Computer Aided Approaches forNanophotnic ResearchesTae-Woo LeeCenter for Computation and TechnologyLouisiana State UniveristyOverview• Numerical approaches in nanophotonics research– Introduction to “Computer Aided R & D”• requirement, future direction• Examples: nanophotonics researches enabled by HPCsimulations– Quasi-3D plasmonic crystals for sensing application– Coherent control of local plasmons for decoding/encoding ofdigitally mastered optical signals• Concluding remarks– CCT, a firm foundation for computer aided R & D environmentIntro. to Computer Aided R & D (CARD)• Computational background– Low cost PCs are doing better job nowadays and will bemuch better in the future– Development of message passing interface (MPI) allows usto acquire unlimited computing power from high performancecomputing (HPC) environments where many cheap PCs areassembled as a cluster.– Techniques for rigorous and realistic numerical simulationsare developed and adopted for HPC environments.• Demands in nanophotonics– Light interactions are complex and unpredictable– Rigorous and realistic description of nanophotoncs system isnecessary to capture all the complexities inherent in thenanophotonics systems.Research environment with CARDprovide optimal designfeedbackHPCsimulationfabrication ofprototypeproductionnovel deviceconceptfeasibilitytestunderlying physicsResearch environment with CARDprovide optimal designfeedbackHPCsimulationfabrication ofprototypeproductionnovel deviceconceptfeasibilitytestunderlying physicsScientific research(discovery, exploration, …)Engineering development(design, optimization, …)Establishing CARD• Computational requirements– Simulation tools and techniques must be rigorous so that onecan safely rely on simulation results– Simulation must be capable of efficiently handling realistic fullsize 3-D problems• General research direction– Leading role: explore novel scientific phenomena andsystems which might results subsequent research projects– Supporting role: provide tools to analyze theoreticallypredicted and/or experimentally observed physicalphenomena.HPC based simulationsStrong collaborations with theory and experimentExample 1: quasi-3D plasmonic crystals• Collaborative work: computer modeling + experimentalworks– Experimental work: John A. Rogers’ group at UIUC– Motivated by T. W. Ebbesen et al., “Extraordinary optical transmittance throughsub-wavelength hole arrays,” Nature 391, 667-669 (1998)• Problem of interestincident lighttransmittanceCross-sectional view of one cell(sealed structure)goldpolymerExample 1: quasi-3D plasmonic crystals• FDTD Modeling setup and HPC computingCross-sectional view:wave excitation and measurement3-D computational domain with periodic boundary in x and y and absorbing boundary in zplane waveexcitationprobesurfaceExample 1: quasi-3D plasmonic crystals3-D computational domain segmented for parallel computing• FDTD Modeling setup and HPC computing– Numerical specs• Computational domain size: grid resolution: 5 nm lateral: 142 ~ 160 grid cells / dimension height: 1200 grid cells• Computational requirements memory: 7 ~ 10 GB # of CPUs: 72 ~ 128 Wall clock time: 2 ~ 3 hrs / 350 fs• Computing resource: Jacquard @ nersc (4.4 GFlops/sec)Example 1: quasi-3D plasmonic crystals• Result: transmittance spectrum– Period = 710 nm, Hole diameter = 450 nm, Relief depth = 350nm, metal thickness = 64 nmFDTD simulationsexperimental measurement(John A. Rogers at UIUC)wavelength (nm)transmittance (%)wavelength (nm)transmittance (%)slight pile up of goldgrain on disk edgeExample 1: quasi-3D plasmonic crystals3-D Intensity distributions at specific wavelengths of interestexperimentFDTD idealFDTD defect– Strong coupling between bottom diskand top hole resonance isresponsible for highly sensitivetransmittance peak of B– Feasible for bio-sensing applicationFor more info. – Proceedings of National Academyof Science, 103, 17143 (2006)Example 2: coherent control of local plasmons forencoding/decoding optical signals• Metal nanoparticle system– Specsd1 (20 nm) < d2 (140 nm)Length = 250 nm– Chirped optical pulsesHEincident light- chirped opticalpulsessilverABd1d2cone shape silvernanoparticleE t f t t tinc( ) ( )sin[ ( / )]= +! " #1( )f t: pulse envelope!: pulse width!: chirp parameter-1-0.500.510510 15 20 25 30t (fs)0510 15 20 25 30t (fs)! "> 0! "< 0Einc(t)Field intensities at A and B aremonitored in time simultaneouslythrough a FDTD simulationLocal plasmons in the system• Shorter resonant wavelength for A (smaller cross section)compared to B (bigger cross section)λ0 = 350 nmλ0 = 365 nm-50 0 50 100 150 200 250 300x (nm)-100-50050100y (nm)AB-50 0 50 100 150 200 250 300x (nm)BABAcw lightwith λ0Spatiotemporal observation of local hot spot0510152025051015202505101520253005007009001100|E(t)|2 envelopetime (fs)ABα = -0.120510152025051015202505101520253005007009001100time (fs)ABα = +0.12300 1100700 900500 300 1100700 900500BAchirpedpulseLocal intensity peaks are delayed in timeApplication: encode/decode digitallymastered optical signalsoriginal signal1 10opticalfiber1 1?Photodetector- convert to electricsignalafter spreading bydispersion200 400 600 800 1000 120000.511.5time (fs)signal (a.u.)1 ? 1after converted to electric signalApplication: encode/decode digitallymastered optical signalsoriginal signal1 10opticalfiber1 1?Photodetector- convert to electricsignalafter spreading bydispersionNote, e-h pair generation is not directly included but assumed to be proportional to lightintensity. Capacitance effect and response time of photodetector are treated as a localaverage of signals200 400 600 800 1000 1200-100102030time (fs)1 0 1signalA - signalBproposed NPN photodetectorwith nanosystemnipin2-DarrayABApplication: encode/decode digitallymastered optical signalsoriginal signal1 10opticalfiber1 1?Photodetector- convert to electricsignalafter spreading bydispersion200 400 600 800 1000 1200-100102030time (fs)1 0 1signalA - signalB200 400 600 800 1000 120000.511.5time (fs)signal (a.u.)1 ? 1Conventional photodetectorPhotodetector with nanosystemFor more info. – Physical Review B, 71, 035423 (2005)Summary for examples of CARD• Two examples of HPC enabled nanophotonics researchoutcomes were presented– Quasi 3-D plasmonic crystal:• Sensitivity of the system was identified by matchingsimulation and experimental measurement• Further in-depth analysis revealed