1. Scanner Architecture Overview
Wafer Stepper vs Scanner
The evolution from steppers to step-and-scan systems represents a fundamental shift in lithography tool architecture. Steppers expose the entire field at once, while scanners expose the field by synchronously scanning the reticle and wafer.
| Parameter |
Stepper |
Scanner |
| Exposure Mode |
Full-field flash |
Slit scan (synchronous) |
| Field Size |
22 x 22 mm (limited) |
26 x 33 mm (larger) |
| Lens Quality |
Must be uniform over full field |
Only slit region critical |
| Dose Uniformity |
Single flash variation |
Integrated over scan |
| Throughput |
Lower (large fields) |
Higher (efficient scanning) |
Reduction Optics Principle
Lithography scanners use reduction projection optics, typically 4× or 5× reduction, to minimize mask defects and enable finer patterning. The reduction factor M relates mask and wafer dimensions:
Reduction Factor: M = 4× (standard) or 5× (advanced)
Wafer Feature = Mask Feature / M
Example: 40nm wafer line → 160nm mask line (4×)
Benefits:
- Mask defect size reduced by M
- Mask writing resolution relaxed by M
- Mask inspection easier
Drawbacks:
- Larger mask (expensive)
- More mask making complexity
Synchronous Scanning Mechanism
In a scanner, the reticle stage and wafer stage move synchronously at velocities related by the reduction ratio. A narrow illumination slit scans across the reticle while the wafer moves in the opposite direction.
Velocity relationship: V_wafer = V_reticle / M
Typical scan speeds:
Reticle: 500-800 mm/s
Wafer: 125-200 mm/s (for 4× reduction)
Slit width: 6-8 mm at wafer
Scan length: 26-33 mm
Scan time: ~130-200 ms per field
Advantage of Scanning: Aberrations and field curvature are averaged over the scan direction, improving uniformity. Only the slit region needs to be well-corrected, allowing larger fields with fewer lens elements.
2. Illumination System
ArF and KrF Excimer Laser Sources
Deep UV lithography employs excimer (excited dimer) lasers that generate UV light through rare gas halide reactions:
| Laser Type |
Wavelength |
Active Medium |
Applications |
| KrF |
248 nm |
Krypton Fluoride |
130nm-250nm nodes |
| ArF |
193 nm |
Argon Fluoride |
45nm-90nm nodes |
| ArF Immersion |
193 nm (in water) |
Argon Fluoride + H₂O |
7nm-45nm nodes |
ArF Laser Specifications:
- Wavelength: 193.368 nm ± 0.0003 nm (FWHM bandwidth < 0.3 pm)
- Pulse energy: 5-15 mJ per pulse
- Repetition rate: 2000-6000 Hz
- Average power: 60-90 W
- Pulse duration: 10-25 ns
- Energy stability: < 0.5% (3σ)
- Beam pointing stability: < 5 μrad
Beam Delivery and Homogenization
The raw laser beam is spatially and temporally inhomogeneous. The illumination system transforms this into a uniform, controlled illumination at the reticle:
- Beam Expansion: Telescope expands beam from ~10mm to ~50-100mm diameter
- Beam Shaping: Diffractive or refractive elements modify beam profile
- Pulse Stretcher: Lengthens pulses to ~100ns, reducing peak intensity and mitigating lens damage
- Energy Monitor: Inline sensor measures pulse energy for dose control feedback
Fly's Eye Homogenizer
The fly's eye integrator creates uniform illumination by dividing the beam into many sub-beams, each forming an image of the source. These sub-images are superimposed at the reticle:
Fly's eye configuration:
- Array of small lenslets (5-10mm pitch)
- Two fly's eye arrays in sequence
- Each lenslet pair creates one source image
- Fourier lens superimposes all images
Homogenization:
- Uniformity: < ±0.5% (3σ) across field
- Pupil fill: Controlled by fly's eye aperture
- Telecentricity: < 1 mrad at reticle
Pupil Aperture Control
Variable apertures at the pupil plane control illumination shape (conventional, annular, dipole, quadrupole, freeform). Modern scanners use programmable illumination with up to 1000+ independently controlled mirror elements or LCD pixels.
