Claude Weisbuch
Transcription
Claude Weisbuch
Claude Weisbuch Materials Department, University of California, Santa Barbara, USA Laboratoire PMC, Ecole Polytechnique, Palaiseau, France Strong coupling 1 : Fermi Golden rule not valid: tri-‐dimensional systems-‐ excitonic polariton, resonant Brillouin sca<ering, resonant polariton fluorescence, RIRS, RRS R G Ulbrich, N Van Hieu, A Nakamura end of 70ies Low D1 quantum wells: free exciton luminescence (2D/3D), interface disorder, R Dingle , A C Gossard, W Wiegmann, PM Petroff, WT Tsang end of 70ies QW laser J Nagle, S Hersee, JP Duchemin, M Razeghi ‘ 80ies Quantum dots, bo<leneck, … end of ‘ 80ies H Benisty, E Bockenhoff Strong coupling 2 cavity polaritons Y Arakawa, M Nishioka, R Houdre, R Stanley, U Oesterle, M Ilegems ‘ 90ies controlling flow of light : weak coupling -‐ Fermi’s Golden rule-‐modify photonic density of states Microcavity LEDs, photonic crystals for telecoms or LEDs H Benisty, D Labilloy, M RaMer, E Schwoob, A David, S Denbaars, S Nakamura, J Speck, E Hu, E MaPoli, E Rangel ‘ 00ies plan 1 Strong coupling I : the environment in late 80’ies 2 Strong coupling II : how was it discovered? 3 microcavity light emi<ers a) strong coupling ? b) weak coupling design rules c) microcavity LEDs the limits no photon recycling 4 visible light emi<ers today Summary 1 Strong coupling I : the environment • QWs/QWWs/QDs – e-‐h – Excitons in 1991: 3D vs 2D • Purcell • Microcavi\es : CQED with atoms Strongly-coupled 3D excitons and photons, excitonic polaritons are the quasi particles of the system Exciton ω.k photon ω.k photons excitons Phys. Rev. B 7, 4568 (1973) In the polariton picture, light scattering becomes a first order process, due to Resonant Brillouin scattering the scattering of an excitonic polariton. By bulk excitonic polaritons Outside photons transform into polaritonswith unit efficiency at crystal interface Efficiency of Brillouin scattering ≈ 10-12 in opaque materials Brillouin scattering is the Need of multipass FP spectrometer to reject scattering of an excitonic polariton photoluminecence by an acoustic phonon. Efficiency of resonant Brillouin scattering ≈ 10-3 Surprise: in quantum wells, free excitons dominate luminescence! Rashba theory of bound excitons GaAS QW Bulk GaAS Explanation: Agranovitch& Dubovskii, 1966 3D exciton-polariton luminescence is a forbidden process (needs break of symetry): 3D excitons transform into 3D photons and vice-versa, back and forth. 3D luminescence is due to the outcoupling of a polariton at the air interface. QW luminescence is the coupling of 2D electronic system with 3D photons = no polariton effect as reversibility is possible, k is not conserved. Exciton luminescence is allowed, and fast! More on low D excitons and 3D photons: Temperature dependance of lifetime of QW and Q wire excitons: Fraction of population within light cone (depends on DOS) ~ T for QWs, ~ T 1/2 for QWRs Gershoni & Katz, Phys.Rev. ‘ 94 Trying to understand luminescence: Linewidth increases with increasing quantum confinement Measuring all samples grown in Bell Labs 1974-1980 No data in this region! Linewidth is due to single atomic layer fluctua\ons! Macroscopic manisfestation Of microscopic disorder Solid state communs 1981 Resonant Raleigh sca<ering of an inhomogeneously broadened transi\on Hegarty et al. Phys.Rev. Lett. 49, 930, 1982 absorption PL Resonant Raleigh scattering Mobility edge Fabry-Perot with internal source Microcavity emission: atoms in a Fabry-Pérot resonator Idea from ~1960 (Kastler, Schawlow&Townes), Applied Optics, 1, p.17 (January 1962) cf single mirror ! rings output mirror mc= Int L λ2 For a point source : ( Fabry-Perot interf.) × (2-beam interf) (metal) 2 2 E = Eo × T1 1− r1r2 e 2iφ 2 × 1 + r2 e 2iφ ' 2 2 = Eo × Exaltation or Inhibition due to the modal structure of the whole cavity T1 1− r1r2e 2iφ 2 × 2 ζ (z, θ ) Factor from 0 to 4 depending on the source location with respect to the mode antinodes at the considered angles Controlled Atomic Spontaneous Emission from Er 3+ in a Transparent Si/Si02 Microcavity Vredenberg, et al. Phys. Rev . Lett.71, 517 (1993) Physics and Device Applications of Optical Microcavities Yokoyama Science 256, 66 (1992) Spontaneous Emission Experiments with Optical Microcavities