Final Report - Materials Science and Engineering - University of

33
Capstone 2011: Microfluidic Lab‐on‐a‐Chip for Delivering Targeted Alpha Therapy Presented to the Materials Science and Engineering Department, University of Maryland by: Tricia Alward, Tunji Godo, Coit Hendley, Iain Kierzewski, Michael Meadows, Ninoska Moratin, William Schoenfelder, Nicholas Strnad, Robert Thompson

Transcript of Final Report - Materials Science and Engineering - University of

Capstone2011:MicrofluidicLab‐on‐a‐ChipforDeliveringTargetedAlphaTherapy

PresentedtotheMaterialsScienceandEngineeringDepartment,UniversityofMarylandby:

TriciaAlward,TunjiGodo,CoitHendley,IainKierzewski,MichaelMeadows,NinoskaMoratin,WilliamSchoenfelder,NicholasStrnad,RobertThompson

2

TableofContents

Title1

TableofContents2

Abstract3

Motivation3

MaterialsScienceandEngineeringAspectsandReviewofPriorWork3

DesignGoals4

TechnicalApproach5

EthicsandEnvironmentalConcerns9

IntellectualMerit10

BroaderImpact11

SimulationResultsandDiscussion11

Conclusions30

AcknowledgementsandReferences31

3

IAbstract:

Thegoalofthisprojectistodesignamicrofluidicdevicethatwilltreattumorcells.TargetedAlphaTherapy(TAT)isusedtodelivertheradiationsourceasindividualradioisotopestothedesiredtumorcellsinthebody.Thechelatingagent,DTPA,willbemixedwithamonoclonalantibody(mAb)inasolutionofwater.Actinium225,whichhasahalf‐lifeof10days,isthentransferredtothedeviceusingalpharecoil,resultinginimplantedBismuth213.BasicmodelingwasdoneforthedissolutionofthesucrosefilmthatcapturestheBismuth,andFluentwasusedtomodelthemixingbehaviorofthefluid.

IIMotivation:

OurmotivationforthisprojectistodesignamicrofluidicdevicethatwilladministerTargetedAlphaTherapy(TAT)quicklyandefficientlytoapatient.TATitselfiscurrentlyinclinicaltrialsforabroadrangeofcancers,butisextremelyexpensive,withthecostofonedoserangingfrom$10,000‐$40,000.AnotherimportantproblemwithTATistheneedtohaveageneratoron‐sitebecauseoftheshorthalflifeoftheradio‐isotope,Bismuth213(45min).Alab‐on‐a‐chipdevicewouldsignificantlyreducethecost(to$7.77pergenerator)andmakethetreatmentavailabletoamuchwiderrangeofhospitalsandasaresult,patients.

IIIMaterialsScienceEngineeringAspectsandImportantPriorWork:

Asmaterialsscienceandengineeringstudentsourmainobjectivewastodesignapracticalsystemwithanemphasisontheinterconnectionbetweenthestructuresofthematerialsweintendedtouse,thewaywewouldprocessthembasedonourdesign,thepropertiestheywouldhaveasaresultoftheprocessingandthewaythematerialsperformedbasedontheirproperties.

Ourdesignrequiredasucrosefilmthathadathicknessof1μmandbasedonour

discussionwithDr.PhaneufandwhatwelearnedfromkineticsENMA471,apolycrystallinefilmwithasmallcrystallitesizewouldbedesirablebecausethesepropertieswillaidthedissolutionrateofsucrosemicroprocessing.BasedonourskillsacquiredfromtakingENMA465,aclassonnanoscaleandmicroscaleprocessingofmaterialswithanemphasisonthinfilmprocessingforadvancedtechnologies,wenowspincoatedsucroseonasiliconwafertomakeafilmof1μmthicknessattheFabLaboncampus.

Materialscharacterization:Wenowhadtoverifyifthefabricationprocessyieldedthe

propertiesofthesucrosefilmwedesired,sobasedonwhatwelearnedfromENMA310,amaterialslaboratory,weusedx‐raydiffractiontodeterminethecrystalstructureand

4

crystallitesizeofthesucrosefilm.Wehadalsolearnedaboutthetheoreticalprinciplesofx‐raydiffractionfromENMA460,solidstatephysics.

OurknowledgeofkineticsandourdiscussionwithDr.Phaneufenabledusto

understandthattheprecipitationandthenheterogeneousnucleationatthesurfaceofthefilmwouldbegovernedbykineticsandthermodynamics,ENMA461.Theteam’sknowledgeofsolidstatephysicsalsoenabledustounderstandthatphononsaregeneratedthroughcollisionsofBiatomsintosucrosefilmbysimulatingionimplantation.ThekineticMonteCarlosimulationofnucleardecayandtimestep,thediffusivityofBiintocollectionareaandthediffusionandconvectionofBismuthacrosschannelsareallmaterialsscienceandengineeringaspectsbasedonkinetics.

TargetedAlphaTherapyinvolvingtheuseofradioisotopeslikeAc225andBi213forbiomedicalapplicationsandparticularlycancerhasbeeninvestigatedextensivelybythescientificcommunity[20].Itiscurrentlyinclinicaltrialsandhasbeenproventowork.TheapplicationofBi213astheradionuclidetobedeliveredasthedosetothepatientanditsgenerationfromAc225isalsowelldocumentedinliterature[22].Thistherapytargetsthetumorcellwithinthebody.Theantibodiesareadministeredtothebodywhichtargetandbindtothetumorcellsandtheradioisotopedecaysandbecomesbenignafterreleasinganalphaparticlewhichkillspartsofthetumorcellperdose.Theseparationofthealpha‐emittingisotopeisdonethroughrecoil‐ionseparation.Asthedecayproceeds,Th229producesanalphaparticleandrecoilRa225ion.Thisoccursthroughconservationofmomentumandenergy.ThisenergyisenoughtodislodgetheRa225ionintoaneighboringsubstrate.ThenthedecaycontinuesonuntilbecomingstableatBi213,whichisbenigntothebody.[18]Thistreatmenthasarangeof70microns.TAThasbeentestedagainstmelanoma,leukemia,colorectal,breast,ovarian,prostate,andpancreaticcancers.ThereactionkineticsofBi213withthechelationagent,DPTAhasalsobeenwelldocumented[23].ThekineticsofadsorptionanddesorptionofBismuthfromaqueoussolutionswasalsostudied[25].Thekinetics,therateofdiffusionandtheroleoftheNernstlayerinunderstandingthedissolutionprocessofsucrosehasalsobeenofinteresttoresearchers[2].ThemathematicalcalculationoftheDiffusioncoefficientinliquidssuchaswaterhasalsobeendoneinthepast[24].Theroleofthecrystallitesizeintherateofsucrosedissolutionandunderstandingtheroleofthephasechangeandmicrostructureofthesucrosefilmwasalsofoundinliterature[26].Thedifferencebetweenthecurrenttechnologyandourprojectisthesizeandcontainmentofthedose.Usinglab‐on‐a‐chiptoadministerthistherapymakesitmorewidelyavailable,affordableandabletobemassproduced.