FlexRay/FlexWave Systems: Advanced illumination uses micromirror arrays or LCD spatial light modulators to create arbitrary source shapes optimized for specific patterns. Each mirror/pixel can be individually controlled to achieve source shapes with < 1% pupil discretization.
3. Projection Lens System
4× and 5× Reduction Lens Design
The projection lens is the most critical and expensive component of a lithography scanner, containing 20-40 lens elements with < 0.1nm surface figure accuracy:
| Parameter |
ArF Dry |
ArF Immersion |
EUV |
| Reduction |
4× |
4× |
4× |
| NA |
0.75 - 0.93 |
1.20 - 1.35 |
0.33 - 0.55 |
| Elements |
~30 lenses |
~40 lenses |
6 mirrors |
| Weight |
~800 kg |
~1000 kg |
~1500 kg |
| Cost |
$5-10M |
$10-20M |
$50-100M |
Catadioptric Design
High-NA lenses often employ catadioptric (refractive + reflective) designs to manage chromatic aberration and packaging constraints. Mirrors fold the optical path, allowing extreme NAs within reasonable physical dimensions.
NA Specifications and Trade-offs
Numerical aperture determines resolution and depth of focus through the fundamental equations R = k₁λ/NA and DOF = k₂λ/NA². Immersion lithography increases effective NA beyond the n=1 limit of air:
Maximum NA in air: sin(θ_max) = 1.0 → NA = 1.0
Maximum NA in water (n=1.44): NA = n·sin(θ) = 1.44
Practical limits:
ArF dry: NA < 0.93 (to avoid total internal reflection)
ArF immersion: NA < 1.35 (limited by water n, resist n, incident angle)
At NA=1.35, water gap: ~2-4mm, controlled to ±1μm
Field Size and Aberration Correction
Larger field sizes require more lens elements and tighter aberration correction. Modern scanners achieve 26×33mm fields with wavefront error RMS < λ/80:
- Aberration Budget: Total RMS < 2.5nm (at 193nm), distributed across Zernike terms
- Adjustable Elements: 6-12 elements with piezo actuators for dynamic correction
- Temperature Control: Lens barrel held at 22.0°C ± 0.001°C
- Pressure Control: Interstitial gas (N₂ or He) pressure stabilized to < 0.1 Pa
Lens Material: Fused silica (SiO₂) is the primary material for DUV optics due to high transmission at 193nm and low CTE (coefficient of thermal expansion ~0.5 ppm/K). Calcium fluoride (CaF₂) is used in some elements for chromatic correction.
4. Reticle Stage
6-DOF Positioning System
The reticle (mask) stage provides six degrees of freedom: three translations (X, Y, Z) and three rotations (Rx, Ry, Rz = tip, tilt, rotation). This enables precise positioning and scanning:
Reticle Stage Specifications:
- X/Y travel: ±50mm (for 6" reticle)
- Z travel: ±5mm
- X/Y positioning accuracy: < 2nm (3σ)
- Scan velocity: 500-800 mm/s
- Scan acceleration: 10-20 m/s²
- Scan synchronization error: < 1nm (dynamic)
- Tilt range: ±100 μrad
- Rotation (Rz): ±500 μrad
Interferometric Position Feedback
Laser interferometry measures stage position in real-time with sub-nanometer resolution. Multiple interferometer beams (typically 3-5 per axis) measure position and tilt:
- Laser Wavelength: Stabilized HeNe laser (632.8nm) or frequency-doubled Nd:YAG (532nm)
- Measurement Principle: Heterodyne or homodyne interferometry with phase detection
- Resolution: λ/1024 to λ/4096, corresponding to 0.15-0.6nm per count
- Bandwidth: 10-20 kHz for position feedback control loop
- Air Compensation: Temperature, pressure, humidity sensors correct for refractive index changes in air
Thermal Control
Reticle and stage thermal management prevents drift and distortion:
Temperature control:
- Reticle: 22.0 ± 0.01°C
- Stage frame: 22.0 ± 0.005°C
- Interferometer air path: < 0.05°C variation
Heating sources:
- Motor dissipation: 50-200W (actively cooled)
- Bearing friction (air bearings: minimal)
- Reticle absorption: ~1-5 mW (laser exposure)
Vibration Isolation
The reticle stage is isolated from building vibrations via active damping systems. Target: < 1nm RMS vibration above 1Hz to maintain overlay and imaging quality.
