1 Some repetition Øegeneration (3R) is needed to scale a network ©caling of all optical networks: No digital optics (so far) ·Optical network elements: Today in essence analog ones (1R or 2R) 2 Network elements: ­Optical cross connects (OXCs) ­Optical add drop multiplexers (OADMs) ­Optical line amplifiers (OLAs) ­Optical line terminals (OLTs) ­........... 3 OXC 4 Need for wavelength conversion 5 6 OADM 7 8 Fig 7.6 9 10 Issues for OADMs ©hould be capable of dropping a certain number of wavelengths · Should allow an operator to drop/add channels under computer control, without affecting other channels · Should have fixed loss, regardless of OADM configuration ( i e how many channels are dropped/passed through) 11 PoP Long distance PoP Metro DWDM NetworkMetro DWDM Network Metro core Business Area Banks Residential Areas Metro Access Network Mgmt Enterprise 2nd generation networks 12 Optical cross connects · Needed for complex mesh topologies and large number of wavelengths · Functions: ­ Service provisioning ­ Protection switching ­ Bit rate transparency (if it is all optical) ­ Performance monitoring ­ Wavelength conversion ­ Multiplexing and grooming 13 14Grooming, regeneration, wavelength conversion 15 In an electronic switch one can monitor signals, measure BER, groom, etc In an optical one, optical power can conveniently be measured, but not much more 16 17 18 ? ? ? ? 19 Wavelength crossconnect General strictly non-blocking wavelength switch: Paths in the wavelength and space domains Any input (space,lambda) port can be connected to any output (space, lambda) port by a control signal (normally electronic signal) => Wavelength conversion { }i { }i { }i { }i { }i 20 Devices in optical networks 21 Photonics in information transfer and in general ·Functionality ­Photonics lacks, currently, RAM type memory and signal processing capability ·Physical size ("footprint") ­ 100s to 1000s of wavelengths in length, order of wavelength in transverse dimension ­ Compare electronics (FET gate lengths < 100 nm), but interconnects important for photonics as well as electronics Èost ­Too expensive (too much handcraft..) ·But there are ways to resolve this 22 Devices in optical networks · Bulk photonics ­ Dominating today, but has in most cases all the drawbacks listed above · Integrated photonics ­ The most interesting candidate for solving these problems, is increasing being deployed in the network (e g arrayed waveguide gratings) · But do not forget MOEMS... Fig. 1.2 Artist's sketch of a monolithic integrated GaAs transmitter. (After Blum et al. (1.54) Laser Fig. 1.2 Artist's sketch of a monolithic integrated GaAs transmitter. (After Blum et al. (1.54) Integrated optics is a single mode technology! Integrated photonics 23 24 Photonic Integrated Circuits (PICs) Integrated photonics has been conditioned on the development of single mode fiber networks, since the most efficient use of PICs requires a single mode system. Why? Most PICs are based on some sort of modal interference, this effect is washed out in multimode structures... Why guided wave optics? · Flexible interconnect with low crosstalk and with no diffraction spreading due to propogation · The small transverse interaction areas and the selectable interaction lengths decrease the energy required to control the devices by orders or magnitude, and ensures compatibility with electron confinement volumes. But there are also drawbacks: · Guided wave optics is(at least today) a planar technology, i e the inherent 3 D parallelism of optics is not utilized. But this is changing! · The polarization sensitivity is in general aggravated in guided wave optics, since modal confinement factors and propagation constants are mode dependent (gratings) 25 Example Drive or switch energy reduction in guided wave modulators in relation to bulk devices 26 2 mod 2 += L w wLVolume ulatorBulk Energy stored in interaction volume, L=length w=beam waist ( )2 mod LVolume ulatoreguideChannelwav = Switch energy ratio=Equal lengths => 2 2 / 2 + L w w 27 20 40 60 80 100 Bulk beam waist , microns 1 2 3 4 5 6 7 Log ratio of switch energy bulkmodulator PLC - modulator , length Hmicrons Las parameter short long 28 Classical PIC problems · Insertion loss (due to propagation, elements in the cct, coupling to the surrounding world /fiber/) ­Typical values: 0.5-3 dB PIC/fiber, 0.1-1dB/cm, ?