The Thresholdless Laser Altering Spontaneous Emission Bulk System dimensionality Optical phenomenon 3D exciton 3D photon conservation de k3D et E(K) Strong coupling Coupled exciton –photon state= Exciton polariton Extrinsic radiative process 2D Quantum 2D exciton Weak coupling wells in 3D 3D photon Intrinsic radiative process: free optical non-‐conservation de exciton luminescence dominates environment k⊥ Quantum wells in microcavity 2D exciton 2D photon conservation de k// Strong coupling Cavity polaritons Extrinsic radiative process Summary 1 Strong coupling I : the environment 2 Strong coupling II : how was it discovered ? & early days Excitons in 2D (QWs) interacting with 2D photons (planar cavity) Energy and wavevector conserved: 1exciton interacts with 1 photon Cavity polaritons Erice, 1993, Eli 2 & Eli 1) Dipole or field 1/ ΩRabi time cw, arakawa et al, PRL ‘ 92 U. Oesterle EPFL Room-temperature cavity polaritons in a semiconductor microcavity Houdre et a., Phys.Rev. B 49, 16 761 (1994) Inhomogeneous Broadening FP mode" Exciton" CavityPolariton" + If Rabi splitting is larger than inhomogeneous broadening All exciton states are coupled together through cavity photon exchange Splitting is hardly perturbed γc /E0 = 3.10−5 , γ/E0 = 1.10−4 , σ/E0 = 1.10−3 coupling strength are 50 (1), 5 (2) and 1 (3) Houdré et al. Phys. Rev. A 53 (1996) 2711 While QW width is >1 meV, polariton linewidth can be as low as 50 µeV QW fluctuations are averaged over polariton wavefunction, i.e. photon Mode size, ≈ tens of µm, instead of exciton size, ≈ ten nm. Phonon broadening Phonon scattering in LPB Is diminished due to small DOS photon-like Measurement of Cavity-Polariton dispersion Curve from AngleResolved Photoluminescence Experiments R. Houdre, Phys. Rev. Lett. 73, 2043 (1994) Cavity-polariton photoluminescence in semiconductor microcavities: Experimental evidence R. Stanley et al., Phys. Rev. B 53, 10995 (1996) photolum absorption Summary 1 Strong coupling I : the environment 2 Strong coupling II : how was it discovered? 3 microcavity light emiCers a) strong coupling ? b) weak coupling design rules c) microcavity LEDs the limits no photon recycling Experimental: Norris, CW, Arakawa 1994 Mode counting & cavity order modes ~ Dirac peaks , antinode factor taken constant for simplicity ko extracted to air kz ko guided modes θc kz mc Airy peaks guided modes k// k// (b) : micro-cavity of order mc ≤ 2n2 ~ Same amount of emission in each mode 1 # mc mc Airy peaks θc (a) : “macro”-cavity of high order mc => Extracted fraction = SINGLE extracted mode #in the ideal limit (R →1) ζextracted mode ζi ∑ all modes 2 → (< 1 !) mc Schubert et al., Science 265, 943 (1993) Impact of Planar Microcavity Effects on Light Extrac\on Part I: Basic Concepts and Analy\cal Trends Benisty et al. IEEE Journal Of Quantum Electronics, 34, 1612 (1998) GaN microcavity emi<er modeling Extracted intensity as function of cavity length and QW position in cavity. Quantum well must be 100nm from top and 150 from bottom with 20nm accuracy(λ/8n) GaAs microcavity LED Ø 400µm 28% QE top contact oxydation trench intracavity contact external QE 0.3 0.2 0.1 0 AlOx DBR in air single-facet substrate-based 0.1 current density (A.cm 1 10 -2 ) M. Rattier, JSTQE 2002 HYBRID ORGANIC INORGANIC QWs IN CAVITY • principle : resonant energy transfer by resonant microcavity photons between electrically-created excitons and radiative excitons Organic layer III-V Frenkel exciton light out photon Wannier-Mott exciton QW • solution : use coupled cavities electrical injection Agranovitch, Benisty, Weisbuch Solid state commun;102 , 631 (1997) organic QW (pentacene, J-aggregates, terrylene,...) GaAs / GaAlAs QW GaAs substrate Summary 1 Strong coupling I : the environment 2 Strong coupling II : how was it discovered? 