Reticle Flatness: Masks are specified to < 0.25μm peak-to-valley flatness over the pattern area. Electrostatic or vacuum clamping flattens the reticle against a reference chuck with < 50nm flatness. Pressure control in the clamp gap maintains < 10nm stability during scan.
5. Wafer Stage
High-Speed XY Scanning
Wafer stage performance directly impacts throughput and overlay. Modern stages achieve > 600 wafers per hour at 26×33mm field size:
Wafer Stage Performance:
- XY travel: ±150mm (for 300mm wafer)
- Max velocity: 800-1200 mm/s
- Max acceleration: 20-40 m/s² (~4g)
- Settle time: < 10ms after step
- Positioning accuracy: < 1.5nm (3σ, on-the-fly)
- Overlay: < 2nm (mean + 3σ, after corrections)
- Throughput: 250-300 wafers/hour (300mm wafer)
Z Focus and Tilt Control
Maintaining the wafer surface within the depth of focus (typically 50-150nm) requires active Z focus and tilt control:
- Z Range: ±1mm for wafer bow, chuck flatness, and focus offsets
- Tilt Range: ±500 μrad to accommodate wafer non-flatness
- Leveling Sensors: Optical sensors measure wafer surface height before exposure (pre-scan leveling)
- Leveling Accuracy: < 20nm (3σ) over exposure field
- Dynamic Focus: Real-time Z adjustment during scan based on pre-measured map
Laser Interferometry
Multiple interferometer axes measure wafer stage position with < 1nm accuracy:
Interferometer Configuration:
- X axis: 2-3 beams (position + Ry tilt)
- Y axis: 2-3 beams (position + Rx tilt)
- Z axis: 3-4 beams (height + Rx/Ry tilts)
Measurement rate: 10-50 kHz
Servo bandwidth: 200-500 Hz (position), 50-100 Hz (tilt)
Non-linearity: < 0.5nm over full travel
Air Bearing or Maglev Technology
Two competing technologies for wafer stage motion:
| Parameter |
Air Bearing |
Magnetic Levitation |
| Friction |
Near-zero (viscous air drag) |
Zero (contactless) |
| Stiffness |
Moderate (air gap ~5-10μm) |
High (actively controlled) |
| Power Consumption |
Low (compressed air) |
Higher (electromagnets) |
| Vacuum Compatibility |
Not compatible |
Compatible |
| Heat Dissipation |
Excellent (air cooling) |
Requires active cooling |
Twin-Stage Architecture: Advanced scanners use two wafer stages: one for exposure, one for wafer exchange and pre-alignment/leveling. While one wafer is exposed, the next is prepared, improving throughput by 40-50% compared to single-stage systems.
6. Alignment Systems
Through-the-Lens (TTL) Alignment
TTL alignment detects wafer marks through the projection lens, ensuring mark and exposure use the same optical path, minimizing baseline errors:
TTL Alignment Specifications:
- Wavelength: Typically 633nm (HeNe) or broadband (500-900nm)
- Detection: Diffraction grating marks, image processed
- Precision: 0.5-1.0nm (3σ) per mark
- Capture range: ±5μm
- Mark types: Box-in-box, grating, Vernier
- Mark size: 10-40μm × 10-40μm
Field Image Alignment (FIA)
FIA systems image alignment marks onto CCD/CMOS sensors, providing high-speed, robust mark detection even on process-heavy wafers with resist, films, or topography:
- Simultaneous Marks: Images multiple marks (4-8) in parallel for speed
- Wavelength Diversity: Multiple wavelengths (RGB or broader) to penetrate film stacks
- Polarization Control: Reduces sensitivity to birefringent films
- Processing: Cross-correlation or template matching in real-time (< 5ms per mark)
Enhanced Global Alignment (EGA)
EGA models wafer distortion (translation, rotation, scaling, orthogonality, higher-order terms) using measurements from a sparse set of alignment marks:
EGA Model:
ΔX = Tx + Sx·X + Ry·Y + Ox·X² + Oxy·X·Y + ...
