AI summary by TwinMind
EUV With Fewer Mirrors?
Asianometry
21m 43s
ASML EUV System Limitations
- Mirror efficiency: 70% reflection per mirror, 30% absorption
- 10-12 reflections total → only 1% photons hit wafer
- Electrical power efficiency <0.2%
- Causes stochastic defects due to insufficient EUV photons hitting resist
- Quantum effects not overcome
Shintake's Proposed System Overview
- Professor Toru Shintake, Okinawa Institute of Science and Technology
- 30 years experience in electron accelerator science
- Built world's second X-ray free electron laser (Spring-8 Angstrom Compact)
- System complements, not challenges ASML
- Only 5 total reflections vs ASML's 12
Commercial EUV System Basics
- 13.5nm wavelength EUV light
- Light source → illumination module → photo mask → optical projection → wafer
- Industry standard exposure field: 26mm × 33mm
- Minimum viable: 13mm width (barely prints Apple A18 Pro chip)
- ASML variants: Low NA (0.33), High NA (0.55)
Historical EUV Development
- Early systems (Kinoshita): 2-mirror spherical Fox field optics
- Field curvature distortion limited exposure field size
- NTT 1992: 2-mirror aspherical equal radii system
- Convex primary + concave secondary mirrors
- Cancelled field curvature distortions
- EUV LLC consortium: 4-mirror projection, 0.1 NA, 24×32.5mm field
- Berkeley micro exposure tool (MET): 0.3 NA, 0.6×0.2mm field
- MET5 (2020): 0.5 NA, 0.2×0.03mm field, 5× magnification
- ASML Alpha demo (2006): 6-mirror system for larger field
- Required 200+ watts vs 10-20 watts for 2-mirror systems
Shintake System Design
- Light path: Source → illumination → mask → M1/M2 projection chamber → wafer
- Projection module: 2-mirror equal radii system (M1 + M2)
- Total reflections: 5 (vs ASML's 12)
- Key innovation: Dual line field illumination system
- Tool height: ~2m (6'6") - semiconductor fab ceiling limit
- Distance: Object-to-image maximized within fab constraints
Dual Line Field Illumination
- 4 smaller light sources bundled into 2 larger sources
- 2 fixed exposure slits (vs ASML's single slit)
- Prevents light reflection collision between sources
- Requires scan-and-step machine with moving mask
- Patent application filed by Okinawa Institute
Performance Specifications
- Current proposal: 0.2 NA → 24nm half pitch
- Upgraded version: 0.3 NA → 16nm half pitch
- Comparison:
- ASML low NA: ~13nm half pitch
- ASML high NA: ~8nm half pitch
- 193nm immersion: 1.35 NA, 40nm half pitch
- Field size: 13mm scan width, 4× reduction ratio
Technical Challenges
- Field curvature: Unavoidable with 2-mirror system
- Curved mask solution: Bend flat mask 30-120 microns
- $4M EUV masks made from ultra-low expansion glass
- Risk of cracking during bending
- Alternative: Manufacture curved blanks
- Mask writers: Need variable depth of focus capability
- M2 mirror: Must be thin and close to wafer while maintaining rigidity
Development Status
- Currently exists only in computer simulation
- Shintake studied 100+ papers over 2 years
- Used optical simulation software for 3 months
- Rewriting paper with industry-standard parameters
- Proof of principle experiment proposed as next step
Industry Context
- Media misrepresentation: "Japan breakthrough to break ASML monopoly"
- Reality: Complementary system, not displacement
- Requires entire ecosystem: light source, mirrors, masks, mechanics, software
- Many technical hurdles remain unforeseen
- Commercial viability requires extensive additional development
