Moving towards generic foundries for photonic integrated
Transcription
Moving towards generic foundries for photonic integrated
PHOTONICS RESEARCH GROUP Moving towards generic foundries for photonic integrated circuits Building Roel Baets on the silicon CMOS technology base Photonics Research Group, Ghent University – IMEC Center for Nano- and Biophotonics, Ghent University roel.baets@ugent.be PHOTONICS RESEARCH GROUP 1 Integrated Optics / Photonic Integration A little bit of history • 1969 + seventies: foundational basis of integrated optics (Marcatili, Marcuse, Yariv, Kogelnik...) • eighties and nineties: • discrete integrated optic devices have a game-changing impact on optical telecoms, optical recording, metrology, high power lasers, … • photonic integrated circuits (PICs) do NOT leave the research labs • 2000-2010: • the dot-com bubble and subsequent bust damage the field badly • emergence of major research initiatives with focus on PICs • small number of industrial PIC product developments PHOTONICS RESEARCH GROUP 2 Barriers causing slow industrial take-up “Roadmap” Large diversity in materials and technology options No “standards” Cost Investment cost in PIC fab + NRE costs for PIC-process are HUGE Food chain Lack of suitable design tools and packaging solutions Knowledge Lack of knowledge about potential of PICs in application fields (other than telecom) 3 PHOTONICS RESEARCH GROUP Mitigating the Barriers “Roadmap” Large diversity in materials and technology options No “standards” Consolidation on FEW “generic” technology platforms: Silicon, InP, glass/nitride Cost Investment cost in PIC fab + NRE costs for PIC-process are HUGE Create access to “generic” technology platforms: develop a “foundry” model Food chain Lack of suitable design tools and packaging solutions Identify and involve all stakeholders Develop standards Knowledge Lack of knowledge about potential of PICs in application fields (other than telecom) Develop demonstrators with convincing performance PHOTONICS RESEARCH GROUP 4 “Generic”: what’s in a name make use of standardized building blocks and a highly reproducible process flow that can serve many functions • no (or little) design freedom in the out-of-plane direction • large design freedom in the in-plane direction • benefits: • few process flows to maintain: high yield and moderate NRE-costs • cost sharing in a MPW-approach both for masks and for processing • result: “do more with less” (and do so at lower cost) PHOTONICS RESEARCH GROUP 5 The generic technology platforms • InP • “all-in-one” • unique for light emission and amplification • existing fab capacity (discrete photonic components) • silicon • CMOS technology base with existing 200-300 mm fab capacity • SOI: standard wafer with built-in high-index contrast • standardization driven by available CMOS process technology and by SOIwafer availability • CMOS community embraces silicon photonics as a potential interconnect solution • glass/nitride • also builds on silicon technology and infrastructure • adjustable index contrast • very low loss • very broad wavelength coverage (including visible) PHOTONICS RESEARCH GROUP 6 Building the “foundry model”: what does it take? Network of Excellence: ePIXnet (2004-2008) Large-scale research projects • consensus building on the roadmap • initiate action in Europe • silicon: HELIOS • InP: EuroPIC, PARADIGM MPW-services • typically limited to the most basic generic process • ePIXfab, IME, OpSIS, JePPIX, Triplex Small volume PIC-manufacturing • typically a generic process + some customization • silicon: IMEC; Si3N4/SiO2: Lionix; InP: several Industry-driven technology development • Optical I/O program at IMEC (serving CMOS industry) Supporting actions • create awareness in various application domains • training Medium to large volume manufacturing • when and where? PHOTONICS RESEARCH GROUP 7 European Network of Excellence on Photonic Integrated Components and Circuits Mission: To provide Europe with a strategic advantage in Photonic Integration by joining forces of Europe’s key players and organising access to unique and expensive technological infrastructure 2004-2009 32 Partners + 18 Affiliates • • • 22 Academic 11 research institutes 17 industrial 1st