ΔY = Ty + Rx·X + Sy·Y + Oy·Y² + Oyx·X·Y + ...
where:
Tx, Ty: Translation
Sx, Sy: Scaling (magnification)
Rx, Ry: Rotation (skew)
Ox, Oy, Oxy, Oyx: Orthogonality and higher-order terms
Typical: 20-40 marks measured → model fits all field positions
Residual: < 2nm (3σ) after model correction
Wafer Grid Optimization
The wafer is modeled as a deformed grid due to process stress, thermal effects, and chuck distortion. Grid optimization determines optimal field positions and corrections:
Alignment Strategy: Sample alignment (measure subset of marks, model full wafer) reduces overhead to 1-3 seconds per wafer. Fine alignment (measure marks near each field) adds 0.2-0.5s per field but improves overlay to < 2nm for critical layers. Die-by-die alignment (measure every die) is used only for ultra-critical layers or R&D.
7. TIS Measurement & Calibration
Tool Induced Shift (TIS)
TIS quantifies the overlay error introduced by the scanner itself, independent of wafer processing. It includes baseline shift, reticle heating, lens heating, and stage errors:
TIS Components:
1. Baseline (fixed offset between alignment and exposure): Δ_baseline
2. Reticle heating (absorption causes thermal expansion): Δ_reticle(dose)
3. Lens heating (absorption changes magnification): Δ_lens(time, dose)
4. Stage repeatability (position noise): Δ_stage
5. Chuck distortion (clamping force, thermal): Δ_chuck
Total TIS: TIS² = Σ(Δi²)
Target: TIS < 1nm (mean + 3σ) for advanced nodes
Long-Term Drift and Stability
Scanner performance drifts over time due to lens element shift, thermal transients, and mechanical relaxation. Daily and weekly calibrations maintain specifications:
- Drift Rate: Baseline drift < 0.2nm/hour (stable after warmup)
- Magnification Drift: < 0.05 ppm/day (temperature-corrected)
- Focus Drift: < 5nm/hour (Z-axis thermal stability)
- Rotation Drift: < 2 μrad/day
Calibration Wafers and Procedures
Reference wafers with accurately placed marks provide ground truth for overlay calibration:
Calibration Wafer Specifications:
- Mark placement accuracy: < 0.5nm (3σ, certified by metrology lab)
- Grid: 5×5 to 9×9 across wafer
- Recertification: Every 3-6 months
Calibration Procedure:
1. Measure calibration wafer with scanner alignment system
2. Compare to certified mark positions
3. Compute correction parameters (baseline, magnification, rotation)
4. Update scanner calibration tables
5. Verify with second measurement (residual < 1nm)
Temperature Compensation
Magnification and overlay vary with temperature. Compensation models correct for measured temperature deviations:
Magnification temperature sensitivity: ~0.5 ppm/°C
Overlay temperature sensitivity: ~2-5 nm/°C (due to stage expansion)
Temperature sensors: 50-200 distributed in scanner
Model: Mag(T) = Mag₀ · [1 + α·(T - T₀)]
Real-time compensation updates every 10-60 seconds
Matching Between Tools: In high-volume manufacturing, multiple scanners must be matched (overlay < 1.5nm scanner-to-scanner). Matching wafers are exposed on each tool, measured, and tool-specific offsets are applied. Monthly re-matching maintains fleet uniformity.