3 microcavity light emi<ers a) strong coupling ? b) weak coupling design rules c) microcavity LEDs the limits no photon recycling 4 visible light emiCers today Light Emitting Diodes LEDs Another important metrics: luminous efficacy of light source for lighting: lumens per watt Luminous efficacy of radiation Spectral efficacy of source Source LER/max LER=411 lm/W History of LED Efficiency Dupuis and Krames, IEEE J. Lightwave Tech. 26, 1154 (2009). Present geometrical op\cs solu\ons Pa<erned substrate Roughened surface shaped substrate Thin film roughened surface Light extrac\on study for different devices a) Plain GaN chip b) Surface roughening c) Roughened GaN chip d) Pa<erned Sapphire substrate e) Flipchip III-nitride photonic-crystal light-emitting diodes " with high extraction efficiency " Jonathan J. Wierer, Jr1, Aurelien David1 * and Mischa M. Megens2 , Philips Lumileds" NATURE PHOTONICS 2009" The optical mode control results in a high-performance" light-emitting diode with an estimated " unencapsulated light extraction of 73 %, higher than any unencapsulated III-nitride LED measured to date. " Lumileds Determining the extracMon efficiency: photonic crystal LEDs Need to determine EQE (external quantum efficiency) and IQE (internal quantum efficiency) ηextraction = EQE / IQE 8 10 Encapsulated Unencapsulated 6 ) % (e EQE measurements (integraMng sphere): IQE measurement: Room temperature EL method*: ηIQE = 7.9% η q e ) W m 4( 5 P 2 0 0 200 400 J (A/cm2) 600 0 800 *Getty et al. Appl. Phys. Lett. 94, 181102, (2009) Extraction efficiency : Very high light extraction! Due to a more efficient extraction than absorption mechanisms* 73% in air however in a poor IQE device… 94% in epoxy dome Matioli et al., Appl. Phys. Lett., 96, 031108 (2010) Appl. Phys. Express, 3, 032103 (2010) Solid-‐State Ligh\ng Research and Development: DOE Mul\ YearProgram Plan March 2011 LED Package and Luminaire Loss Channels & Efficiencies Best: 70% in blue. 10mA /3V=20mW i.e. equiv. in IR 9mW, in 2π st =1.5 mW/st 15 times the 1993 Science MicroCavity LED! Light efficiency in LED: why do we want more? If power injection is limited by heat extracted from chip, improved efficiency allows to inject more current on given chip surface, hence diminishing cost Going to 80% efficiency from 60% increases the output by 166%! Cost is decreased by 62% 50 LEDs Magazine June 2011 Japanese conversion to LED lights would slash energy usage Replacing all the incandescent lights with LED lamps is more feasible, since this would have an initial cost of 850 billion yen ($10.5 billion), but would save 27.3 billion kWh annually, worth the output of four nuclear reactors, according to the institute of energy economics (Japan). Summary 1 Strong coupling: the basic proper\es were evaluated in the early 90’ies and shown to be sound for quantum behaviour -‐ Long coherence \mes & distances -‐ Strong interac\ons between photon like par\cles -‐ Many profound discovered, more later in the conference 2 Applica\ons: a big challenge! Supplementary slides Lighting life cycle analysis: fabrication energy requirements Lighting: energy requirements over life @10 lm/W @50 lm/W @50 lm/W Cost of energy saving: 2600kWh@10c= US $ 260! Why LED lamp so expensive today? 2” substrate ≈ 20 cm2 @30 A/cm2 (droop conservative) = 0.6 kA = 1.8 kW electrical input @ 3V 0.72 optical kW @ 40% Wallplug Efficiency 140 klm @ 200 lm/W color efficiency = 140 x 1klm light bulbs! (@ 80lm/W) [≈ 4mmx4mm each] Today: processed wafer cost $ 1000 Wafer cost per bulb: $7!(≈ ASP 4GB flash memory) Diminishing costs requires diminished processing costs and current density increase A glimpse into the future: Production throughput Suppose large factory: 100 machines each1 1000 runs/year2 1000 x 100 x 42x 20= 8 107 cm2/year Aixtron produc\on MOCVD machine 42 substrates 2” 1 substrate ≈ 20cm2 Pump power, full wafer surface made into LEDs ≈ 8 107 cm2x 1253 A/ cm2 x 3V≈ 3.1010 W ≈ 6 1010 W saved power (replaces lamps with x 3 efficiency, 150lm/W) 3hrs lamp use4/365 days => 6 1010 kWh saved 1 GW nuclear power plant: 1.2 106kW x 8000h produces ≈ 1010 kWh LED producMon each year saves energy of 6 new (or old) electricity plants! Value of saved energy: 6 1010 kWh = 6 109US $ @ 10c LEDs produced: 8 107 cm2 = 8 109 1mmx1mm LEDs5 Comments 1. 1200 MOCVDs were ordered in 2010,not all that large and with this throughput 2. 3 runs per day- should be possible with high throughput machines 3. Conservative number to avoid too much droop. Might become better 4. 3 hrs/day on average for every LED produced-maybea little too much 5. i.e. 8 billions per year-big factory- today’s LED production 50-60 billions , but it should largely increase for wide scale lighting use.