IVDesignGoals:

Ourgoalsforthisdeviceareprimarilylimitedbytime,becauseweonlyhave3halflives(2.25hours)tomakeanddelivertheradio‐labeledantibodysuccessfullytothepatient.However,westillhaveseveralsubstantialgoalsforthedeviceasawhole.Becauseitisamicrofluidicdevicewemustthereforebeabletofitseveralonastandardwafer,nomorethan3.5squarecentimetersforthedissolutionchamberandmixingchannels.Additionally,inorder

5

tomaximizetheoveralldeviceefficiency,andgivenwehaveagoodamountofsizetoworkwith;wemake100%mixingourgoalwithinthedevice.Also,sinceweplanonusinganIVpump,thedevicemustbeabletohandleavolumeflowrateof5‐20mL/hr.Thesegoalsareallcomparablewithpreviousdevices[3,4,5,8,16].However,thegoalsforthenewestpartofourdesign,thesucrosedissolutionchamber,havebeensetbasedonourmodeling.WewantthesucroselayertodissolvequicklyenoughsothattheBismuthcaneasilydiffuseacrossthesystemandbemixedwithplentyoftimetoreactwiththechelate‐antibodyconjugate(lessthan2or3minutes).

VTechnicalApproach:

OakRidgeNationalLabsellsActinium225for$1,450permilliCurie,whichgreatlysimplifiestheprocessofpurificationbeforeinsertionintothedevice.Wewillonlyneedtodoonestepofrecoilseparation:fromtheactiniumthin‐filmgeneratortothesucrosecollectionlayer.Inordertoprovethatoursystemisfeasibleintermsofthismaterialcost,wedidasimulationtocalculatetheavailablequantityofBismuthfordosages.

ThefirststepinknowinghowmanyBismuthatomshavebeenimplantedintothesugaristoknowhowmanyatomshaverecoiledinaparticulartimeintervalwithaparticularstartingamountofAc.Thesimplesolutiontothisproblemistostartfromafilmof100%pureAcandprojectthequantitiesofeachrespectivedaughterastheycontinuedownthechain.TheresultingchartisaPoissondistribution,andsincetheAchasthesmallestdecayrate,thisdecaybecomesthelimitingfactor.

Thekineticsoftherecoilseparationprocessarenontrivialandresultinacertainlossbetweenthegeneratorandthesugar.Howeverthissimulationisconcernedonlywiththe'ceiling'ofavailableBiasdecidedbytheexponentialnucleardecay.Thusfarintheproject,thenumberscalculatedherearemostusefulforfeasibilityestimates,suchasthepotentialsizeofthesucrosefilmonthechip,ortheestimatedcostoftheAcperchip.

Sincetheisotopesareconstantlyinastateoffluxfromhighertolowerorderisotope(atleastuntilBi209),wecreatedabasicKineticMonteCarlosimulationtomodelindividualatoms.Thetimethatacertainatomexistsasanisotopebeforedecayingisdefinedusingthelogarithmofarandomnumberdividedbytheconstantdecayrate,λ.

(4)

6

Bysimulating10millionatoms,wecanseethattheamountofBi213existinginthemixtureofelementsapproachesasteady‐stateatabout3ordersofmagnitudelessthanthequantityofActinium.ThisnumberallowsustocalculatethemaximumamountofBiinthesugarfilmasafunctionofthestartingamountofAcinthegenerator.

Generator‐CollectorInteractionTheprevioussectiondetailsthetheoreticalmaximumamountofactiveisotopethatis

availabletooursystem,whichwecanholdconstantduetothelimitsofthedecayrates.WhilewecanchangethestartingamountofAc225inourgenerator,theavailableBi213willalwaysbeafractionofthatquantity.

Inordertonowunderstandthelossofthealpha‐recoilmethod,wesimulatethe

interactionwiththeenergeticBi213ionsastheytravelthroughthethreeregimesoftherecoilsteponthechip:thegeneratorwhichholdstheAc225,theairthatseparatesthetwofilms,andthesucroselayerthatcollectstheions.Theenergyoftherecoilparticlesisaconstantduetothenatureofthealphadecay‐thedecayreleasesenergyfrom5.6to5.8MeV,andduetoconservationofmomentumtherecoilparticlemusthaveanenergyfrom101to105keV.

Figure1:Simulatedratioofisotopedensities,beginningwithaquantityof1unitofAc225.Thesteady

stateregimeoftheBi213isotopeisseenat2.5hours.

7

UsingthepackageStoppingandRangeofIonsinMatter(SRIM),wecanfindthepenetrationdepthvalueofBiionsat100keVtravellingthrougheachofourthreeregimes.Wecanusethedepthvaluestocalculateanenergylossperdistancetravelled,dE/dx,orstoppingpower.InourPMMAgeneratorfilm,thestoppingpowerwasfoundtobe1380keV/µm;inair,1.3keV/µm,andinsucrose,1250keV/µm.

UsingMatlab,wesimulatedthedimensionsofthetwofilmsastheywouldbewhilethe

sucroselayerwas‘charging’onthegenerator,asseeninfigure2a.

Assumingthattheparticlesdecayinrandomdirections,weknowthatmorethanhalfof

theparticleswillfireinthewrongdirection.Eithertheparticlesgointhedirectionoppositetothecollectionplaneortheyareclosetotheedgeofthegeneratorandfireoutwards.Sincetheaspectratiooftheplaneareaversusthedistanceisverylarge,closeto200,wecanclassify

Figure2:(a):SchematicofCollector‐GeneratorArrangementshowingrecoilandthefacethatisdisplayedintheMonteCarloSimulation(b)MonteCarloSimulationshowingrelativedensitiesfortwo

distinctgeneratorthicknesses.Simulationwasbasedonstoppingpowersfoundinliterature.

(a)

(b)

8

thesecasesassomewhattrivial.Thisimmediatelylowersourcollectionefficiencytoroughlyhalfoftheparticlesthatdecayonthegenerator.Furthermore,weknowthatacertainpercentofionswilldecaywithadirectionveryclosetoparalleltotheplates,andsomewillneverhaveenoughenergytotraversethedistancefromthebottomofthegeneratortothetop.Thesecasesalsoconstitutelosses.ThewaytominimizetheselosseswassimulatedbyreducingthefilmthicknessofthePMMAgenerator.Bychangingthisdepth,particleshaveamuchhigherprobabilityofleavingthefilm.Withathicknessof50nm,approximately20.3%oftheionsremainstuckinthegenerator.Witha10nmfilm,thisnumberisreducedto4%.SimilarlywecanreducethestoppingpowerdE/dxordensityofthegeneratorfilmtomaximizeourionfluxbetweenfilms.Therelativedensityofionsinthethreerangesaregiveninfigure2.

Forthe10nmgeneratorfilm,still41.4%oftheionslosetheirenergywhileinthespace

(air)betweenthefilms.Howeversincetheyarechargedparticlesandthesucrosefilmisbiasedto‐200V,wecanassumethatthemajorityoftheionswillmoveacrossthegapwiththeelectricfield.BehaviorofBiIsotopewithinChannelsAfterthesucrosefilmhasbeendissolved,thefluidbeginsdownthechannelpathinthesucrose.Thegoalofthechannelsistomixthebismuthwiththechelatedantibodyasthoroughlyaspossible.Originallywewantedconvectivemixing,butthisisimpossibleonthelengthandflowratescalesdesiredonsuchasmallchip.Theflowconditionsinsidethechannelareentirelylaminar,whereviscousforcesdominateoverinertial.Sincethefluidismovingslowly,wecannotassumeconvectivemixingofthebismuth.Instead,weusedtheHayduk‐LaudierelationtoapproximatethediffusivityoftheBismuthionswithinthewater.Withthiswecangetanapproximatemasstransfer.

