Modeling Of CMP

University of California at BerkeleyModeling of CMPDavid DornfeldCMP researchers: Jihong Choi, Sunghoon Lee, Dr.Hyoungjae Kim, Dr. Dan EchizenyaDepartment of Mechanical EngineeringUniversity of CaliforniaBerkeley CA 94720‐1740http://lma.berkeley.edu Laboratory for Manufacturing Automation, 2005

University of California at BerkeleyOverview Background on modeling Review of work to date Some new developments pattern/feature sensitivity pad design Laboratory for Manufacturing Automation, 20052LMA

University of California at BerkeleyNew Book on Modeling Chemical Mechanical Planarization (CMP)“Integrated Modeling of Chemical Mechanical Planarization for Sub-Micron IC Fabrication:From Particle Scale to Feature, Die and Wafer Scales,” J. Luo and D. A. DornfeldWritten by researchers at UC-Berkeley, this monographreviews CMP modeling literature (from Preston to presentday efforts) and develops, with a strong emphasis onmechanical elements of CMP, an integrated model ofCMP addressing wafer,die and particle scale mechanismsand features. Special emphasis is on abrasive sizes,distributions and resulting material removal rates anduniformity resulting over all scales.175 Figures and 14 tablesISBN 3-540-22369-x Springer-Verlag 2004For 9-X.Or contact: dornfeld@berkeley.edu Laboratory for Manufacturing Automation, 20053LMA

University of California at BerkeleyChemical Mechanical PlanarizationCMP Team in FLCCDornfeld, et alDoyle, et alTalbot, et alChemicalPhenomena Laboratory for Manufacturing Automation, 2005MechanicalPhenomenaInterfacial andColloidPhenomenaLMA

University of California at BerkeleyScale Issues in CMPMechanical particle forcesParticle enhanced Tool mechanics,Load, SpeedChemicalReactionscritical featuresnmmmµm5MechanismwaferdiesFrom E. Hwang, 2004 Laboratory for Manufacturing Automation, 2005PadGroovesLayoutScale/sizeLMA

University of California at BerkeleyCMP Process SchematicF : down forceOscillationconditionerw w : wafer rotationslurry feedRetainerringBackingfilmheadWafer CarrierWafertablepadw p :pad rotationElectro plated diamondconditioner Laboratory for Manufacturing Automation, 2005PoreWallPadAbrasive particleTypical pad6LMA

University of California at BerkeleyAn overview of CMP research in FLCCCu CMP[oxidizer],[complexing agent],[corrosion inhibitor],pH Bulk Cu CMPBarrier polishingW CMPOxide CMPPoly-Si CMPBulk Cu slurryBarrier slurryW slurryOxide slurryPoly-Si slurryDoyleChemical reactionsTalbotDornfeld modelMechanical material removal mechanism in abrasive scalePhysical models of material removal mechanism in abrasive scalePatternTopographyMIT modelPad properties in die scaleSlurry supply/ flow patternin die scaleWafer scale pressure NUWafer scale velocity profileModels ofWIDNUModels ofWIWNUSmall dishing & erosionReducing scratch defectsReducing ‘Fang’Reducing slurry usageUniform pad performancethru it’s lifetimeLonger pad life timeWafer bending with zone pressuresSlurry supply/ flow patternin wafer scalePad grooveBetter planarization efficiencyBetter control of WIWNUSmaller WIDNUmodeldesign goalUltra low-k integrationAbrasive type,size and concentrationPad asperity density/shapePad mechanical propertiesin abrasive scalePad designFabricationtechniqueFabricationTestE-CMPPad development Laboratory for Manufacturing Automation, 2005LMA

University of California at BerkeleyThe 4‐component system Hypotheses:– all polishing processes can be described as a 4 componentsystem;– Understanding the components and their interactions (pair-wise,triplets, etc) provides a structure to catalog our knowledge (andignorance)“Granule”?Lap (rigid)Deliberately sought a word thatcovers the range of particlesused without implying anythingabout size, hardness, or removalWorkpiece Carrier fluid GranuleLap mechanism: µm to nm sizerange; from hard (diamond) toPlatensoft (rouge);Pad}Source: 86. Evans, J., Paul, E., Dornfeld, D., Lucca, D., Byrne, G., Tricard, M., Klocke, F., Dambon, O., and Mullany, B.,“Material Removal Mechanisms in Lapping and Polishing,” STC “G” Keynote, CIRP Annals, 52, 2, 2003. Laboratory for Manufacturing Automation, 20058LMA

