AI summary by TwinMind
ASML’s Breakthrough 3-Pulse EUV Light Source
Asianometry
17m 57s
ASML EUV Roadmap Discovery
- Attended ASML presentation at Semicon Korea
- Key slide revealed new approach: three laser pulses instead of two
- Goal: grow EUV power source to 800+ watts
- Previously unclear how to achieve this beyond increasing tin pellet speed
EUV Light Generation Fundamentals
- 13.5 nanometer UV photon = 92 electron volts energy
- Visible light photons = 1.5-3 electron volts
- Requires highly ionized tin atoms (8-12 electrons removed)
- Tin atomic number = 50 electrons in neutral state
- Need high proportion of tin ions = plasma state
ASML's Power Scaling Equation (2024)
- CO2 laser power delivered to tin droplets
- Product of drive laser power × tin droplet repetition rate
- Tens of thousands droplets per second
- Conversion efficiency (CE)
- Measures inband UV light emitted/collected vs power input
- Critical for cost of ownership
- Power consumption example (Martin Vandenbrink):
- 1.5 megawatts from grid → 30 kilowatts laser → 100 watts EUV → 1 watt on wafer
- Additional factors: transmission losses, stable power level maintenance
Historical Development: Single Pulse to NOMMO
- Early LPP: single high-intensity CO2 laser, 50,000 pulses/second
- Droplet velocity: 130-260 mph (5-10x European swallow cruising speed)
- Challenge: hitting fast-moving droplets
- NOMMO (No Master Oscillator) architecture late 2000s:
- Droplet replaced one laser cavity mirror
- Self-timing system using droplet reflectiveness
- Elegant but failed: insufficient time for light buildup
- Tin debris accumulation in cavity
- NXE 3100 NOMMO produced only 10 watts (late 2010)
MOPA Development and Single Pulse Limitations
- Master Oscillator Power Amplifier (MOPA) replaced NOMMO
- Generated seed laser pulse, amplified to high power
- Problem: conversion efficiency struggled above 1%
- Issues with spherical droplets:
- Too dense - plasma light reabsorbed before reaching collector
- Required very powerful laser for small droplet ignition
- Risk of drilling holes or blasting droplets backward
- Surface plasma reflection of laser energy
- Potential equipment damage from reflected light
Two-Pulse System Development
- Added pre-pulse to reshape droplet before main blast
- Pre-pulse ablation process:
- Vaporizes surface material
- Pressure flattens droplet into disc
- Expands from 30 micrometers to 500 micrometers
- Center thickness: few nanometers
- Main pulse arrives nanoseconds later
- NXE 3100 MOPA + pre-pulse (2013): 50 watts
- NXE 3300 (2014): 80-100 watts with additional amplification
Current Scaling Limitations
- CO2 laser power increases:
- Extracting more gain from drive laser
- Droplet velocity: 50,000 → 62,000 droplets/second
- Total CO2 power increase: ~18%
- Plasma wind effect limitation:
- Droplet-to-plasma creates "fart-like" wind
- Preceding droplet wind impedes following droplet
- Limits droplet frequency scaling
Three-Pulse Innovation: Rarefaction Pulse
- New middle pulse: rarefaction pulse
- Purpose: reduce tin disc density before main pulse
- Creates lower density pre-plasma plume of preheated tin atoms
- Benefits vs solid disc:
- Reduced main pulse energy reflection
- Maximum tin atom absorption of main pulse energy
- Better ionization and UV light emission
- Minimal reabsorption of emitted EUV light
- Pulse characteristics:
- Same wavelength as pre-pulse (1 micrometer)
- Possibly shorter for better ionization
- Similar energy to first pre-pulse
Technical Implementation Challenges
- Critical timing requirements:
- Cannot hit pre-plasma plume immediately after second pulse
- Must wait for dispersion and rarefaction
- Cannot wait too long (density becomes too low)
- Both delays: microseconds long
- Adds substantial complexity with second microsecond delay
- Patent application details:
- Cameras measure droplet size, position, shape after pulses
- Computer feedback for laser alignment
- Lasers move in X, Y, Z directions
- Some recipes require droplet rotation
- Dynamic beam size adjustment for evolving plasma
Machine Calibration Workflow (Patent)
- Three-sequence startup process:
- Sequence 1: Low duty cycle (<1% droplets)
- Main pulse laser only, calibrate positioning
- Add pre-pulse laser after threshold
- Add rarefaction pulse laser last
- Sequence 2: EUV optimization
- Start with rarefaction pulse off
- Optimize beams 1 and 3 for EUV emission
- Bring back beam 2, adjust beams 2 and 3
- Sequence 3: High-duty refinement
- 80-90% droplet hit rate
- Optimize all three lasers for consistency and maximum UV energy
Expected Performance Gains
- Conversion efficiency improvement: 30-50%
- Significant debris generation reduction
- Target power: 800-1,000 watts within reach
- Potential shift from EUV photon starvation to abundance
- Major implications for semiconductor manufacturing scaling