Hayduk‐Laudieequation ThediffusivityallowsustosolveFick’s2ndlawandcreateaconcentrationcurveasafunctionoftimeanddistancefromthebottomofthechannel.Sincethesourceofbismuthisfinite,thesolutiontoFick’slawisagaussianfunctionandapproachesauniformconcentrationwithtime.Ourdiffusivitywashigh,soafteronlythreeminuteswecanassumeanearlyuniformgradient.Fick’sandGaussianEQs

9

VIEthicsandEnvironmentalConcerns:

Thesocietalimpactrepresentedbysuccessfulimplementationofthissystemforcancertreatmentisobvious.Targetedalphatherapyhasbeenshowntobeeffectiveintreatingcancerbutbecauseoftheneedforaradioisotopesourceinhouseithasprovenveryexpensiveandthefacilitieswhichcanprovidesuchtreatmentaregreatlylimited.Oursystemwouldcentralizetheproductionofradioactivespeciesandthenecessarymaterialstoadministertargetedalphatherapyintoonecompactmicro‐fluidicsystem.Despitethesemajorsocietalbenefitsthefactthatoursystemisdisposablepresentsethicalconcernsintheformofwastegeneration.Materialsthatwereimplementedintothedesignwereevaluatedtoensurethattheyarenotharmfultotheenvironmentandwillnotcauseanylongtermdisposalissues.Polydimethylsiloxane(PDMS)isusedinthechanneldesign.PDMSdoesnotposeanyenvironmentalthreatswhendisposed.PDMShasshowntodegradetolowermolecularweightcompoundswhenincontactwithsoil.ItwilldegradetoMe2Si(OH)2afteronlyafewweeks[21].WeusedSiliconbasedwafersforthefabricationofthechip.Siliconisnotknowntocause

Figure3:SimulationofBismuthConcentrationGradientacrossthechannellengthbasedontheDiffusivityfromHaydukLaudieequationandassumptionoffinitesource.

10

anyadverseeffectsontheenvironmentortothebodybecauseitisaninertmaterial.Waterandsucrosearenotharmful.SmallamountsorUVepoxycontainingurethaneacrylatewereusedinthedesignofthechip.Urethaneacrylateifreleasedintoanyaquaticenvironmentcancauselongtermeffectsinaquaticorganisms.TheMaterialsSafetyDataSheetforurethaneacrylatestatesthatitisimportanttoavoiddisposalintodrains,soilorsurfacewater.Disposalofthisagentissubjectofregulationsandrequireschemicaldisposalthatensuresregulatorycompliance.SECgelisagarbased,andsobiodegradesreadily.TheBi213isotopedegradesreadilytoanon‐radioactivespeciesBi209whichisnon‐toxic,nonbio‐accumulative,andistheleasttoxicofallheavymetals.Theconcentrationofitissosmallthatitwillnotposeanysignificantdangers.TheisotopegeneratorisprovidedbyOakridgeNationalLab(ORL)andisembeddedinPMMA.Appropriateprecautionsshouldbetakenwhenthegeneratorhastobedisposedofbecauseofremainingradioactivespeciesandtheirheavymetaldecayproductsthatcouldremainasremnantsinthegenerator.Thebiologicaleffectswereconsideredthoughtthedesignprojecttoensurethatallmaterialsthatweregoingtobeinjectedintothebodyarecompatible.Thisisveryimportantbecausepatientscouldhavesideeffectsorpossibledeathifthematerialswerenotdeemedcompatible.Overallthechipdesignandprocessprovesnottoinduceanymajorenvironmentaleffectsorsafetyproblems.

VIIIntellectualMerit:

Priortothiswork,theuseofalpharadiationforcancertreatmentshasbeenlimitedtoonlyahandfulofhospitalswhichcanproducetheseshortlivedsourceson‐site.Themainintellectualmeritofthisprojecthasbeentheconsiderationofmaterialsscience,nuclearphysics,andfluiddynamicstostreamlinetheprocessintoamicrofluidicsystem.

Therecoilseparationprocess,withwhichwedesignedourisotopeseparationgroup,hasverylittlepreviousresearch.Weinvestigatedthepossibilityofusingitonalab‐on‐chipanddecideditdefinitelyhasthepossibilityofcompetingwithresin‐orstrongacid‐basedsystems.Itstillhasmajordetractions:halfoftheionsrecoilawayfromthecollectorfilm,andthelackofconvectivefluidmovementinsidethechannelsmakesmixingverydifficult.Recoilseparationisanunresolvedtopic;mostlybecausethealphaemittersthatarerequiredtotestsuchasystemarerareandexpensive.OurtreatmentofthesubjectmakesastrongargumenttodelegateacertainamountofAc225tocancerresearchtoputoursystemtotrial.Sinceourcostanalysisputsdosagesontheorderof$10,thereisagreatmonetarybenefittoexploringourdesign.

Onebasicinsightthatwasgainedduringtheprocesswastheabilitytocoatathinfilmofsucroseonasiliconwafer.Fromwhatwesaw,spincoatingsucrosehadnotbeendonebefore.

11

Whileitwasarelativelysimpleprocess,wefoundthatitwasquiteeasytomakethinanduniformcoatingsofsucrosedowntoonemicron.Thisabilitycanbeusedinrecoilseparationapplicationsandpossiblyotherapplicationswhereabiocompatiblesacrificiallayerisneeded.

Oursystembenefitsmedicineandsciencebecauseitexpandsthebaseofknowledgeforlab‐on‐chipdevicesingeneral.Wehavefurtheredtheunderstandingofhowmicrofluidicpropertiescanpermitthedeliveryofdrugs.Ourconceptallowsforscientistsanddoctorstorethinkthetypesoftreatmentsthatarepossible.Itreducestheneedtoworryabouttimesensitivityandprovidesaneffectivemodelforhowsuchtreatmentscanbemadeontheindustrialscale.

VIIIBroaderImpact:

Shouldthisdesignprovesuccessful,itwillhavesignificantbroaderimpact.Itgoeswithoutsayingthatagreatamountofresearchtowardscancertreatmentisrequired.Thepotentialforenormousreductionsincostandincreasesinavailabilitysuggestedbythisdesigncouldprovideanewstandardfordeliveryofcomplexmedicaltreatments,especiallyforcancertherapy.Currentlythemosteffectivecancertreatmentsareveryexpensive,andthosethatarehighlyeffectiveandlowriskevenmoreso.Thisdesigncouldheraldinanewtrendofmakingthosetopendtreatmentscheaperthroughminiaturization.