University of California at BerkeleySix possible pair-wise interactions le-padpad-fluidFluid-granule Laboratory for Manufacturing Automation, 20059LMA

University of California at BerkeleyThree-way interactions (triplets) e-granule-padFluid-pad-granule Laboratory for Manufacturing Automation, 200510LMA

University of California at BerkeleyStribeck Curve and Characteristics of slurryfilm thicknessDirect contactSemi-direct contactHydroplane sliding Laboratory for Manufacturing Automation, 2005Friction coefficientPolishing padFilm tHydroplaneslidingElastohydrodynamic ionHersey number( Viscosity Velocity)PressureStribeck curve11LMA

University of California at BerkeleyGap effects on “mechanics”Eroded surface bychemical reaction--- softeningSilicon waferDelaminated by brushing‘Small’ gapAbrasive particlePolishing padPad-based removalSilicon wafer‘Big’ gapAbrasive particlesSlurry-based removal Laboratory for Manufacturing Automation, 2005Polishing pad12LMA

University of California at BerkeleyIdealized CMP‘Softened’ surface bychemical reactionSilicon waferAbrasive particlePolishing padPad asperityMechanical Aspects of the Material Removal Mechanism inChemical Mechanical Polishing (CMP) Laboratory for Manufacturing Automation, 200513LMA

University of California at BerkeleyInteractions between Input VariablesFour Interactions: Wafer‐Pad Interaction; Pad‐Abrasive Interaction;Wafer‐Slurry Chemical Interaction; Wafer‐Abrasive InteractionVelocity VVolChemicallyInfluenced WaferSurfaceAbrasive particlesin Fluid (Allinactive)WaferAbrasiveparticles onPolishing padPad asperityContactarea withnumber NActive abrasiveson Contact areaSource: J. Luo and D. Dornfeld, IEEE Trans: Semiconductor Manufacturing, 2001 Laboratory for Manufacturing Automation, 200514LMA

University of California at BerkeleyFramework Connecting Input Parameters with MaterialRemoval RateBasic Equation of Material Removal: MRR N VolNSlurry AbrasiveWeightConcentration CAverage AbrasiveSize XavgφXavgVolg Fraction ofActive AbrasivesX avg‐aFraction of Active Abrasive:1‐φ((g‐Xavg)/ σ) where g isthe minimum size of activeabrasivesProportion ofActive AbrasivesPad Topography& Pad Material Laboratory for Manufacturing Automation, 2005Abrasive SizeDistribution φ15Force F & VelocityActive AbrasiveSize Xavg‐aWafer HardnessHw / SlurryChemicals &Wafer MaterialsDownPressure P0LMA

University of California at BerkeleyExperimental Verification of Pressure Dependenceof Material Removal Rate (MRR)MRR N Vol K1 {1‐φ(1‐K2P01/3)}P01/2.Ke1 (K1 84148, K2 0.137)Advantage overPreston’s Eq. MRR KePV MRR0:What input variablesand how they influenceKe is predicableKe2(K1 8989, K2 0.3698)SiO2 CMP Experimental Datafrom Zhao and Shi,Proceedings of VMIC, 1999 Laboratory for Manufacturing Automation, 200516LMA

University of California at BerkeleyAbrasive Size Distribution Dependence of MRR:Particle Size Distribution [1](%) FrequencyFive Different Kinds of Abrasive (Alumina) Size Distributions for Tungsten CMPAbrasive Size X (Log Scale)1.Bielmann et. al., Electrochem. Letter, 1999 Laboratory for Manufacturing Automation, 200517LMA

Normalized Material Removal RateUniversity of California at BerkeleyRelationship between Standard Deviationand MRR Based on Model Prediction21.81.61.41.210.80.60.40.20Xavg 0.29umXavg 0.38umXavg 0.60umXavg 0.88umXavg 2umSize influencedStd dev influenced00.050.10.150.20.250.3Standard Deviation (10-6m) Laboratory for Manufacturing Automation, 200518LMA

University of California at BerkeleyPattern-Density Dependency ModelpadpadMRRoxideoxideUp AreaK/densityTimeKDown Area0InterLevel Dielectric Case (single material)Source: MITSame Pattern DensityDifferent Orientations Laboratory for Manufacturing Automation, 200519LMA

University of California at BerkeleyFramework of a CMP Topography Evolution Model Laboratory for Manufacturing Automation, 200520LMA

