OmniSim - UK Silicon Photonicsomnisim -英国硅光子学

Thenextgenerationinphotoniclayout&simulation“propagatingideas”WhatisOmniSim?TrulydirectionagnosticlayoutandsimulationtoolTimedomainengine2D&3DFDTDgeneratespectralresponseinasinglesimulationFrequencydomainenginenewalgorithmbringsunparalleledspeedAutomaticoptimisationevolvesyourdesignwhileyoueatlunch!MaskexportusingGDSII“propagatingideas”OmniSimFeaturesSummaryOmniSimcontainsasophisticatedlayouteditorComplex3DdevicescanbebuiltupusingphysicallayersandprocessconceptsElementscanberotatedandlinkedtoeachotherAwiderangeofsourcesandsensorsareavailableCanimportaFIMMWAVEmodeforlaunchingormeasuringmodepower–substantiallyincreasessimulationefficiency.ThefrequencydomainmethodoffersarapidtechniqueforsimulatingdevicesTheFDTDtechniqueisabletosimulateeitheratCWoroverabroadbandKallistoscanbeusedtooptimiseadeviceThefinaldesigncanbeexportedasGDSIIdataOmniSimFDTDEngineFeaturesGeneralThefastestFDTDengineinthefieldSub-gridding–substantiallyreducememoryandtimerequiredc.f.variablegridFDTDs.ManymemoryreducingtechnologiesClustersystemfordistributingsimulationacrossmultipleCPUs/PCsMaterialsAnisotropy–generalsymmetrictensor.Dispersivematerialsincludingmetals–Drude/Debye/Lorentzianmodelswithautomaticfittingtool(youjustprovideadispersioncurve).Chi2andChi3non-linearityPermeability(magneticmaterials)BoundaryConditionsHighperformancePMLson6facesDispersivePMLs–e.g.continueametallayerintothePMLMetal,magneticandperiodicboundaryconditionsSub-griddingPhotonDesignfirsttodemonstrateanefficient,stablesub-gridalgorithminFDTD–allpreviouspublishedalgorithmsnotstable.Increaselocalresolutionby2x,4x,8x,…Canacceleratea3Dsimulationby64xormore.Muchmoreefficientthan“variablegrid”FDTDs–adiagonalofsmallfeatureswouldcreateafinegrideverywhere.ModeExcitorsandSensorsImportprecisemodefromFIMMWAVEformodelaunchDittoimportformodesensor=>Substantiallydecreasesimulationvolume/time.modeexcitormodesensormodecomputedinFIMMWAVEExample-ringresonatordesign/simulationEditorallowsveryeasylayoutofeventhemostcomplexringstructuresSimulationwithtimeorfrequencydomain“propagatingideas”FreerotationofalmostallobjectsPowerfulconstraintsystemHierarchicalframeworkn-levelundo/redoSupportsetchandgrowthprocessesExporttomaskfileTheMaskEditorFirstdefinethephysicallayersDefinethedeviceprocesssteps,suchasetchandgrowthLayout-SOIgratingexampleUsethephysicallayersandprocessstepstobuildupthedeviceUsearraysofsubelementsforperiodicstructuresAddviewpointstoconfirmstructureisasexpectedLayout-SOIgratingexampleLayout-SOIgratingexampleUseviewpointstoconfirmstructureisasexpectedLayout-CouplerExampleConstraintslinkobjectstogetherCannowadjustwaveguideseparationjustbymovingasinglelowerwaveguideFinaldesigncanbeexportedasGDSIIAnewpowerfulstateoftheart2DMaxwellsolverforpropagationofEMfieldswithinanarbitraryphotonicstructure.CombinestheproblemsolvingcapabilityofFDTDwiththespeedandaccuracyofthefrequencydomainExceptionallyfastOrderofmagnitudefasterthancompetingtools.BasedonnewefficientnumericaltechniquesHighdelta-ncapabilityIntegratedwithOmniSim-TDengineWiderangeofsourcesandsensorsOmniSim-FD'sspeedandlownumericalnoisemakeitidealforautomaticoptimisationOmniSim-FDTakeaninitialdesignUseKallistosandthefrequencydomainformultiwavelengthoptimisationFinetunethedeviceusing3DFDTDtoproduceanoptimiseddesignOmniSim-DesignSystemFDTDisaccurateandrobust.ThesourcesoferrorinFDTDcalculationsarewellunderstoodFDTDcansimulatelighttravellingatanyangleUseFDTDwhereBPMandEMEfailFDTDtreatsimpulseresponsereadily.FDTDdirectlycalculatestheimpulseresponseofanelectromagneticsystemAsinglesimulationcanprovidethesteadystateresponseatanyfrequencywithintheexcitationspectrumFDTDaccountsforallreflectionsTransmissionandreflectioncanbecalculatedinasinglesimulationComputermemorycapacitiesareincreasingrapidly.FDTDdiscretizesspaceoveravolume,inherentlyneedslargeamountsofRAMWhyuseFDTDFDTDisdirectdiscretisationofMaxwell’sEquationIn1DMaxwellgivesus:Discreteformgives,foraspaceandtimeintervaldx,dt:WhatisFDTDFDTDgivesspectralresponseoverawideband,byFourierAnalysis.Thetimestepdtisapproximatelydx/cRunasimulationfortimetTotwithtimestepdt,thenSpectralresolution:df=1/tTotSpectralrange:fTot=1/dtStructureswithsmallfeaturesfinegridsmalltimestepForhighspectralresolutionthesimulationrequiresmanytimesteps.FDTDuseslotsofmemoryThealgorithmsusedinOmniSimareverymemoryefficientMirrorplaneshavebeenintroducedtoreducememoryrequirementsNewmethodtoreducememoryrequirementsbyuptoafactorof8or16SomeFDTDguidelinesThespectralwidthofthesourceisgivenbyFourieranalysisACWsourcetransformstoadeltafunctioni.e.ithaszerospectralwidthTolookatabroadspectralresponseweneedtolaunchashortpulseAfinitepulsetransformstoafinitespectralwidthIfT=pulselengththenthespectralwidthFisgivenby1/TInwavelengthterms=-2/cTForexamplea20femtosecondpulseat1.55umhasaspectralwidthof700nmSpectralwidthofsourceAhighlyefficientFDTD(finitedifferencetimedomain)engine2Dand3DFDTDengineveryfastspeedoptimisedalgorithmMorememoryefficientthancompetingproductsPMLandmetalboundariesManyopticalsourcesavailableincludingwaveguidemodeanddipoleRuntimemonitoringofevolvingfieldsWavelengthresponsespectraComplexrefractiveindexMaterialdatabaseusesDebye,DrudeorLorentzianmodeltoautomaticallyfittoawiderangeofmetalsRight:FDTDsimulationofopticalpulsesscatteringoffcylindricalobjectsOmniSim-FDTDCornerreflectorexampleDefineacornermirrorusingdifferentmasklayersCornerreflectorexampleSilicawaveguideshavetoolowarefractiveindexfora90degreecornerreflectorButthereflectorworksquitewellfor60degrees!CouplerdesignexampleA20fssinusoidalpulsecentredat1.55umwavelengthContainsspectralinformationover700nmWavelengthresponseofdeviceCouplerdesignexampleUseCWat1.55umwavelengthtostudythefielddistributionContainsspectralinformationover800nmRingresonatordesignexampleUsecouplerasbuildingblockforringresonatorSimulationtimecomparisonsContainsspectralinformationover800nmThefrequencydomainmethodisveryfastanduniquetoOmnisimForsomedevicesitmightbebettertodomultiplefrequencydomainsimulationsratherthanasinglebroadbandFDTDRingresonatordesignexampleUsecouplerasbuildingblockforringresonatorKallistos-automaticoptimisationKallistoscanevolvedesignsthataresimplynotpossiblebytraditionaltechniques.Choiceofsophisticatedoptimisationalgorithms-localandglobalObjectivefunction(determines“good”and“bad”)canbeanyexpressionCanoptimiseany3-5parametersgloballyCanoptimiseany30+parameterslocallyMulti-wavelengthoptimisationtomaximisedevicebandwidthKallistos-RingresonatorexampleThewavelengthpeaksaredeterminedbythelengtharoundtheringinsertvariablestraightlengthLbetweenthetwoarcsTheextinctionratioisdeterminedbythecouplerVarythegapbetweenthewaveguidesSetupoptimisationexampletotuneresonatortransmissionto1.7umandmaximisetheextinctionratioKallistos-RingresonatorexampleDefinetwoindependentvariablesLengthwillrepresentthelengthofwaveguidebetweenthearcsGapwillrepresenttheedgetoedgeseparationbetweenthearmsofthecouplersKallistos-RingresonatorexampleLinktheindependentvariablestothestructureCommandscanbefoundveryeasilybypressingthetabkeyKallistos-RingresonatorexampleDefineanobjectivefunctionInthisexamplethewavelengthissetto1.7anFDcalculationisperformedthefluxinthethroughandthedroparmiscalculatedthewavelengthissetto1.72umthecalculationisrepeatedTheobjectiveisdefinedastheextinctionratioatthetwowavelengthswhichisthentobemaximisedTheresonatorwillbetunedto1.7umandthefreespectralrangesetto40nmKallistos-RingresonatorexampleRuntheoptimisationUsingglobaloptimisationSeveralpeaksarefoundcorrespondingtodifferentspectralordersKallistos-RingresonatorafteroptimisationTransmissionat1.7umTransmissionat1.72umActiveFDTDEngineModelgaininFDTDsimulationIncludesarate-equationforcarrierdynamicsMulti-Lorentziangainmodel(firsteverinWorld)User-defined“contacts”forcurrent-injectionApplications:PhotoniccrystallasersNano-cavitylasersRight:aLittrowmodeofaphotoniccrystallasersimulatedwithCrystalWave.ConclusionsOmniSimprovidesoneoftheeasiesttouseinterfacesintheindustry,withpowerfulfeaturessuchastiltedetchingandgrading.AdvancedfastFDTDengineboastsmanyWorld-firsttechnologiesincludingsub-griddingandrealisticgainmaterials