IXSimulationResultsandDiscussion:

Thesolidconsistsofasquarefilmofsurfaceofareax2andthicknessf,soS=x2

Ifthesolidmassiswrittenintermsofthedensity,ρ,thenm=ρx2f

Fromthis, ,andtheratecanbemodeledas

FromAntonel

ρ=1.4g‐cm‐3

Cs=0.67g‐cm‐3

12

D=6.1x10‐6cm2‐s‐1

FromFluentSimulations,onecandeterminetheNernstLayerbasedonthemaximumvelocity.Itwasassumedthatvmax/10wouldprovidetheminimumvelocityrequiredfortransportlimitingkinetics.PriortotheNernstLayerduetothenoflowboundary

condition,thesurfacereactionisratelimitinganddissolutiontakestheformofanerrorfunction.Byestimatingthisasalinearconcentrationgradient,onecaneasilyobtainthedistanceofthisboundary.AftertheNernstLayer,theconcentrationofthesolidinthebulksolutionremainsconstantinawellstirredsolution.ThefollowingfiguredisplaysaschematicoftheNernstlayer.

Bycomparingone‐tenthofthemaxvelocity(fromvelocityprofilesobtainedwithFluentsimulations)withthediffusivityofsucroseinwater,onecanobtainacharacteristiclength,h,thatwehavedefinedastheNernstBoundaryLayerasdescribedabove.

Thisresultsinasucrosedissolutionrateof0.27µg‐s‐1assumingentiresurfacecoverageinawellstirredsolution.With14ngofsucrose(1cmx1cmx1µmfilm),thedissolutiontimeshouldtake52ms.

Crystallitesizealsohasaneffectondissolutionrate.UsingtheAvramiequation,onecansimulatetheeffectofgrainsizeonthedissolutiontime.Bydefiningdissolutionastheoppositemechanismofcrystallization,thefractionofsolidpresentataspecifictimeandtemperatureduringcrystallizationisgivenby:

Figure4:TheNernstDiffusionlayerexistsalongthechannelfromx=0tox=h.Bycomparingthediffusivitytothesimulatedvelocity

profiles,thecharacteristiclengthoftheboundarycanbeobtained.

13

whereNisthenucleationrateandvisthecrystalgrowthrate.Thisequationwasderivedbasedontheassumptionthatnucleiarenucleatedthroughoutthetransformationataconstantrate,Nandareexpandingspheres,alsoataconstantrate,v.Atsmalltime,t,thisequationtendstoward0andatlargetime,fstendstoward1asexpected.Inordertomodeldissolutionwiththisequation,wemusttake1‐fs.Theequationcanbewrittenintheform

wherekis .Tosimplifythemodel,visassumedconstantduringanisothermal

transformation.Inthiscase,kisdirectlyproportionaltothenucleationrate,N,whichisinverselyproportionaltograinsize(moresitesfornucleationresultsinsmallerandmorenumerousgrains).ByplottingthedissolutionequationfromtheAvramiequationfordifferentvaluesofk,wecanmakeaqualitativecomparisonofhowgrainsizeeffectsdissolutiontime.Thefollowingplotshowscurvesforvaryingordersofmagnitudeofk.Itisapparentthataskincreases(Nincreases)thedissolutiontimedecreases.SinceNisincreasing,grainsizemustbe

Figure5:QualitativeresultsofthedissolutionmodelasafunctionofnucleationrateandgrowthrateappliedtoanAvramiequation.Increasingvaluesofksimulateincreasingnucleationrateforaconstantgrowthrate.Increasingnucleationrateleadstosmallergrainsize.Fromthefigureitisapparentthatasgrainsizedecreases,sodoesdissolutiontime.

14

smaller.Therefore,smallergrainsizeswillresultinfasterdissolutiontimes.

FluidDynamicsSimulationsandCalculations

Thesimulationsteamhascontributedseveralessentialportionsoftheproject.Theseincludeunderstandingfluiddynamics,verifyingtheFluentandGambitsimulationsoftware,modelingtheradioactiveisotopedecay,modelingtheisotopeconcentrationprofileinthesucrosefilm,anddesigningandmodelingthefluidflowinthechip.Eachofthesecontributionswillbesummarized.

WehavedecidedtouseFluent©tomodelfluidflowandmixinginourmicrofluidicschip,andGambit©forgeometryandmeshgeneration.Thedecisiontousetheseparticularpiecesofsoftwarewasbasedentirelyonavailability.Wecontactedseveralexperts(seeprofessorsvisited),andthetakeawaypointswerethatCOMSOL©wouldbeoverallmoreuseful,andmoreflexible.COMSOL©MultiphysicsClasskitiscurrentlyavailableonthecomputersinthematerialsengineeringcomputerlab.However,aftercontactingtheCOMSOL©licensingdepartment,itwasdeterminedthatthelicensehadalreadyexpiredandthesoftwarewasbeingusedabusively.ThecostforastudenttoobtainasinglelicenseforCOMSOLMultiphysicsis$1,595.Additionally,theadd‐onfluiddynamicspackageis$1,595perlicense.Thepriceofthissoftwaregreatlyexceededourbudget,sowedecidedthatFluent©andGambit©wouldbeacceptablealternativesbecausetheyareavailable.

Thefirststepforthesimulationscommitteewastofamiliarizeourselveswiththegeneralprinciplesoffluiddynamics.Todothis,westudiedtheNavier‐Stokesequations,andderivedanexactsolutionfor2DPoiseuilleflow.2DPoiseuilleflowassumesthatwehavehardwallboundariesintheupanddownydirections,andinfinitelyspacedwallsinthezdirection.TheflowisinthepositivexdirectionbythecoordinatesysteminFigure11.ThesolutiontotheNavier–Stokesequationpresentedbelowalsoassumesano‐slipboundarycondition,whichmeansthatthereisano‐flowregiondirectlynexttothewallsofthechannel.Alsoassumedfortheanalyticsolutionisthatwehavecompletelylaminarflow.Duetothedimensionsofthechannel(1mmheightand1mlength)andtheaveragevelocityofthefluid,wecanbesurethatweareinthelaminarregime.Foranon‐circularduct,theReynoldsnumbercanbeapproximatedby(source“fluidflowresistance”blackboarddocument)Re=(ρ*v*D)/µwhereρisthedensityofthefluid,vistheaveragevelocity,DisthecharacteristicDiameter,andµisthefluidabsoluteviscosity.Forourduct,thecharacteristicdiametercanbeapproximatedbyD=2*h=2mm.Forwater,wehaveµ=.001003(kg*m^‐1*s^‐1),ρ998.2(kg*m^‐3)andsettheaveragevelocityto.001m/s.ThereforeweexpectaReynoldsnumberofapproximately2,whichputsuswellwithinthelaminarflowregime.

15

Wehavedecidedtouse2dflowtomodelourchannelsduetotimeandcomputerpowerlimitations.BasedontheReynoldsnumber,weexpectthatdiffusionalmixingwillbethemajorcontributingfactortothemixingofourreagents,andwewillthereforeneedtohavelongchannelssothatourfluidswillbeincontactforalongtime.Longchannelsnecessitatelotsofnodes,anditiseasytoseehowa3dmeshwouldquicklystrainourcomputers.The2dmodelswillthereforeprovideuswithafundamentalunderstandingofthemixingandflowparameters,butspecificswillstillneedtobemeasuredandtestedinthelab.

Theflowbetweentheupperandlowerplatescanbemodeledbythefollowingapparatusinfigure11.Thechannelsareassumedtobehorizontalwiththeflowinthex‐directionandthewidthbetweenthechannelsinthechannelinthey‐direction.

