Busscher, H. J. & van der Mei, H. C. Bacterial strains isolated from different niches can exhibit different patterns of adhesion to substrata. Applied and environmental microbiology 70 , 3758 3760 (2004). 88 Bos, R., van der Mei, H. C., Meinders, J. M. & Busscher, H. J. A quantitative method to study co adhesion of microorganisms in a parallel plate flow chamber: basic principles of the analysis. Journal of microbiological methods 20 , 289 305 (1994).
226 89 Bos, R., van der Mei, H. C. & Busscher, H. J . Influence of ionic strength and substratum hydrophobicity on the co adhesion of oral microbial pairs. Microbiology 142 , 2355 2361 (1996). 90 Sjollem, J., Busscher, H. J. & Weerkamp, A. H. Experimental approaches for studying adhesion of microorganisms to solid substrata: applications and mass transport. Journal of microbiological methods 9 , 79 90 (1989). 91 Van der Mei, H., Bos, R. & Busscher, H. A reference guide to microbial cell surface hydrophobicity based on contact angles. Colloids and surfaces B: B iointerfaces 11 , 213 221 (1998). 92 Callow, M. E. & Fletcher, R. L. The influence of low surface energy materials on bioadhesion a review. International biodeterioration & biodegradation 34 , 333 348 (1994). 93 Lindner, E. A low surface free energy approach in the control of marine biofouling. Biofouling 6 , 193 205 (1992). 94 Ista, L. K. & LÃ³pez, G. P. Thermodynamic analysis of marine bacterial attachment to oligo (ethylene glycol) terminated self assembled monolayers. Biointerphases 8 , 24 (2013). 95 Bennett , S. M. et al . The role of surface energy and water wettability in aminoalkyl/fluorocarbon/hydrocarbon modified xerogel surfaces in the control of marine biofouling. Biofouling 26 , 235 246 (2010). 96 Rosenhahn, A., Schilp, S., Kreuzer, H. J. & Grunze, M. T chemistry in marine biofouling prevention. Physical Chemistry Chemical Physics 12 , 4275 4286 (2010). 97 Gudipati, C. S., Finlay, J. A., Callow, J. A., Callow, M. E. & Wooley, K. L. The antifouling and fouling release perfomance o f hyperbranched fluoropolymer (HBFP) poly (ethylene glycol)(PEG) composite coatings evaluated by adsorption of biomacromolecules and the green fouling alga ulva. Langmuir 21 , 3044 3053 (2005). 98 Gudipati, C. S., Greenlief, C. M., Johnson, J. A., Prayongpa n, P. & Wooley, K. L. Hyperbranched fluoropolymer and linear poly (ethylene glycol) based amphiphilic crosslinked networks as efficient antifouling coatings: an insight into the surface compositions, topographies, and morphologies. Journal of Polymer Scien ce Part A: Polymer Chemistry 42 , 6193 6208 (2004). 99 Zhang, Z. et al . Polysulfobetaine grafted surfaces as environmentally benign ultralow fouling marine coatings. Langmuir 25 , 13516 13521 (2009).