University of California at BerkeleyDishing and Erosion in Copper Damascene ProcessTrenchSiNSiO2Via(a)(b)(c)(d)Fabrication steps in dual damascene process (a)deposition of SiN, SiO2 and etching trenches and vias inSiO2 (b) deposition of barrier layer (c) copper fill (d) CMPand deposition of SiN (courtesy of Serdar Aksu) Laboratory for Manufacturing Automation, 2005LMA

University of California at BerkeleyDefinition of Feature‐Scale TopographySHcuHOxideErosion eCopperThinningWoxWcuHox Hox0H CopperDishingd SHox(a)(b)(a) Feature scale topography before dielectric material isexposed and (b) feature scale topography after dielectricmaterial is exposed Laboratory for Manufacturing Automation, 200522LMA

University of California at BerkeleyModels of Polishing PadEEEηη1ηE2(a)(b)Linear Elastic andLinearViscoElastic ModelsE1η2(d)(c)Separated Models ofPad Bulk and AsperitiesKdKf Laboratory for Manufacturing Automation, 200523LMA

University of California at BerkeleyThree Stages of Wafer‐Pad ContactS S0H Hcu0 Hox0DfHcu01S1 Df1H Hstage12Hox0Only upper part ofstep is in contact Laboratory for Manufacturing Automation, 2005Erosione3Two differentBoth upper andbottom parts of step materials areremovedis in contactsimultaneously24LMADishingd

University of California at BerkeleySimulation Results of Step HeightEvolution for Different Pattern Density500Step Height S p Height S (nm)400PDi 0.1PDi 0.2PDi 0.3PDi 0.4PDi 0.5PDi 0.6PDi 0.7PDi 0.8PDi 0.9450Step height S (nm)450Step height S (nm)500PDi 0.1PDi 0.2PDi 0.3PDi 0.4PDi 0.5PDi 0.6PDi 0.7PDi 0.8PDi 0.95020406080100 120 140Polishing Time t (second)1601800200Planarization time (sec)20406080100120140Polishing Time t (second)160180Planarization time (sec)Linear Elastic PadWcu 100 microns Laboratory for Manufacturing Automation, 2005025Linear ViscoelasticPadLMA200

University of California at BerkeleyCopper Dishing as a Function of Pattern Densityusing commercial pads Laboratory for Manufacturing Automation, 200526LMA

University of California at BerkeleyCopper Dishing as a Function of Selectivity Laboratory for Manufacturing Automation, 200527LMA

University of California at BerkeleyEffect of Pattern Density - PlanarizationLength (PL)High-density regionGlobal stepLow-density regionILDMetal linesPlanarization Length Laboratory for Manufacturing Automation, 200528LMA

University of California at BerkeleyModeling of pattern density effects in CMPEffective pattern densityPlanarization length(window size) effecton “Up area”a 320um Test pattern a 640uma 1280um Post CMP filmthickness prediction atdie-scale Laboratory for Manufacturing Automation, 2005 Effective density map 29LMA

University of California at BerkeleyDie scale modeling of topography evolution during CMPContact wear modelInitial pressure distributionMRR modelTopography evolutionIterationwith timestepContact wear modelNew pressure distribution Laboratory for Manufacturing Automation, 2005LMA

University of California at BerkeleyFeature level interaction between pad asperities and pattern topographyPADZ(x,y)Z padReference height (z 0)Z ( x, y) Z padF(x, y) Kp (asperity density) (PDF(z dz) PDF(z)) (Z(x, y) z)0F tent F ( x , y ) dxdydiedzF tent F die ?YesZ(x,y) Z padzNoNoF tent F die ?Yes--Z padZ padZ pad Laboratory for Manufacturing Automation, 200531LMA

University of California at BerkeleyChip level interaction between pad and pattern topographyk2k1k1k 2Kpad k2k1 k 2rPLMIT model : approximation of contact wear modelw4 (1 γ 2 ) qrw(r ) πE40umPattern40umPDE i, j40um x 40um cell Laboratory for Manufacturing Automation, 200532 π20r21 sin 2 θ d θ2PL w ( i , j ) PD ( i , j ) w (i, j ) i, j LMA

University of California at BerkeleySimulation result20% 33% 50% 100%50%t 0 sect 10 sect 20 sect 30 sect 40 sect 50 sec33%20%t 60 sec Laboratory for Manufacturing Automation, 2005t 70 sect 80 secLMA