Figure6:Poiseuilleflowgeometry

(5)

Verification

Anexactsolutionfor2DPoiseulleflowwasobtainedfromtheNavier‐Stokes(NS)Equation.ThederivationfromtheNSsolutionassumedanoslipboundaryconditionwhichwaslikewisemodeledintheFluentsimulation.Thevelocityprofilesareplottedbothanalyticallyandnumerically.ForlowReynoldsnumbers,thereisstrongagreementbetweentheresults,however,astheReynoldsnumberincreases,slightdiscrepanciesoccurwhichisthoughttobecausedbyturbulenteffects.Anextensionofthisderivationtoa3DcasealongwithcorrespondingFluentsimulationswillbeusedtoverifythefinalchipgeometrywithappropriatechannelcrosssection.Thechannelparametersusedintheverificationincludeachannelwidthof0.001m,achannellengthof0.1m,fluidpropertiescorrespondingtowater(µ=0.001003kgm‐1s‐1andρ=998.2kgm‐3).Fromtheverification,the2Dmixingsimulationimplementingaswitchbackchannelgeometrycanbeconfidentlymodeledtomaintainawellmixedsolutionfor

16

8‐10minutesforappropriatechelation.ThepressuredropacrossthechannelisexpectedtobelinearaccordingtoPoiseulle’sequationwhichiswhatresultedfromtheFluentsimulations.Figures12and13displaytheplottedanalyticalcalculationsandnumericalsimulations.

Figure7:Velocityprofilesfortheindicatedchannelforinletpressuredifferencesof8.016Pa,80.16Pa,and160.32Pa.

17

Figure8:Pressureprofileacrossthelengthofthechannel.Thisprofileshowsthepressurechangeforthe8.016Painletpressuredifference.

UponhavingsimulatedandverifiedPoiseulleflowofasinglefluidinarectangularchannel,analyticallywithMatlabandnumericallywithFluent,itisnecessarytomodelthemixingoftwoseparatefluids.Multiphaseconsiderationswithinthefluids(suspensions/solutions)mayalsoberequiredforaccuratemodeling.

Undertheassumptionthatweareabletoachieveturbulentmixing,thechannelsmaybefabricatedonalargerscaleandadiffusionalmixingmodelwillnotbenecessary.Inthecasewhereturbulentmixingcannotbeachieved,channelscanbefabricatedsufficientlynarrow,andofsufficientlengthtoallowforoptimaldiffusionalmixing.Previousworkhasshownthatazig‐zagpatternedchannelcausesthegreatestdegreeofmixingmostefficiently[Jeon,W.2009].Figure14fromJeon[Jeon]showvariousgeometricalconfigurationsthatcausemixing.Alsoshownarecontourplotsdisplayingvolumefractionofphaseinfluid(goldnanoparticlesinthiscase).Fromthecontours,onecanquicklyobservethedegreeofmixingresultingfromthevariouschannelgeometries.Thiscanbefurtherdescribedquantitativelyforamoreconcretemixingcapabilityanalysis.

18

Figure9:Mixingcontours(displayingvolumefractionofgoldnanoparticles)showingeffectsofdifferinggeometries(asshowninFigure10)atspecifiedtimeintervals.Byinspectionitisclearthatthezig‐zagtypegeometryismostefficient[Jeon].

Toestablishaquantitativemeanstodescribethedegreeofmixing,Fluentiscapableofproducingcontourplotsofvolumefractionandmolarconcentrationofcomponents,which,alongwithadiscretescale,canbeusedtomeasurethedegreeofmixingbasedonacertainchannelgeometry.Onceavaluecanbeassignedtothedegreeofmixing,thiscanbecomparedtothesizeofthechanneltodetermineamixingefficiency.Channelgeometrycanbealteredinordertooptimizemixingefficiency.Wedefinemixingefficiencyashavingahomogeneousconcentrationacrossthechannel,whichisdiscussedbelowinthesection“ChipDesign.”

FigurestofollowdisplaytheresultsfromtheFluentmixingsimulation.Velocityvectorandcontourplotsclearlyshowlaminarflowwiththeno‐slipboundarycondition.Thechannelwidthwasassumedtobeonemillimeterandthefluid,liquidwater.Onesampleofwaterwasdesignatedatinlet1andanothersampleofwaterwasintroducedatinlet2.Themassfractionofwatercorrespondstothefirstsample(i.e.100%water(inlet1)atinlet1correspondsto0%water(inlet2)atthisinlet.

ChipDesign

Theresultsfromtheisotopedecaysimulationsindicatethatwewillonlyhaveausabledoseinthechipforabout45minutesonceitisremovedfromtheactiniumsource.Theactiniumsourcemustbebroughttoeveryhospitalthatplansontreatingpatientsusingthechip,andthismeansthatthephysicalchipassemblymustbedoneinhospitaltoensurethatthedoseisdeliveredinatimelymanner.Theantibodiesandchelatingagentswillbemixedandreactedinaverysmallcontainer(ontheorderof1mLinvolume)offthechiptosavetime,and

19

willbesubsequentlypumpedintothechipviaIVpumps.Themixedsolutionwillthendissolvethesucrosecontainingtheisotopes,andwillthenbemixedandreactedinswitchbackmixingchannels.Thedesignofthechannelswasinfluencedbythefactthataccordingtoasource(McDevittetal)thechelatingagentmustbeinawell‐mixedenvironmentwiththeisotopesfor8‐10minutestoachieve80%reactionefficiency.Also,theflowconstraintsofcertainIVpumpsweretakenintoconsideration;forexample,theAbbotLabsPlumXLDPumpIVInfusionhasaminimumvolumetricflowoutputof1ml/hour.Intotalweexpectthechiptobe3.5cmx3.5cm,withmostofthechipareacorrespondingtothemixingdomain,butwithonesquarecentimeterinthe“upperleft”correspondingtothesucrosefilmdissolutionchamber.Aschematicofanapproximateflowvelocityandmixingareshownbelowinfigures15,16and17.Thechannelwidthwasassumedtobeonemillimeterwide.Forflowratepurposes,thechannelsareassumedtobe1mm.Onesampleofwaterwasdesignatedatinlet1andanothersampleofwaterwasintroducedatinlet2.Bothoftheseareconnectedtogetherinthebeginningofthemixer.Themassfractionofwatercorrespondstothefirstsample(i.e.100%water(inlet1)atinlet1correspondsto0%water(inlet2)atthisinlet.Thisallowsforthemixingtobedisplayedasafunctionofmassfraction.

20

Figure10:2Dsimulationofpurewaterthroughthemixingchamberatvolumetricflowrateof10ml/hour.Thisfigureshowstheexampleflowvelocityprofile.

21

Figure11:2Dsimulationsofpurewaterthroughthemixingchamberatvolumetricflowrateof2ml/hour.Thisfigureshowstheblownupvelocityprofile.