8. Environmental Control
Temperature Stability (±0.01°C)
Lithography scanners require extreme temperature stability to maintain overlay and CD control. Typical fab cleanroom setpoint: 22.0°C, but scanner requires tighter local control:
Temperature Control Zones:
- Lens barrel: ±0.001°C (tight control, direct cooling)
- Reticle stage: ±0.005°C (active water cooling)
- Wafer stage: ±0.01°C (water-cooled base)
- Metrology frame: ±0.01°C (reference structure)
- Ambient (scanner enclosure): ±0.05°C
Cooling system:
- Chiller capacity: 20-50 kW heat removal
- Water flow: 50-100 L/min
- Water temperature: 20.0 ± 0.01°C
- Parallel cooling loops for different zones
Barometric Pressure Control
Air pressure affects interferometer readings (refractive index of air) and needs compensation or active control:
Pressure sensitivity: ~0.3 ppm per mbar (for interferometry)
Ambient pressure variation: ±10 mbar (daily weather)
→ ~3 ppm error if uncompensated → 3nm overlay at 1mm distance
Solutions:
1. Active compensation: Pressure sensors + correction formula
n_air = f(P, T, humidity) → apply to interferometer readings
2. Pressure-controlled enclosure: Maintain P = 1000 ± 0.1 mbar
Modern scanners: Compensation accurate to < 0.05 mbar equivalent
Vibration Isolation (< 1nm RMS)
Building vibrations must be attenuated to sub-nanometer levels. Multi-stage isolation achieves this:
- Passive Isolation: Air spring mounts (resonance ~1-3 Hz) provide base isolation
- Active Isolation: Servo-controlled actuators (bandwidth 1-100 Hz) cancel vibrations
- Metrology Frame: Separate from base frame, provides vibration-isolated reference for interferometers
- Target Performance: < 0.5nm RMS vibration 1-100 Hz, < 1nm total up to 1000 Hz
Cleanroom Requirements (ISO Class 1-2)
Particles on wafer or reticle cause defects. Stringent cleanroom and contamination control are mandatory:
Cleanroom Classification:
- General fab: ISO Class 4-5 (10,000-100,000 particles/m³ > 0.1μm)
- Litho bay: ISO Class 2-3 (100-1,000 particles/m³ > 0.1μm)
- Wafer/reticle handling: ISO Class 1 (< 10 particles/m³ > 0.1μm)
Contamination Control:
- HEPA/ULPA filters: 99.9999% efficiency for > 0.1μm
- Laminar flow: 0.3-0.5 m/s downflow velocity
- Air changes: 300-600 per hour
- Wafer stored in FOUP (Front Opening Unified Pod): Sealed, purged with N₂
- Reticle in SMIF pod (Standard Mechanical Interface): Sealed environment
Acoustic Noise: Scanner motors, air bearings, and vacuum pumps generate acoustic noise that can couple to stages as vibration. Acoustic enclosures and quiet pump designs reduce noise to < 70 dBA at operator position and minimize vibration coupling.
9. Lithography Process Flow
Complete Process Sequence
A typical DUV lithography process involves 10-15 steps from wafer arrival to final inspection:
- Wafer Input: FOUP delivered to loadport, wafer extracted by robot
- HMDS Prime (Adhesion Promoter): Hexamethyldisilazane vapor treatment, 90-150°C, 30-60s. Replaces surface -OH groups with hydrophobic -Si(CH₃)₃, improving resist adhesion
- Resist Coat: Spin coating at 1000-4000 RPM, 30-60s. Target thickness: 50-300nm depending on application
- Soft Bake (PAB - Post-Apply Bake): 90-130°C, 60-90s on hotplate or in oven. Removes casting solvent, densifies resist film, reduces standing waves
- Edge Bead Removal (EBR): Solvent spray removes resist from wafer edge (2-3mm) to prevent contamination and chuck particles
- Wafer Transfer to Scanner: Automated track-to-scanner interface
- Pre-Alignment & Leveling: Optical sensor maps wafer surface height, 1-3 seconds
- Alignment: Measure alignment marks, compute corrections, 2-5 seconds
- Exposure: Scan each field with calibrated dose, 20-60 seconds per wafer
- Post-Exposure Bake (PEB): 90-130°C, 60-90s. Acid diffusion and deprotection occur during PEB for chemically amplified resists
- Development: Immersion or spray of 0.26N TMAH (tetramethylammonium hydroxide) developer, 30-90s. Rinse with DI water
- Hard Bake: 100-150°C, 60-120s. Stabilizes resist for etch, improves adhesion and etch resistance
- Post-Development Inspection (ADI - After Develop Inspection): Optical defect inspection, CD metrology on test structures
- Wafer Output: Return to FOUP, transfer to etch
Process Parameters and Optimization
| Parameter |
Typical Range |
Impact |
| Resist Thickness |
50-300 nm |
Thinner → less LER, worse etch resistance. Thicker → opposite |
| PAB Temperature |
90-130°C |
Higher → more outgassing, lower sensitivity. Lower → incomplete solvent removal |
| Exposure Dose |
15-40 mJ/cm² |
Higher → smaller CD, more acid, faster development |
| PEB Temperature |
90-130°C |
Higher → more diffusion, larger CD loss, smoother LER |
| Development Time |
30-90 s |
Longer → cleaner field, potential CD loss from over-development |
Track Integration: Modern coat/develop tracks are fully integrated with scanners via automated wafer transfer. A cluster tool configuration (4-6 tracks per scanner) maintains high throughput (> 200 wph) while minimizing queue time and environmental exposure.
