AI summary by TwinMind
Visiting IMEC’s ITF World 2024 in Belgium
Asianometry
11m 49s
Conference Overview
- Attended IMEC 2024 in Antwerp, Belgium - semiconductor industry's "watering hole"
- 40th anniversary of IMEC founding (1984-2024)
- Prestigious speakers, gala dinner with dance show
- Booths demonstrating new technologies, many focused on medicine/biology
New Era of Scaling Challenges
- Industry spoiled by lithography gains in 1980s-1990s
- Easy scaling through wavelength switching
- Early 2000s: 193nm lithography progress slowed to "turtle's crawl"
- Intel chased 157nm but never reached high volume production
- EUV now mature technology running several years
- No credible future wavelength progression planned
- Need alternative scaling methods without "biggest arrow in quiver"
Advanced Packaging Status
- Substantial products already using advanced packaging
- Major advancement with substantial remaining flaws
- Traditional silicon die: single plane, easy inspection/heat dissipation
- 2.5D integration: silicon interposer maintains advantages
- 3D integration: target destination but not ready
- Heat dissipation problems
- Inspection challenges
- Multiple companies working on solutions, still early stage
- Heterogeneous integration adds substantial complexity
- Industry consensus: one tool among many, not panacea
AI/Machine Learning Impact
- AI boom remains significant demand driver
- European skepticism vs Silicon Valley "scale pilled" mentality
- Focus on social issues/privacy
- Geographic distance from development
- More pessimism detected
- ML revolutionizing semiconductor manufacturing
- Optical Proximity Corrections evolved from simple tick marks to "abstract art"
- Conservative approach: foundries unlikely to use GPT-5 type systems
Sustainability Concerns
- Major discussion point in Europe
- Semiconductor manufacturing extremely energy intensive
- Fabs run 24/7, drawing power from carbon-intensive sources
- IMEC estimate: semiconductor manufacturing more carbon intensive than beef production or flight travel
- Data sharing problem: companies unwilling to share information
- IMEC developing cryptological platform for confidential data sharing/comparison
Forever Chemicals (PFAS) Challenge
- Semiconductor industry increasing attention to PFAS
- Companies like ASML drilling down supply chain exposure
- US opportunity per China Talk policy paper
- Two major concerns:
- Photo resist chemicals use PFAS in huge quantities
- Fundamental to semiconductor process
- Recent developments: EUV photo resists, metal oxide resists as alternatives
- PFAS extraordinarily prevalent across dozens of chemicals
- Few alternatives available, difficult choices required
- Industry proactive due to expected government regulation
- SAIA collecting information for potential roadmap
Lithography Future - Hyper NA EUV
- Former ASML President/CTO Martin van den Brink presentation
- High NA EUV: NA = 0.55
- Hyper NA EUV: NA = 0.75, roadmap for 2032 (8-9 years out)
- Previous skepticism about market room for larger lenses
- Economic feasibility concerns for continued NA increases
- No other optical lithography options at 13.5nm
- Potential re-evaluation of:
- Electron Beam projection lithography
- Electron Beam direct lithography
- Alternative approaches
Silicon Photonics Development
- Industry seeking "killer app" for volume/research stimulation
- Comparison to mobile smartphones driving chip scale packaging investment
- Potential interconnect space applications
- Light: less lossy, more power efficient communication
- Industry precedence for light replacing copper
- Current barriers: insufficient R&D or demand
- TSMC photonic engine technology "Coupe" as potential breakthrough
EU Chips Act Funding
- €2-3 billion allocated during show
- IMEC receiving portion of funding
- Less attention than US Chips Act or Asian initiatives
- Mixed reactions from industry:
- Skepticism: Europe lacks major demand drivers (mobile, data centers)
- Exception: automotive sector
- National security argument for advanced fab construction
- "Europe spending money anyway, why not semiconductors"
European Semiconductor Assessment
- Strong technology and research capabilities
- Smart people working on far-out technologies
- Much research won't commercialize (expected for nonprofits/universities)
- Learning process includes discovering what doesn't work
- Effective use of grad students vs expensive R&D personnel