22

Figure12:2Dsimulationsofpurewaterthroughthemixingchamberatvolumetricflowrateof10ml/hour.ThisfigureshowstheExamplemixing

ThesimulationsshowninFigure15,16and17weredonewithpurewater,whichisnotthefluidwewillactuallyuse.Theactualfluidwillclosetotheviscosityofwaterbecausethereactantconcentrationsareontheorderofnanogramspermilliliter.Futuresimulationswillincludethesolutionwiththeappropriateviscosity.Forallofthesimulations,amassflowinletwasusedbecausethatisthecapabilityofFluent.Thevolumetricflowwasconvertedtothemassflowratebyusingthedensityofwater,whichis998.2071kg/m^3.

Fixingtheinletflowrateto10ml/houryieldsamaximumcenter‐linevelocityof5.25mm/s,andwithatravellengthof534.5mm,weexpectaminimumreactiontimeofabout2min.Thisfallsshortofthe8‐10minutesrequiredformixing.However,wehavedecreasedtheflowrateto2ml/hour(whichiswithinthecapabilityofstandardIVpumps).Figure16showsthat(usingafinermesh)thatthemaximumvelocitychangesto1.13mm/s,whichputstheminimumreactiontimenearidealtoat7.88min,andtheaveragedirectlyintheidealrange.MoreexpensiveIVpumpscangetfinerflowcontrol,andcouldbeimplementedtobetter“center”theflowvelocityintotheidealrange.

23

Figure18shownbelowisaschematicofthesucrosedissolutionchamber.Thischamberwascreatedinordertogetsufficientsurfacecoverageanduptakeofsucroseandbismuth.Thelargeopenareainthemiddleisfortheinletholethatwillbepunchedintheglass.

Figure13:3Ddesignofthesucrosedissolutionchamber.Thesmallsquareinthemiddleisa2mmx2mm

Themain“square”ofthechamberis1cmx1cminarea,and1mminheight.Thechannelsareall.5mmx1mmincrosssection.Thechannelsweredesignedsothatonequarteroftheflowwouldbedirectedintoeachchannel.Duetotimeconstraints,wewerenotabletocomplete3Dfluentsimulationsforthisdesign.However,weassumedthattheflowwouldbeonequarterofthetotalflow(2ml/hour),andweranaMatlabsimulationassuminglaminarflowtodisplaythevelocitycross‐section.Figure19isthevelocitycontourofthissimulation.

24

Figure14:Velocitycontourofthechannelsinthesucrosedissolutionchamber.TheNernstlayercorrespondstothewhiteborderaroundthecoloredcontours.

ThewhiteborderaroundthecoloredprofilerepresentstheNernstlayerwhichwillbediscussedinthenextsectiononsucrosedissolution.TheNernstlayerrepresentssectionofthevelocityprofilethatfallbelow10%ofthemaximumvelocity.TheMatlabcodewaswrittenbasedonanexercisebyMartinPederson(MartinPedersen,TechnicalUniversityofDenmark).Flowanalysisindicatesthatforthechannelregion,weexpectaReynoldsnumberofabout½.Thisputsuswellwithinthelaminarflowregime,whichisexactlywhatweseeinthesimulation.Itisabitdistortedfromatraditionallaminarprofilebecausethecross‐sectionisarectangle,notasquareorcircle.Thissimulationindicatesthatthemaximumvelocityisabout.4mm/secinthesechannels.AlthoughtheFluentsimulationsdidnotdirectlyrepresentmaterialsdesignaspects,theywereintegraltotheoveralldesignofthechip.Theflowprofilesallowedforyieldcalculations,andmixingestimationscouldbeusedtodeterminereactionefficiency.GambitservedbothasmeshgenerationsoftwareforFluent,andforCADapplicationsshowingtheschematicofthesucrosemixingchamber.

Facilities,Materials&Prototyping

Thedesignandfabricationofourtargetedalphatherapylab‐on‐chipdevicewillutilizenumeroussoftwareandlaboratoryresourcestoensuredesiredoperation.IndesigningourdevicethesimulationscommitteehasperformedworkusingtheANSYSsoftwarepackageFluent.Thegoalofthesesimulationshasbeentwofold,firsttoconfirmtheapplicabilityofthe

25

softwareforourmicrofluidicdesignandsecondtomodeloursysteminsuchawayastoensurethatproperdesignparametersaremet.Theparametersofinterestincludeflowrate,mixingandgeneraldimensionalrequirements.Tooptimizetheseparametersfactorssuchassystemtopology,channeldiameterandlength,etcareconsideredinourcomputersimulations.Fromthesesimulationsthefinaldevicedesignwillbedeterminedandfabricationwillfollow.

Ourdevicewillbefabricatedusinglithographictechniquesanditthereforebecomesnecessarythatalithographicmaskbedesignedandprocuredsothatproductioncanbegin.WehavedesignedandsentoutthemasktoFineLinePrototypingforfabricationandarecurrentlyawaitingitsarrivalfordeviceproductiontobegin.EspeciallyimportantinthisearlyworkwillperfectingthepatterningofzigzaggedchannelsintotheSisubstrateasthemixingwhichoccursinourdevicewillrelyalmostentirelyonthesepatterns.Wemustbesurethatweareabletocreatethedesired1mmchannelsizeonSiandthattheywillbefreeofobstructionsforproperflow.OurmaskwasdesignedusingtheCorelDRAWX4softwarepackage.

Ourmicrofluidicsystemwillbefabricatedonasiliconchip.Therearefourmajorfabricationaspectswhichrequireconsideration:chipsurfacefeatures,thesugar/Bifilm,SizeExclusionChromatography(SEC)chamberandglasstop.AmaskofthefeaturesisshowninFigure20ThevarioussurfacefeaturesrequiredfortheproperfunctioningofthesystemwillbemoldedusingthepolymerPDMS.ThispolymerwillbepatternedbyusingaSiwaferwiththechipdesignetchedintothewafer’snativesilicondioxidelayer.ThisprocesswasexplainedtousbygraduatestudentMarianaMeyerandfromherguidancewehavelearnedthatallofthepatterningrequiredcanbeperformedrelativelysimply.Inordertocompletepatterningwefirstsurroundthepatternedwaferwithanaluminumfoil‘boat’andthenpourthePDMSoverthewafersurface.OncecuredthepatternedPDMScanbepeeledfromtheSimoldandinspectedforimperfections.Thisprocessrepresentsthelion’sshareoffabricationtimeandeffortduetoitsinherentlycomplicateddesignandneedforprecisionwhencomparedwiththeotherconstructionaspects.Thechip(s)usedforfabricationwillbesuppliedthroughtheFABLABwhichmaintainsalargestockofblankSiwafers.

26

Figure15:Projectedmaskdesign.Contact(topleft)andchannels(right,bottom)

Thesecondmajorconsiderationliesinthesugar/Bifilmwhichwillcontaintheactivebismuthforalphatherapy.OncethedissolutionbehaviorofthisfilmisdetermineditshouldberelativelysimpletospincoatasugarfilmwhichcanbeimplantedwithBiatomsforintroductiontoourchip.Wehavedeterminedthatwecaneasilypatternsuchafilmwiththedesiredthickness(~1micronrange)byspincoating.Thusfarwehavecompletedtrialsusing33and50wt%sucroseinwatersolutionsandbothhaveviscositypropertieswhichhaveresultedinfilmsinthemicronrangeaccordingtomeasurementsmadeusingthen&kanalyzerdeviceintheFABLAB.Thesucrosefilmswerespincoatedat4000rpmfor40sec.,basedonaknownprocessforasimilarly‐viscous1812photoresist.Inthesetrialswewerealsoabletodeterminethatbyusingplasticbackedvacuumtapewecouldsimplycoverthepartsofthechipthatwedonotwishtohavesucroseon,spinonthefilmandremovethetape.UsingthistechniqueweareabletoquicklyandeasilymakeanysizedsucrosefilmontheSiwafer.YoucanseeafinishedsucrosefilminFigure21.