/cct element · Polarization 29 Polarization The ordinary standard single mode fiber does not preserve a defined state of polarization Solutions: · Strictly polarization independent devices: Devices independent of the state and degree of polarization, i.e. rapid polarization fluctuations are allowed. · Polarization control, i.e. feedback stabilization, which requires an assessment of the state of polarization (performed in coherent systems at the time) · Polarization diversity, i.e. separating the polarizations and processing them separately (probably only solution for photonic crystal devices) · Polarization holding (PH) fibers: Expensive, nonstandard fibers, but new photonic crystal fibers emerging?? · Polarization scrambling, i.e. the polarization is randomized before entering the device in question 30 Photonic integrated circuits (PICs) · "Active": Manipulation of the real and imaginary parts of the refractive index · Passive 31 The Kramers Krönig relations - - = dPV )(1 )( - - -= dPV )(1 )( 32 Physical Mechanism Operation mode Characteristics Materials Index change through Pockel effect (linear EO effect). Voltage applied over dielectric or reverse-blased pn-junction Speed usually limited by Walk-off or RC- constants. Small nonresonant effect. Dn/n ~0.0001 typically, no associated absorption. All materials lacking inversion symmetry: LN, III-V semiconductors Index change through nonresonant Kerr effect (quadratic EO effect) See above See above. Usually very small All materials Free carrier, intra-band transitions ("plasma" effect). Index and absorption change Carrier injection or depletion through injection and reverse blasing, respectively Usually smaller than other carrier-induced effects. Nonresonant effect, increases with the square of the wavelength All semiconductors and other materials in which free carriers can be induced Absorption/gain change through bandfilling Unipolar through reverse blasing of pin-junction (depletion), bipolar Unipolar operation is fast (RC-limited) but cannot give gain, bipolar operation limited by carrier recombination. Strong dependence. III-V (and other SC) Index change through bandfilling See above See above. Large index changes achievable, of the order An/n ~0.001-0.01 Strong dependence. See above Absorption change by Quantum Confined Stark Effect (QCSE) Reverse blased pin-junctions RC ­ or transport ­ time limited response (fast, depends on number of wells and barrier heights etc). Strong dependence. Quantum Well SC Index change through QCSE See above RC-limited response. An/n 0.001-0.01 Strong dependence. See above Absorption change by Franz- Keldysh (FK) effect See above RC ­ or transport time limited response. Strong dependence. All SC Temperature induced absorption and index changes Change of external temperature, and heat dissipation associated with carrier recombination Large effect that partially counteracts the index and absorption changes induced by electronic effect. Depends on heat sinking and the ratio between radiative and non- radiative recombination channels inSCs All materials 33 However, MOEMs! ·Microoptoelectromechanical systems ·Huge "index changes" (material-air, i e 0.5 to 2!) Ìlectronically controlled but SLOW (> 1 s) 34 Materials for integrated photonics Ferroelectrics Semiconductors (III-V) Dielectrics Polymers Si Detector Yes Yes Amplifier/Laser - El pumped Yes ?? - Opt pumped Yes Yes Yes Yes Yes Index change 1 Yes Yes (Yes) Yes Yes (plasma) Abs change 1 Yes ?? Opt nonlinearity Yes Yes Yes Yes Electronics integrate Yes Yes 1Electronically controlled _______________________________________________ Ferro electrics: LiNbO3, LiTaO3, strontium barium niobate (SBN)... Semiconductors: GaAs, InP, GaN ... ... Dielectrics: Glass, SiO2/Si... 35 Why LiNbO3? · Easy fabrication of low loss waveguides · Large electrooptic effect · Good RF properties 36 Semiconductor PICs: Bulk, quantum wells, quantum wires, quantum boxes: material engineering: manipulation of energy levels, density of states, matrix elements... L ...3,2,1; 2 * == nnL mE h 37 Requirements on semiconductor materials for photonic integrated circuits · Control of composition, doping and dimensions over areas > 1cm2 ( at least currently) · Mixing of "active" and "passive" elements on one chip · Mixing of forward and reversed biased pn-junctions on one chip => semiinsulating materials · Integration of electronic and photonic elements on one chip?? (immature technology and maybe questionable approach) 38 Devices in optical network elements · Couplers · Isolators and circulators · Multiplexers and filters · Optical amplifiers (covered in the photonics course, only network issues treated here) · Switches · Wavelength converters · Tunable devices · Transmitters and receivers · Boldface: Covered in this course 39 Types of photonic integrated circuits (PICs) Active · Switches · Modulators · Amplifiers · Filters ·(De)Multiplexers · Splitters, couplers ·Isolators, circulators ·... ·.... · (Optically) Nonlinear devices · Passive 40 Couplers 41 B Couplers 42Power question 43 Equations for directional coupler (codirectional coupling) Example of use of coupled mode equations (see eg T Tamir (ed), Integrated optics, Springer, 1975) 44 ( ) ( )zjAjB zjBjA 2exp 2exp -= --= directionnpropagatioisz,2t,coefficiencouplingis 21 -= ( ) ( )zjSBzjxpA exp,Re Set =-= ( ) 222222 2222 /)sin(zcosR(z) /)sin()( :Solutions ++++= ++-= zj zjzS 45 46 ©tepped coupler H. Kogelnik and R. V. Schmidt, IEEE J. Quantum Electron. QE-12, 395 (1976). ·1S 47 Modes of the composite structure ("supermodes") = · -- - = S R k S R M S R M S R ii ii S R + + 1 -:rsEigenvecto :sEigenvalue 22 22 i 48 49