Microphotonics Hardware for the Information Age

Transcription

1 Microphotonics Hardware for the Information Age E-P synergy - Integration - Standardization - Cross-market Platforms Lionel C. Kimerling Materials Processing Center MIT Microphotonics Center

2 Acknowledgements CTR Fellows Editorial Advisory Board and Technology Working Group (TWG) Chairs Jerry Bautista, Intel, TWG Chair Richard Clayton, Clayton & Associates, TWG Chair Thomas Dudley, Pelorus Strategies, TWG Chair Louay Eldada, DuPont Photonics, TWG Chair Lionel C. Kimerling, MIT Randolph Kirchain, Director CTR, MIT Michael Morse, Intel, TWG Chair Dominic O'Brien, University of Oxford, TWG Chair Edward Sargent, University of Toronto Michael Schabel, Lucent Technologies, TWG Chair Jeffrey Swift, Independent, TWG Chair Christopher Weaver, Media Technology / MIT Fred Welsh, Radcliffe Advisors Katherine Butler Erica Fuchs Genta Gusho Sarah Kaplan Andjelka Kelic John McGrath Michael Speerschneider Yiwen Zhang CTR Staff Jon Bartels, Administrative Officer Mindy Baughman, Industry Consortium Program Manager Mark Beals, MPhC Associate Director Deborah Houben, Media Specialist George Kenney, MPhC Associate Director Tamarleigh Lippegrenfell, Publications Editor Trisha Montalbo, CTR Program Manager Frances Page, Office Manager

3 The CTR Program Technology Working Groups Current State of Transceiver Technology Silicon Optoelectronics Integration in III-V Materials Organics in Optoelectronics Objectives Evaluate communications technology Serve as a guide for R&D investment Perform analyses for a rational restructuring of the industry Deliverable: report documenting findings Technology Working Groups Analyze the photonics industry with support from MIT faculty and student. Gather industry leaders (40 companies) to discuss technology evolution. Editorial Advisory Board Build a consensus among the TWGs Prepare roadmap publications Lead formulation of an action agenda Support from the Economics Department and the Sloan School of Management. Market drivers and affordability are critical to sustainable innovation.

4 CTR Key Findings 1. Photonics technology will be driven by electronicphotonic convergence and short (<1km) reach interconnection. This direction will ignite a major shift in the leadership of the optical component industry from information transmission (telecom) to information processing (computing, imaging). 2. The skill set required for this path does not exist at any single institution. 3. We recommend that the Microphotonics Industry Consortium expand its focus toward the creation of the necessary competence and the recommendation of standards. Phase 2 meeting: 11/16-17/05 Our Goal: Building the path

5 Electronic-Photonic Convergence Relative Information Capacity (bit/s) OPTICAL FIBER SYSTEMS Communication Satellites Carrier Telephony first used 12 voice channels on one wire pair Telephone lines first constructed Year Multi-channel (WDM) 10 Mb/s km Advanced coaxial and microwave systems Early coaxial cable links Single channel (ETDM) 1. Distance bandwidth product > 10 Mb/s km 2. EMI, Crosstalk, Power, Weight, Footprint are important.

6 The End of the Roadmap: GHz off-chip and global clock rates 10 GHz local clock rates Low latency Low power Low crosstalk? 5000 pins 100 Tb/s on-chip 40 Tb/s off-chip ITRS 2000 CMOS: Frequencies won t exceed 10 GHz (power efficiency). performance scaling by parallelism System-level: Interconnects and photonics are getting more important with time. Cost per Gb/s for m long optical interconnects must reach $1 per Gb/s.

7 Cluster computing Cluster architectures provide value, and require lots of interconnect now the most common architecture for top 500 machines Top-tier (6 PF) machine of 2010: ~2.5M Gb/s Budget of $20 40M for optical transceivers in one machine! Alan Benner, IBM

8 Proliferation: Companies OFC 2005 Exhibition Too many suppliers Unable to achieve economies of scale Photonics industry is over-subscribed Death Spiral dynamic Dispersion Management Fibers and Cables Fiber Switches Communication Lasers Optical Amplifiers Passive Optical Devices Optical Cross Connects 401 Unique Companies Technology Optical Transmitters and Receivers Current companies with supporting products Amplifiers Attenuators Detectors Fiber Lasers Technology Modulators Wavelength Mux Dispersion Compensators Optical Switches Transceivers 276 Unique Companies Current companies with supporting products Lightreading Business Index

9 Proliferation: Technologies Optical Transceiver Variants ~ Application SAN, LAN, WAN/MAN Bit Rate Gb/s, Gb/s, 1 Gb/s, 2.5 Gb/s, 10 Gb/s Wavelength 850 nm, 1310 nm, 1550 nm Reach SR, IR-1, IR-2, LR Form Factor SFF, SFP, GBIC, XFP, MSA, Other Approx. number of logical exclusions M. J. Speerschneider, "Technology and Policy Drivers for Standardization: Consequences for the Optical Components Industry," MS Thesis in Materials Science and Engineering. Boston: MIT, Differentiation to preserve market share Prevents achieving economies of scale 1M units shipped 100,000 threshold Support at most 10 transceiver solutions

10 The Death Spiral Dynamic Proliferation loop earnings (-) - Economies of Scale Costs - Communications and Interconnects Perceived Benefits of Optical - R - Product Variety Transceiver Sales Standardization B Desire to Differentiate from Competitiors total addressable market (-) Transceiver Demand Efforts to Gain Market Share Efforts to Increase Total Market Perceived Need to Do Something - Sufficiency of Business (Or, Total Capacity Ultiliaztion) Speerschneider, MIT