27

Figure16:Sucrosefilmpatternedonpolishedsideofsiliconwafer.Squarearea(~1x1cm) tapedoffusingvacuumtape.

Wealsofoundthatafterafewdaysthefilmsbeganshowingregionsofcrystallitenucleation.ToconfirmordenythisobservationwecharacterizedthesucrosefilmswithXRD.Wetookmeasurementsoftheregionsofthefilmthatwerestillas‐spunandoftheregionsthatappearedtobecrystallizing.Theas‐spunregionshowednosignificantpeaks(excludingthelargepeaksfromtheSisubstrate),suggestingthattheas‐spunsucrosefilmisamorphous.ThecrystallizedregionsshowednoconclusiveresultsfromaregularXRD,sowetookawideranglerangescanofthatregioninapowderdiffractometer/XRD.Thediffractometerwouldbeabletodetectcrystallinepeaksforapoly‐crystallinesample,wherethenormalXRDwouldaverageoverallofthecrystallitesandshowinconclusiveresultsaswesaw.ThepowderdiffractometryresultsareshowninFigure22.ThecrystallizedregionhadmanypeaksincommonwiththesucrosereferenceintheXRDdatabanks,suggestingthatthecrystallizedregionwasinfactpoly‐crystalline.Thestrongpresenceofpeaksrelatedtothe(100)familyofsucrose,whichalignsat(400)withSi(002)suggeststhattheremaybesomeepitaxialeffectsonthenucleationfromtheSisubstrate.Furthertestingalongthisideawouldhavetobedonetoconclusivelyconcludeonewayoranother.

28

Figure17:XRDResults.Thebottomscanistheas‐spunregion,andthetopscanisthecrystallizedregion.

TheSECchamberisanotherarearequiringmajorconsiderationinthefabricationofourdevice.FirstofallthedimensionsofthischambermustbedialedinsothatthedesiredseparationcanoccurbytheSECgelused.Thischamberwillbeplacedoffofthechipinordertoensurethattheproperdimensionscanbeachieved.Otherwiseitwouldbedifficultifnotimpossibletohavebothdimensionsandvolumefortheseparationcolumn.Secondlythepropergelmustbeselectedforbothexclusionsizeandcoarsenessagainforresolutionandalsoforflowratethroughthedevice.WehavechosentouseBio‐GelA‐0.5mSECgelfromBio‐Radwhichisaagarosebasedmaterialcommonlyusedfortheseparationofantibodiesandotherproteins.Fordesiredresolutionweneed5‐10:1length‐to‐diameterand4‐10:1bed‐to‐samplevolumeratiocolumn.Basedontheinformationfromoursimulationswehavedeterminedthatweneedacolumnwhichisapproximately5x1x1mmwhichiscorrectforbothnecessaryratios.

29

Crystallitesizewasestimatedusingopticalmicroscopy.Figure23showstwopicturestakenwiththeopticalmicroscopeintheteachinglabintheFabLab.Thesmallestnucleatingcrystalliteswereneedle‐likestructuresapprox.5uminwidth.Largercrystallitestructureswereupto800um,withgroupsofradially‐orientedneedle‐likestructuresformingcirclesmultiplemillimetersindiameter.Thesmallestcrystallitesizeistheimportantsizewithrespecttodissolution,sinceitwouldbeonthatscalethatthesucrosewouldbedissolvingintothewatersolution.

Figure18:OpticalMicroscopepictures.5xmag(left)showssomenucleatingneedle‐likestructures.20xmag(right)showsfully‐grown,larger,moregrain‐likestructures.

ThefinalmajorconsiderationinfabricatingourdeviceisinfusingthetopPDMStothebottomSiwafer.Thisprocessshouldberelativelysimple,butitisnecessarytobeverycarefulandprecisetoensurethatthefinaldevicedoesnotleakinternallyorexternally.TobondthePDMSandglasswehavechosentouseaUVcuringepoxyUV30‐27seriesfromLoxeal.Thischoicewasmadefor3mainreasons:fastcuringtime(~5minorless),highbondstrength(20‐30N/mm^2)andmedicalusecertification(ISO10993).Speedisimportantbecauseoncethesucrosefilmisactivatedwithbismuththeclockbeginscountingdownuntiltherequireddosageofradiationisnolongerpresent.Highbondstrengthisofobviousimportanceasthedevicewillbesubjecttopressuresasfluidsflowthroughandmixwithinthesystem.Thisstrengthguaranteesthatthedevicewillneitherleaknor,evenworse,failtotally.Finallythemedicalusecertificationensuresthatwearenotintroducingpotentiallyharmfulcomponentstothesystembyoverlookingthecompositionofthesealantusedforthedevice.Theprototypemaybesealedwithpartially‐curedPDMSinsteadoftheepoxyforeaseandavailability(nothavingtopurchaseandwaitforspecialtyepoxy).

30

FortestingwehavediscussedusingstreptavadinandbiotinasmodelantibodyandchelatewithDrPhaneuf.Thesespecificmoleculeswillbeusefulfortestingbecauseoftheirfluorescencepropertiesoncechelationhasoccurred.Asuccessfultrialwillresultinhighfluorescenceintheoutflowstreamfromourchipindicatingthatthebiotin/streptavadinchelationhasoccurred.

Totestthesucrosedissolutionrateagainstourmodels,wewouldhavetorunanexperimenttomeasurethemassdissolvedperunittime.Oneideawastoputthesucrosefilmincontactwithmixedwaterforafixedamountoftime.Aftereachtimeinterval,thefilmwouldbetakenout,dried,andmeasuredforthickness.Knowingthesurfacearea(1x1cmsquare)anddensity,wewouldplotdm(dt)andcomparewiththeresultsofthetheory.Oneproblemwiththismethodisintroducingwatertothefilminacontrolledmanner.Amicrofluidicchannelsystemwouldaccomplishthis,butrequiresextraworkaheadoftimecreatingone.Asecondproblemisquicklyandevenlydryingthefilmwithoutaffectingthefilmitself.Afterbeingsubmergedinwater,thefilmwillbepartiallyhydrated.Dryingthefilmfastenoughtoeffectivelyceasedissolutionwouldbedifficult,especiallysosincethefilmcannotbetakenpast90Cwithoutdegradingthesucrose.

Overthecourseofthesemestertheprototypegoalshavechangeddrastically.Atthemidtermpointwedecidedtofocusourenergiesonthemodelingandprototypingofthesucrosefilmwhileputtingtheproductionofanactualmicrofluidicchannelsystemasasecondaryobjective.Wemadesureateverystepofthewaythattheprototypingaspectofourprojectdidnotsacrificetimeormanpowerfromtheprimaryfocusofthecapstoneproject:thedesign.Theaspectsoftheprototypethatwefinallydecideduponwerethosethatwebelievedwouldhelpverifyandsupportthedesignportionsoftheproject.

