27420064CHINESEJOURNALOFSEMICONDUCTORSVol.27No.4Apr.,2006*ProjectsupportedbytheNationalOutstandingYoungScientistsFoundationofChina(No.50325519)Correspondingauthor.Email:zhouzaifa@yahoo.com.cnReceived17January2006,revisedmanuscriptreceived15February2006c2006ChineseInstituteofElectronicsPhotolithographyProcessSimulationforIntegratedCircuitsandMicroelectromechanicalSystemFabrication*ZhouZaifa,HuangQingan,andLiWeihua(KeyLaboratoryofMEMSoftheMinistryofEducation,SoutheastUniversity,Nanjing210096,China)Abstract:Simulationsofphotoresistetching,aerialimage,exposure,andpostbakeprocessesareintegratedtoobtainaphotolithographyprocesssimulationformicroelectromechanicalsystem(MEMS)andintegratedcircuit(IC)fabricationbasedonthreedimensional(3D)cellularautomata(CA).Thesimulationresultsagreewellwithavailableexperimentalresults.Thisindicatesthatthe3DdynamicCAmodelforthephotoresistetchingsimulationandthe3DCAmodelforthepostbakesimulationcouldbeusefulforthemonolithicsimulationofvariouslithographyprocesses.ThisisdeterminedtobeusefulforthedevicesizedfabricationprocesssimulationofICandMEMS.Keywords:cellularautomata;processsimulation;photolithographysimulation;model;TCADEEACC:2570;2560;2550ECLCnumber:TN4Documentcode:AArticleID:02534177(2006)040705071IntroductionAsthedimensionsofintegratedcircuits(IC)arescalingdowntothenanometerregimeandthecomplexityofmicroelectromechanicalsystem(MEMS)designandfabricationisincreasing,threedimensional(3D)photolithographysimulationhasbecomenecessaryforaccurateanalysisofcomplexstructuressuchascontacts,corners,islands,and3Ddefects[1].Simulationaidsinunderstandingtheeffectsofparametersduringphotolithographysothatmaskscanbeoptimizedandtheaccuracyofcriticalstructuresandthereliabilityofdevicescanthereforebeimproved[2].Thesimulationofphotolithographyisacomplextaskthatusuallyincludesprecisemodelinginaerialimagesimulation,exposuresimulation,postbakesimulation,andphotoresistetchingsimulation(developmentsimulation)[3].Theetchingsimulationisusuallythemosttimeconsumingstepandgreatlyaffectstheaccuracyofthewholephotolithographysimulation.Soafastandaccurate3Dmodelforphotoresistetchingisgreatlydesiredforeffectivephotolithographyprocesssimulation.Duetoitsadvantagessuchaseaseinhandlingtopologicalchanges,adaptivemeshmethods,andthesimplestmodelsforimplementationinMEMSandICfabricationprocesssimulation,thecellularautomatamodel,whichwasfirstpresentedbyvonNeumannfollowingUlamssuggestions[4],hasbeensuccessfullyappliedtothesimulationofvariousfabricationprocessessuchasphotoresistetching[5~7],deposition[8,9],andsiliconanisotropicetching[10].The3DdynamicalCAmodelforphotoresistetchingsimulation[11],whichisextendedfromthe2DdynamicalCAmodel[5],hasbeenpresentedandtestedusingwellknownetchingratedistributionfunctionsandisdemonstratedtobeaccurate,fastandstable.Inthispaper,simulationsofthephotoresistetchingprocessusingthe3DdynamicCAmodelandthepostbakeprocessusingthe3DCAmodelareintegratedtogetherforthefirsttimewithanaerialimagesimulationandanexposureprocesssimulationtoaccuratelydescribetheeffectsofparametersduringthephotolithographyprocess.Thesimulationresultsagreewellwithavailableexperimentalresults.Thisindicatesthatthe3DdynamicCAmodelforphotoresistetchingsimulationand3DCAmodelforpostbakesimulationareaccurate,fast,andcanbeintegratedwithotherphotolithographysimulationsteps.Thiswillbeusefulforthesimulationofthedevicesizedfabrication27processofICandMEMS.23DdynamicalCAmodelThephotoresistisdividedintol!m!nidenticalcubiccellswithsidelengtha,andthe3DMooreneighborhoodisadoptedinthe3DdynamicCAmodel,asshowninFig.1.Thereare6adjacentcubiccells,12diagonalcubiccells,and8pointcubiccellsintheneighborhoodofcell(i,j,k).AboundarycellbetweentheinternalphotoresistandetchantwillbeetchedbyetchantflowingfromitsFig.1Mooreneighborhoodofthecellularautomataneighbors.Theoretically,theeffectofetchantfromallneighborsshouldbetakenintoaccountwhenthestateofcell(i,j,k)isupdated.Butthe8pointdiagonalcellsdonotsignificantlyaffectthechangeofthestateofcell(i,j,k).Sotheeffectofthepointdiagonalcellsisneglectedinthismodel.ThelocalstateCi,j,k(t)ofeachcellattimetisdefinedastheratiooftheetchedvolumeVe(t)tothetotalvolumeVc:Ci,j,k(t)=Ve(t)Vc(1)TheupdateruleisthengivenbyRef.[11]:Ci,j,k(t1+T)=Ci,j,k(t1)+Ve(t1+T)adjacent+Ve(t1+T)diagonala3(2)Ve(t1+T)adjacent=(a-dih-dil)!(djldkl+djldkh+djhdkl+djhdkh)+(a-dyh-dyl)(dildkl+dildkh+dihdkl+dihdkh)+(a-dkh-dkl)(dildjl+dildjh+dihdjl+dihdjh)+(a-dih-dil)(a-dkl-dkh)(djl+djh)+(a-djh-djl)(a-dkl-dkh)(dil+dih)+(a-dkh-dkl)(a-djl-djh)(dil+dih)+dildjldkl+dildjldkh+dildjhdkl+dildjhdkh+dihdjldkl+dihdjldkh+dihdjhdkl+dihdjhdkh(3)Ve(t1+T)diagonal=D2R2i,j,k(T+Tci+1,j,k+1(t1))2+D2R2i,j,k(T+Tci+1,j,k-1(t1))2+D2R2i,j,k(T+Tci-1,j,k+1(t1))2+D2R2i,j,k(T+Tci-1,j,k-1(t1))2+D2R2i,j,k(T+Tci,j+1,k+1(t1))2+D2R2i,j,k(T+Tci,j+1,k-1(t1))2+D2R2i,j,k(T+Tci,j-1,k+1(t1))2+D2R2i,j,k(T+Tci,j-1,k-1(t1))2+D2R2i,j,k(T+Tci+1,j+1,k(t1))2+D2R2i,j,k(T+Tci+1,j-1,k(t1))2+D2R2i,j,k(T+Tci-1,j+1,k(t1))2+D2R2i,j,k(T+Tci-1,j-1,k(t1))2(4)whereVe(t1+T)adjacentandVe(t1+T)diagonaldescribetheeffectsfromthe