227 100 Chiang, Y. C., Chang, Y., Higuchi, A., Chen, W. Y. & R uaan, R. C. Sulfobetaine grafted poly (vinylidene fluoride) ultrafiltration membranes exhibit excellent antifouling property. Journal of Membrane Science 339 , 151 159 (2009). 101 Aldred, N., Li, G., Gao, Y., Clare, A. S. & Jiang, S. Modulation of barnacle (Balanus amphitrite Darwin) cyprid settlement behavior by sulfobetaine and carboxybetaine methacrylate polymer coatings. Biofouling 26 , 673 683 (2010). 102 Grozea, C. M. et al . Water stable diblock polystyrene block poly (2 vinyl pyridine) and diblock poly styrene block poly (methyl methacrylate) cylindrical patterned surfaces inhibit settlement of zoospores of the green alga Ulva. Biomacromolecules 10 , 1004 1012 (2009). 103 Magin, C. M. et al . Antifouling performance of cross linked hydrogels: refinement of an attachment model. Biomacromolecules 12 , 915 922 (2011). 104 Zhou, M. et al . High throughput discovery of new fouling resistant surfaces. Journal of Materials Chemistry 21 , 693 704 (2011). 105 Wenzel, R. N. Resistance of solid surfaces to wetting by wat er. Industrial & Engineering Chemistry 28 , 988 994 (1936). 106 Cassie, A. B. D. & Baxter, S. Wettability of porous surfaces. Transactions of the Faraday Society 40 , 546 551 (1944). 107 Brandon, S. & Marmur, A. Simulation of contact angle hysteresis on chem ically heterogeneous surfaces. Journal of colloid and interface science 183 , 351 355 (1996). 108 Schmidt, D. L., Brady, R. F., Lam, K., Schmidt, D. C. & Chaudhury, M. K. Contact angle hysteresis, adhesion, and marine biofouling. Langmuir 20 , 2830 2836 (200 4). 109 Ucar, I. O. et al . Effect of contact angle hysteresis on the removal of the sporelings of the green alga Ulva from the fouling release coatings synthesized from polyolefin polymers. Biointerphases 5 , 75 84 (2010). 110 Verran, J. & Boyd, R. D. The r elationship between substratum surface roughness and microbiological and organic soiling: a review. Biofouling 17 , 59 71 (2001). 111 Long, C. J., Finlay, J. A., Callow, M. E., Callow, J. A. & Brennan, A. B. Engineered antifouling microtopographies: mapping preferential and inhibitory microenvironments for zoospore attachment. Biofouling 26 , 941 952 (2010). 112 Scardino, A., Harvey, E. & De Nys, R. Testing attachment point theory: diatom attachment on microtextured polyimide biomimics. Biofouling 22 , 55 60 ( 2006).
228 113 Scardino, A. J., Guenther, J. & De Nys, R. Attachment point theory revisited: the fouling response to a microtextured matrix. Biofouling 24 , 45 53 (2008). 114 Wu, A. H., Nakanishi, K., Cho, K. & Lamb, R. Diatom attachment inhibition: limiting su rface accessibility through air entrapment. Biointerphases 8 , 1 10 (2013). 115 Friedlander, R. S. et al . Bacterial flagella explore microscale hummocks and hollows to increase adhesion. Proceedings of the National Academy of Sciences 110 , 5624 5629 (2013). 116 Schumacher, J. F. et al . Engineered antifouling microtopographies effect of feature size, geometry, and roughness on settlement of zoospores of the green alga Ulva. Biofouling 23 , 55 62 (2007). 117 Long, C. J. et al . A model that predicts the attachme nt behavior of Ulva linza zoospores on surface topography. Biofouling 26 , 411 419 (2010). 118 Efimenko, K., Finlay, J., Callow, M. E., Callow, J. A. & Genzer, J. Development and testing of hierarchically wrinkled coatings for marine antifouling. ACS applie d materials & interfaces 1 , 1031 1040 (2009). 119 Watson, G. S., Cribb, B. W. & Watson, J. A. How micro/nanoarchitecture facilitates anti wetting: an elegant hierarchical design on the termite wing. Acs Nano 4 , 129 136 (2010). 120 Bai, X. et al . Study on b iomimetic preparation of shell surface microstructure for ship antifouling. Wear 306 , 285 295 (2013). 121 Greco, G. et al . Microtopography of the eye surface of the crab Carcinus maenas: an atomic force microscope study suggesting a possible antifouling po tential. Journal of The Royal Society Interface 10 , 20130122 (2013). 122 Baum, C., Meyer, W., Stelzer, R., Fleischer, L. G. & Siebers, D. Average nanorough skin surface of the pilot whale (Globicephala melas, Delphinidae): considerations on the self cleani ng abilities based on nanoroughness. Marine Biology 140 , 653 657 (2002). 123 Decker, J. T. et al . Engineered antifouling microtopographies: an energetic model that predicts cell attachment. Langmuir 29 , 13023 13030 (2013). 124 Schumacher, J. F. et al . Spec ies specific engineered antifouling topographies: correlations between the settlement of algal zoospores and barnacle cyprids. Biofouling 23 , 307 317 (2007). 125 Carman, M. L. et al . Engineered antifouling microtopographies correlating wettability with cel l attachment. Biofouling 22 , 11 21 (2006).