XConclusions:

Ourdesignmetwithgreatsuccesswhenourresultsarecomparedwithourdesigngoals.Wemetthemall,andareabletoprovideadoseofradio‐isotopeatminimalcostandpotentially,wideavailability.Oursimulationsresultedinrapidmixing(100%mixingwithin2mmofchannellength),thesucrosedissolutionconfirmedarapiddissolutionrate,andtheBismuthdiffusionconfirmedasignificantspreadacrossthechannelgiventheassumptionsmadeforourmodel.

31

XIAcknowledgementsandReferences:

Acknowledgements:Thankyouto:Dr.RaymondPhaneufDr.RobertBriberDr.ManfredWuttigDr.IchiroTakeuchiDr.KeithHeroldDr.DonDevoeDrSameerShahProfessorEmeritusGeorgeHelzProfessorAndreiVedernikovProfessorGareginPapoianJohnHummelJimO’ConnorTomLoughranJohnAbrahamsMarianaMeyerRichardSuchoskiReferences:[1]Allen,BarryJ.,ChandRaja,SyedRizvi,YongLi,WendyTsui,DavidZhang,EmmaSong,Chang

FaQu,JohnKearsley,PeterGraham,andJohnThompson."TargetedAlphaTherapyforCancer."PhysicsinMedicineandBiology49.16(2004):3703‐712.Print.

[2]Antonel,P.et.al.TheKineticsofDissolutionRevisited.JournalofChemicalEducation.Vol.80No.9.2003.

[3]Anderssona,Helene,WouterVanDerWijngaart,PeterNilssonb,PeterEnokssona,andGoÈRanStemmea."AValve‐lessDiffuserMicropumpforMicroˉuidicAnalyticalSystems."SensorsandActuatorsB.72(2001):259‐65.Print.

[4]Chen,Jia‐Kun,andRuey‐JenYang."ElectroosmoticFlowMixinginZigzagMicroshannels."Electrophoresis28(2007):975‐83.Print.

[5]HolgerBecker&ClaudiaGärtner,"Polymermicrofabricationtechnologiesformicrofluidicsystems",AnalBioanalChem(2008)390:89–111

[6]I.J.Brownetal,"ElectroplatingandelectrolessplatingofNithrough/ontoaporouspolymerinaflowcell",JournalofAppliedElectrochemistry,31:1203‐1212,2001

[7]Imam,S."AdvancementsinCancerTherapywithAlpha‐emitters:aReview."InternationalJournalofRadiationOncologyBiologyPhysics51.1(2001):271‐78.Print.

[8]Jeon,Wonjin.Shin,C.B.Designandsimulationofpassivemixinginmicrofluidicsystemswithgeometricvariations.ChemicalEngineeringJournal.vol152.pp575‐582.2009.

[9]Karakaya,I.,Thompson,W.T.,“TheAg‐Bi(Silver‐Bismuth)System”JournalofPhaseEquilibria(Vol.14,No.4,525‐530),1993.

32

[10]FluidFlowResistanceELMSReference[11]Kossman,S.E.,D.A.Scheinberg,J.G.Jurcic,andEtAl."APhaseITrialofHumanized

MonoclonalAntibodyHuM195(anti‐CD33)withLow‐DoseInterleukin2inAcuteMyelogenousLeukemia."ClinicalCancerResearch(1999).Web.

[12]Lee,FookT.,AngelaJ.Mountain,MarcusP.Kelly,andEtAl."EnhancedEfficiencyofRadioummunotherapywith90Y‐CHX‐A‐DTPA‐hu3S193byInhibitionofEpidermalGrowthFactorRecepter(EGFR)SignalingwithEGFRTyrosineKinaseInhibitorAG1478."ClinicalCancerResearch(2005).Web.

[13]Mori,Sadao,andHowardG.Barth.SizeExclusionChromatography.Berlin:Springer,1999.GoogleBooks.

[14]Murphy,R.M.,Slayter,H.,Schurtenberger,P.,Chamberlin,R.A.,Colton,C.K.,Yarmush,M.L.“Sizeandstructureofantigen‐antibodycomplexes.Electronmicroscopyandlightscatteringstudies”JournalofBiophys(45‐56),1988.

[15]Okamoto,H.“Bi‐Ni(BismuthNickel)”JournalofPhaseEquilibriaandDiffusion(Vol.29,pg.203).2008.

[16]Pan,Yu‐Jen,andRuey‐JenYang."AGlassMicrofluidicChipAdhesiveBondingMethodatRoomTemperature."JournalofMicromechanicsandMicroengineering16(2006):2666‐672.Print.

[17]Regino,C.A.S.,Wong,K.J.,Milenic,D.E.,Holmes,E.H.,Garmestani,K.,Choyke,P.L.,Brechbiel,M.W.”PreclinicalEvaluationofaMonoclonalAntibody(3C6)SpecificforProstate‐SpecificMembraneAntigen”CurrentRadiopharmaceuticals,(2009,2,9‐17).

[18]Ruddy,F.H.,A.R.Dulloo,J.G.Seidel,andB.Petrovic."SeparationoftheAlpha‐emittingRadioisotopesActinium‐225andBismuth‐213fromThorium‐229UsingAlphaRecoilMethods."NuclearInstrumentsandMethodsinPhysicsResearchB213(2004):351‐56.Web.

[19]Geankoplis,ChristieJohn.TransportProcessesandSepartationProcessPrinciples.UpperSaddleRiver:PrenticeHall,2003.194‐95.Print.

[20]McDevitt,etal,PreparationofA‐Emitting213Bi‐LabeledAntibodyConstructsforClinicalUse,THEJOURNALOFNUCLEARMEDICINE,Vol.40,No.10,October1999

[21]"AnOverviewofPolydimethylsiloxane(PDMS)FluidsintheEnvironment."Environmental Informaton;DowCorning(2009).[22]LaneA.Bray,JoelM.Tingey,JaquettaR.DesChane,OlegB.Egorov,andThomasS.

TenfordeDevelopmentofaUniqueBismuth(Bi‐213)AutomatedGeneratorforUseinCancerTherapyInd.Eng.Chem.Res.Vol39,3189‐3194,2000

[23]DianeE.Milenic,MarioRoselli,SaedMirzadeh,C.GregPippin,OttoA.Gansow,DavidColcher,MartinW.Brechbiel,andJeffreySchlom,InVivoEvaluationofBismuth‐LabeledMonoclonalAntibodyComparingDTPA‐Derived.BifunctionalChelatesJournalofCANCERBIOTHERAPY&RADIOPHARMACEUTICALS.Volume16,Number2,2001

[24]PeterW.LinderetalTheDiffusionCoefficientofSucroseinWater.JournalofChemicaleducationVolume53,Number5,May1976

[25]A.B.MacKenzieandR.D.ScottSeparationofBismuth‐210andPolonium‐210from

33

AqueousSolutionsbySpontaneousAdsorptiononCopperFoils.Analyst,December,1979,Vol.104,pp.1151‐1158[26]AvramiMelvinGranulation,PhaseChange,andMicrostructure,KineticsofPhaseChange.IIIJOURNALOFCHEMICALPHYSICSVOLUME9FEBRUARY1941