10. Tool Qualification
Installation Qualification (IQ)
IQ verifies that the scanner is installed correctly per specifications:
IQ Checklist:
- Utilities: Power (±2%), compressed air (6-8 bar), N₂ (99.999% purity), DI water, exhaust
- Floor vibration: < 50 nm RMS (< 100 Hz) under the tool footprint
- Temperature: 22 ± 0.1°C ambient, ± 0.5°C variation over 24 hours
- Humidity: 40-50% RH ± 5%
- Cleanroom class: ISO 2-3 verified by particle counter
- Safety interlocks: Laser safety, E-stop, door interlocks functional
- Communication: SECS/GEM host connection, recipe download verified
Operational Qualification (OQ)
OQ tests scanner subsystems and verifies they meet functional specifications:
- Laser Performance: Energy stability, wavelength bandwidth, pulse duration
- Illumination Uniformity: < 1% (3σ) across slit, verified with dose mapper
- Stage Performance: Velocity, acceleration, settling time, interferometer accuracy
- Alignment Precision: Repeatability on reference wafer < 1nm (3σ)
- Focus/Leveling: Accuracy vs. reference surface < 20nm (3σ)
- Image Field: Size, distortion, field curvature within specs
Performance Qualification (PQ)
PQ uses process wafers to verify lithographic performance metrics:
PQ Tests:
1. Overlay: Expose test pattern on multiple layers, measure overlay
Target: < 2nm (mean + 3σ) scanner contribution
2. CD Uniformity: Expose dense line array, measure CD at 49-81 sites
Target: < 1.5nm (3σ) intra-field, < 2.5nm wafer-level
3. Focus: Expose focus-defocus matrix, determine best focus and DOF
Target: DOF > spec (50-150nm), best focus stable < 10nm
4. Dose: Expose dose series, measure CD vs. dose sensitivity
Target: Dose control < 1% (3σ), linear response in process window
5. Throughput: Measure wafers per hour at production recipe
Target: > 90% of rated throughput (e.g., > 225 wph for 250 wph tool)
Specification Checks and PM Schedules
Ongoing maintenance ensures performance remains within specification:
| Interval |
Maintenance Activity |
| Daily |
Quick overlay check (golden wafer), laser energy calibration, particle count |
| Weekly |
Full overlay and CD qualification, focus check, baseline stability |
| Monthly |
Illumination uniformity, stage calibration, interferometer verification, aberration measurement |
| Quarterly |
Lens cleaning (if needed), alignment sensor calibration, full PQ re-qualification |
| Yearly |
Major PM: replace consumables (filters, lamps, O-rings), full re-calibration, vendor support |
Tool Uptime: Modern scanners achieve 90-95% uptime (scheduled + unscheduled). Mean Time Between Failures (MTBF) > 200 hours, Mean Time To Repair (MTTR) < 2 hours. Preventive maintenance is scheduled during low-demand shifts to maximize productive time.