229 126 Chung, K. K. et al . Impact of engineered surface microtopography on biofilm formation of Staphylococcus aureus. Biointerphases 2 , 89 94 (2007). 127 Schumacher, J. F. et al . Engineered nanoforce gradients for i nhibition of settlement (attachment) of swimming algal spores. Langmuir 24 , 4931 4937 (2008). 128 Schumacher, J. F. et al . Engineered nanoforce gradients for inhibition of settlement (attachment) of swimming algal spores. Langmuir 24 , 4931 4937 (2008). 129 Scardino, A. J. & de Nys, R. Mini review: Biomimetic models and bioinspired surfaces for fouling control. Biofouling 27 , 73 86 (2011). 130 Efimenko, K., Finlay, J., Callow, M. E., Callow, J. A. & Genzer, J. Development and testing of hierarchically wrinkl ed coatings for marine antifouling. ACS Applied Materials \ & Interfaces 1 , 1031 1040 (2009). 131 Hoipkemeier Wilson, L. et al . Antifouling potential of lubricious, micro engineered, PDMS elastomers against zoospores of the green fouling alga Ulva (Enteromo rpha). Biofouling 20 , 53 63 (2005). 132 Xiao, L. et al . Hot embossed microtopographic gradients reveal morphological cues that guide the settlement of zoospores. Langmuir (2012). 133 Baxter, R. J. Exactly solved models in statistical mechanics . (DoverPubl ications. com, 2007). 134 Carl, C., Poole, A. J., Williams, M. R. & de Nys, R. Where to Settle Settlement Preferences of Mytilus galloprovincialis and Choice of Habitat at a Micro Spatial Scale. PloS one 7 , e52358 (2012). 135 Marmur, A. Underwater superhyd rophobicity: theoretical feasibility. Langmuir 22 , 1400 1402 (2006). 136 Magin, C. M. et al . Antifouling Performance of Cross linked Hydrogels: Refinement of an Attachment Model. Biomacromolecules (2011). 137 Carman, M. L. et al . Engineered antifouling mic rotopographies -correlating wettability with cell attachment. Biofouling 22 , 11 21 (2006). 138 Long, C. J., Schumacher, J. F. & Brennan, A. B. Potential for tunable static and dynamic contact angle anisotropy on gradient microscale patterned topographies. Langmuir 25 , 12982 12989 (2009). 139 Schultz, M., Bendick, J., Holm, E. & Hertel, W. Economic impact of biofouling on a naval surface ship. Biofouling 27 , 87 98 (2011).
230 140 Kirschner, C. M. & Brennan, A. B. Bio inspired antifouling strategies. Annual Revie w of Materials Research 42 , 211 229 (2012). 141 Genzer, J. & Efimenko, K. Recent developments in superhydrophobic surfaces and their relevance to marine fouling: a review. Biofouling 22 , 339 360 (2006). 142 Granhag, L., Finlay, J., Jonsson, P. R., Callow, J. & Callow, M. Roughness dependent removal of settled spores of the green alga Ulva (syn. Enteromorpha) exposed to hydrodynamic forces from a water jet. Biofouling 20 , 117 122 (2004). 143 Finlay, J. A. et al . Settlement of Ulva zoospores on patterned fluo rinated and PEGylated monolayer surfaces. Langmuir 24 , 503 510 (2008). 144 Xiao, L. et al . Hot Embossed Microtopographic Gradients Reveal Morphological Cues That Guide the Settlement of Zoospores. Langmuir 29 , 1093 1099 (2013). 145 Decker, J. et al . Engine ered Antifouling Microtopographies: An Energetic Model that Predicts Cell Attachment. Langmuir (2013). 146 Puliafito, A. et al . Collective and single cell behavior in epithelial contact inhibition. Proceedings of the National Academy of Sciences 109 , 739 7 44 (2012). 147 Treiser, M. D. et al . Cytoskeleton based forecasting of stem cell lineage fates. Proceedings of the National Academy of Sciences 107 , 610 615 (2010). 