11 The Revenue Dynamic Broadband Demand Forecast Demand for Broadband Costs - - New Optical Network Build B Bit Rate * Distance POTS vs performance scaling Network Build lifecycle? Technology obsolescence? R&D to Reduce R&D to Improve Costs Performance 2e014 Total Broadband Demand Transceivers Sales Transceiver Revenue 800 M Revenue 1.75e M 1.5e M legacy dominance build complete 1.25e M 1e Time (Year) Speerschneider, Kelic, MIT Time (Year) Total Broadband Demand : step 1.25e13 newtable (6-7) m*mbps Revenue : step 1.25e13 newtable (6-7) dollars/year Total Broadband Demand : eq newtable (6-7) m*mbps Revenue : eq newtable (6-7) communications technology dollars/year roadmap

12 Photonics Economics Cannot wait for upturn in investment in telecom: yr network build intervals Currently too many firms in industry Standards and consolidation will help decrease cost and price However, unlikely to solve the industry problem because of low cost share of final product Need to develop new demand drivers that will shift out the demand curve. Hausman, MIT

13 Emerging Markets beyond Telecom Designing a System for 2015 Networks and servers (interconnection) Automotive (interconnection, DVI) The digital home (interconnection, DVI) Entertainment (interconnection, DVI) Government and defense (rf signal processing) Who will make the transceiver? What will the cost be? When will it be available? What will its performance be? Cross-market standardization?

14 Materials Platform: III-V Potential InP better suited for optics than silicon Monolithic or hybrid integration with silicon Barriers Industry suffers from over-capacity Fragmented market and supply base Likely to remain dominant materials system for high-performance devices

15 Materials Platform: Organics Potential Used for packaging and transmission media Already demonstrated complex circuits on polymer waveguide platform Process at lower temperature than semiconductors Rapid manufacturing Integration with electronics Barriers Need effective pick-and-place technology for hybrid integration Likely to play key role in commercial, hybrid photonic circuits

16 Materials Platform: Silicon Potential Integration with CMOS using mainstream processes Leverage from silicon processing infrastructure Very low cost integration New applications with EP convergence Barriers Light sources Packaging and interconnection infrastructure Has potential to provide the majority of low-cost photonic interconnects in the medium to long term

17 Building a Technology Platform Market penetration by scalability Functionality (EP synergy) Performance (Hz/W. cm 2 ) Yield (infrastructure) Cost (volume production) Performance = bandwidth/(power x area) Performance/cost > 100x/10yrs

18 EW AS-EPIC Program Objectives To demonstrate the world s first densely integrated Application Specific Electronic Photonic Integrated Circuit (AS-EPIC) using an electronic warfare (EW) application as a demonstration vehicle. Approach Integrate the best technology and designs from BAE Systems, Lucent Technologies, MIT, and AWR to realize our AS-EPIC chip. This involves combining CMOS compatible, low loss, high index contrast (HIC) waveguides and electro-optic components to form optical filters, modulators, and detectors. 4.5X Increase IBW 95X Reduction Size 80X Reduction in Weight 5X Reduction in *Power* 100X Reduction in Cost Nickel Size EPIC RF Photonic Channelizer Chip Bell Laboratories Lucent Technologies PAGE 18

19 Ge-on-Si Photodetector Integration (Bandwidth) x (Quantum efficiency) Size : 5µm 20µm Q.E: 90% transit time limit Waveguide-Integrated, EPIC Photodetector RC time limit d=0.5um d=1.0um d=1.5um d=2.0um Detector Size (µm 2 ) Discrete, free-space Photodetectors Cannon, Ahn, Michel, MIT

20 Electronic-Photonic Convergence PHOTONICS Driver Fiber, lasers, detectors MUX, EDFA Metro-fiber, PLC Transceiver µph ICs FTTH Pervasive, µph ICs Transmission Application ETDM WAN DWDM WAN Security Acces/SAN/LAN 1 Gb/s Access 10 Tb/s WAN Optical switching systems Trend Fiber Fiber pigtail Boards, Servers Optical MCM Optical Nodes ELECTRONICS Driver IC: Al/SiO 2 GaAs IC: Cu/SiO 2 InP Optical bus On-chip optical interconnects Optical switch Processing Application S/DRAM, ASIC, µproc MIMIC DSP µproc TIA Parallel processing EP signal conditioning EP signal processing Trend Yield Yield Shrink Yield Optical interconnection EP design Photonic logic Timeline for commercial deployment

21 Convergence Success Strategy: Vision 2015 Standard component platform. Common manufacturing infrastructure. Productive R&D reduces development cycle. Common platform across industry sectors. Grow platform to $20B/yr revenue. Technology obsolescence triggers sales through performance scaling.

22 Missing Infrastructure Materials Design Process tools Packaging Test Packaging $2 chips, but $22 connections Fiber platform to a chip platform Light from chip to board to backplane Chip carrier without permanent fiber attach

23 CTR: Phase 2 Technology Targets Materials large area wafers; component integration compatibility Processes tool standardization, common processes, process control, process integration Packaging Infrastructure an optical chip carrier without a permanent fiber attach chips; boards; backplanes; interbox; LAN; FTTH; MAN; WAN Test common test platform; wafer level testing; global standard test Design common design tools; common form factors; reduce complexity; focus on functionality methodology for electronic/photonic partitioning photonic circuit theory to support appropriate simulation tools