European Photonic Integration Forum, December 10 2008 ePIXnet Platforms • Organize access to: • Very expensive and unique infrastructure • Offering stable and mature technologies • For research on photonic integration • With economic viability through cost sharing • With a roadmap towards industrial foundries 1st European Photonic Integration Forum, December 10 2008 ePIXnet: integration platforms JePPIX ePIXfab Nanostructuring InP-based active/passive integration process - Silicon photonics Nanostructuring platform for Photonics Integration - PIC design and training - PIC design support -Wafer scale processing in CMOS lines - Electron Beam lithography - PIC Fabrication - IMEC and LETI 200mm pilot lines (24/7) ePIXpack Persyst Cluster Computing - Flip-chip and assembly - High speed component characterisation CLUSTER with 247 CPUs (part of BEgrid) - Fiber pig-tailing - Packaging - Dry etching - MEEP - Telecom system transmission tests 1st European Photonic Integration Forum, December 10 2008 - CAMFR ePIXnet: integration platforms JePPIX ePIXfab Nanostructuring InP-based active/passive integration process - Silicon photonics Nanostructuring platform for Photonics Integration - PIC design and training - PIC design support - PIC Fabrication -Wafer scale processing TripleX in CMOS lines - Multi-project of 200mm Silicon - IMEC andruns LETI nitride/silicon oxide PICs pilot lines (24/7) - Electron Beam lithography - Dry etching - Based on LioniX technology ePIXpack Persyst - Coordinated by U. Twente - High speed component characterisation - Flip-chip and assembly - Fiber pig-tailing - Packaging Cluster Computing CLUSTER with 247 CPUs (part of BEgrid) - MEEP - Telecom system transmission tests - CAMFR 1st European Photonic Integration Forum, December 10 2008 ePIXfab MPW shuttle service send in design users mask integration fabrication Some facts (since 2006) wafers distributed 16 MPW shuttles, 187 designs 65 different users: 44 universities, 12 research institutes, 7 SMEs, 1 larger company 37 Europe, 15 N-America, 10 Asia, 2 Australia, 1 Middle-East 5 training events, 4 workshops >650 website visitors, >10000 website visits 12 ePIXfab Access to silicon photonics technology for academic and industry R&D MPW service Training Outreach events CAD Transfer to production Partially funded by the EC www.epixnet.org www.photonfab.eu 13 Generic Foundry Model Partners: Europe’s leading • Chip manufacturers & pack. • Photonic CAD companies • Equipment Manufacturers • Research Labs Oclaro, FhG-HHI, III-V Lab, CIP Phoenix, Photon Design, Filarete ASML, Aixtron, Oxford Plasma Technologies COBRA – TU/e (coordinator), Cambridge Program: Step 1 Small scale access to COBRA process for research purposes (from 2007) Step 2 Investigate the feasibility of industrial foundry operation - EuroPIC (2009, 5.5 M€, 17 partners, platforms Oclaro, HHI) - PARADIGM (2010, 13 M€, 17 partners, platforms Oclaro, HHI & COBRA) - Dutch STW & IOP (2011, many partners, > 6 M€ total budget) Target: commercial foundry operation before 2014 A revolution in Photonic Integration TNO, Eindhoven, 30 mei 2011 14/33 Diversity in Genericity • A generic technology approach limits the technical and scientific degrees of freedom, BUT… • The generic processes evolve over time • Endless potential in “hybrid” approaches: generic process enriched with custom post-processing • III-V on silicon hybrid approaches (FP7-HELIOS) • Silicon-Organic-Hybrid (FP7-SOFI) • Si3N4 waveguides on silicon waveguides + Ge detectors (Bell Labs – IME) PHOTONICS RESEARCH GROUP 15 Integrated Optics / Photonic Integration A little bit of history • 1969 + seventies: foundational basis of integrated optics (Marcatili, Marcuse, Yariv, Kogelnik...) • eighties and nineties: • discrete integrated optic devices have a game-changing impact on optical telecoms, optical recording, metrology, high power lasers, … • photonic integrated circuits (PICs) do NOT leave the research labs • 2000-2010: • the dot-com bubble and subsequent bust damage the field badly • emergence of major research initiatives with focus on PICs • small number of industrial PIC product developments • 2011-2020: • generic PIC-technologies in a foundry-like model become a reality PHOTONICS RESEARCH GROUP 16