148 Schultz, M. P., Finlay, J. A., Callow, M. E. & Callow, J. A. A turbulent channel flow a pparatus for the determination of the adhesion strength of microfouling organisms. Biofouling 15 , 243 251 (2000). 149 Finlay, J., Callow, M. E., Schultz, M., Swain, G. & Callow, J. Adhesion strength of settled spores of the green alga Enteromorpha. (2002) . 150 Finlay, J. A., Fletcher, B. R., Callow, M. E. & Callow, J. A. Effect of background colour on growth and adhesion strength of Ulva sporelings. Biofouling 24 , 219 225 (2008). 151 Stafslien, S. J. et al . Combinatorial materials research applied to the d evelopment of new surface coatings I: a multiwell plate screening method for the high throughput assessment of bacterial biofilm retention on surfaces. Journal of Combinatorial Chemistry 8 , 156 162 (2006). 152 Casse, F. et al . Combinatorial materials resea rch applied to the development of new surface coatings V. Application of a spinning water jet for the semi high throughput assessment of the attachment strength of marine fouling algae. Biofouling 23 , 121 130 (2007).
231 153 Chisholm, B. J. et al . Combinatoria l materials research applied to the development of new surface coatings: VIII: Overview of the high throughput measurement systems developed for a marine coating workflow. Applied Surface Science 254 , 692 698 (2007). 154 Stafslien, S., Daniels, J., Chishol m, B. & Christianson, D. Combinatorial materials research applied to the development of new surface coatings III. Utilisation of a high throughput multiwell plate screening method to rapidly assess bacterial biofilm retention on antifouling surfaces. Biofo uling 23 , 37 44 (2007). 155 Stafslien, S. et al . Combinatorial materials research applied to the development of new surface coatings IV. A high throughput bacterial biofilm retention and retraction assay for screening fouling release performance of coating s. Biofouling 23 , 45 54 (2007). 156 He, J. et al . Combinatorial materials research applied to the development of new surface coatings X: a high throughput electrochemical impedance spectroscopy method for screening organic coatings for corrosion inhibition . Journal of Combinatorial Chemistry 10 , 704 713 (2008). 157 Hook, A. L. et al . Combinatorial discovery of polymers resistant to bacterial attachment. Nature biotechnology 30 , 868 875 (2012). 158 Chisholm, B. J., Christianson, D. A. & Webster, D. C. Combin atorial materials research applied to the development of new surface coatings: II. Process capability analysis of the coating formulation workflow. Progress in organic coatings 57 , 115 122 (2006). 159 Chisholm, B. J. et al . Combinatorial materials research applied to the development of new surface coatings VII: An automated system for adhesion testing. Review of Scientific Instruments 78 , 072213 072213 072219 (2007). 160 Kugel, A. J. et al . Combinatorial materials research applied to the development of new surface coatings XII: Novel, environmentally friendly antimicrobial coatings derived from biocide functional acrylic polyols and isocyanates. Journal of Coatings Technology and Research 6 , 107 121 (2009). 161 Unadkat, H. V. et al . An algorithm based topogr aphical biomaterials library to instruct cell fate. Proceedings of the National Academy of Sciences 108 , 16565 16570 (2011). 162 Petrone, L. et al . Effects of surface charge and Gibbs surface energy on the settlement behaviour of barnacle cyprids (Balanus amphitrite). Biofouling 27 , 1043 1055 (2011). 163 Maki, J., Rittschof, D., Costlow, J. & Mitchell, R. Inhibition of attachment of larval barnacles, Balanus amphitrite, by bacterial surface films. Marine Biology 97 , 199 206 (1988).
232 164 D., O. & McCarthy, T. J. Ultrahydrophobic surfaces. Effects of topography length scales on wettability. Langmuir 16 , 7777 7782 (2000). 165 Khan, M. M. T., Ista, L. K., Lopez, G. P. & Schuler, A. J. Experimental and theoretical examination of surface energy and adhesion of nitr ifying and heterotrophic bacteria using self assembled monolayers. Environmental science & technology 45 , 1055 1060 (2010). 166 Butt, H. J., Cappella, B. & Kappl, M. Force measurements with the atomic force microscope: Technique, interpretation and applica tions. Surface science reports 59 , 1 152 (2005). 167 Christ, K. V. & Turner, K. T. Methods to measure the strength of cell adhesion to substrates. Journal of Adhesion Science and Technology 24 , 2027 2058 (2010). 168 Lu, H. et al . Microfluidic shear devices for quantitative analysis of cell adhesion. Analytical chemistry 76 , 5257 5264 (2004). 169 Butt, H. J. & Jaschke, M. Calculation of thermal noise in atomic force microscopy. Nanotechnology 6 , 1 (1995). 170 Choi, W., Tuteja, A., Mabry, J. M., Cohen, R. E. & McKinley, G. H. A modified Cassie Baxter relationship to explain contact angle hysteresis and anisotropy on non wetting textured surfaces. Journal of colloid and interface science 339 , 208 216 (2009). 171 Scheen, G. et al . Simultaneous fabrication of sup erhydrophobic and superhydrophilic polyimide surfaces with low hysteresis. Langmuir 27 , 6490 6495 (2011). 172 Schulte, A. J., Droste, D. M., Koch, K. & Barthlott, W. Hierarchically structured superhydrophobic flowers with low hysteresis of the wild pansy ( Viola tricolor) new design principles for biomimetic materials. Beilstein journal of nanotechnology 2 , 228 236 (2011). 173 Krumpfer, J. W. & McCarthy, T. J. Contact angle hysteresis: a different view and a trivial recipe for low hysteresis hydrophobic surf aces. Faraday discussions 146 , 103 111 (2010). 174 Johnson, K., Kendall, K. & Roberts, A. Surface energy and the contact of elastic solids. Proceedings of the royal society of London. A. mathematical and physical sciences 324 , 301 313 (1971). 175 Tabor, D. Surface forces and surface interactions. Journal of Colloid and Interface Science 58 , 2 13 (1977).
233 176 Boccaccini, A., Thomas, B., Brusatin, G. & Colombo, P. Mechanical and electrical properties of hot pressed borosilicate glass matrix composites contain ing multi wall carbon nanotubes. Journal of materials science 42 , 2030 2036 (2007). 177 Greenwood, J. Adhesion of elastic spheres. Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences 453 , 1277 1297 (1997). 178 Chen, J. J., Struk, K. N. & Brennan, A. B. Surface Modification of Silicate Glass Using 3 (Mercaptopropyl) trimethoxysilane for Thiol Ene Polymerization. Langmuir 27 , 13754 13761 (2011).
234 BIOGRAPHICAL SKETCH Joseph Thomas Decker was born in Green Bay , Wisconsin in 1988 to Tom and Jill Decker. He spent his childhood in Green Bay where he enjoyed playing s ports and spending time in the n orthwoods with his family. Joseph attended Valley View elementary and Parkview mid dle school before attending and finally graduating from Ashwaubenon High School in 2006. Joseph developed an interest in medicine and engineering while in high school and went the University of Wisconsin Madison to study biomedical engineering in the fall of 2006. While at the University of Wisconsin he worked as a teaching assistant for freshman engineering design and human physiology. In the summer of 2008 he had the opportunity to travel to Hangzhou, China to study thermodynamics and technical writing. He began his first research project as an intern at the USDA Forest Products Lab studying the effect of humidity on postage stamps. Joseph received a Graduate School Fellowship to attend the University of Florida to pursue a doctorate Materials Science a 2010. In 2012 he married his wife, Ann, and held the reception on the UF campus. He was involved in the Society for Biomaterials while at UF, serving as Treasurer and President and helping to found the annual Bi omaterials Day. He has had the opportunity to teach numerous labs and classes and has given many presentations at both local and national conferences during his time as a graduate student. After graduation, Joseph will be joining the lab of Peter Ma in t he School of Dentistry at the University of Michigan.