All Reflective Fly s Eye Illuminators for EUV Lithography Blake Crowther, Donald Koch, Joseph Kunick, James McGuire Optical Research Associates Robert Harned, Rick Gontin ASML Presented by Kevin Thompson/ Optical Research Associates 1 st International EUVL Symposium October 15-17, 2002 Dallas, Texas 1
Outline Reflective Fly s Eye Concept First Generation EUV Condenser: ETS Next Generation EUV Condensers: Reflective Fly s Eye Some Simulation Methods for Beta Generation Reflective Fly s Eye Optimization Conclusions 2
Background Beta Generation EUV Illuminators Fly s Eye Technology looks to be the leading candidate for Beta EUVL Field Array Reflective Fly s Eye Homogenizer Coupler System Mask Integrator/Relay Elements Pupil Array Second Focus & Source Image Source Source & Collector Collector 3
Fly s Eye Radiometrically efficient Provide uniform irradiance over a range of source parameters 4
Transmissive Fly s Eye Concept Source Image Plane Collector Field Array (FE1) Pupil Array (FE2) Integration Lens Functional description Field array imaged to image plane Overlapped Source imaged to pupil array Discrete pupil 5
Reflective Fly s Eye Parent Surfaces FE2 Reference Surface Imaging Facets FE1 Reference Surface Imaged Facets Source or Source Image Source or Source Image Overlapping Vertex Location Overlapped Facet Image Form Here Beginning point for a fly s eye design Paraboloid example reference surfaces Free of spherical aberration and coma Central ray paths for typical facets shown 6
Simplified Reflective Fly s Eye FE2 2 FE1 1 Source or Source Source or Source Image Image 7 Overlap Area Overlap Area Matched pair of facets Vertex locations correspond to central ray intersections with reference surfaces Vertex tilts match that of local reference surface Facet Imaging FE1 facets image source at corresponding facet on FE2 FE2 facets image corresponding FE1 facets to image plane Integration occurs by summing irradiance across facets at overlap
Fly s Eye Elements and Parent Surfaces Superimposed FE2 2 1 FE1 Source or Source Image Overlap Area Overlap Area FE1 facets can have shape of desired target area (e.g. arcuate) FE2 relayed into imager pupil 8
Top Level Design Parameters Radiance distribution of the source Tolerance to changes in source radiance distribution Illumination specifications Target area cross-scan uniformity and in-scan shape Numerical Aperture (e.g. 0.25) Partial coherence factor (V) Pupil fill Uniformity Polarization (angles of incidence) Telecentricity 9
ETS Condenser (Sweatt 1994) Köhler (azimuthal) & critical (radial) illumination Compound elliptical concentrator collector mirror segments Roof mirror pairs Toroidal field mirror 10
ETS Pupil Fill Early Prediction Six discrete segments Asymmetric pupil fill H/V bias 11
ETS Reticle Irradiance Very Early Prediction 12
Representative Beta Reflective Fly s Eye Condenser COLLECTOR SOURCE Pupil Fill: Images of Source on F2 facets SECOND FACETED MIRROR, FE2 intermediate Focus OB2 M1 M2 M4 Scan Direction WAFER M6 FIRST FACETED MIRROR, F1 OB1 M5 M3 RETICLE 13
Monte Carlo Methods for Pupil Uniformity Must be Accelerated Performance predictions for 2nd generation illuminators require 1% accuracies, this makes conventional Monte Carlo methods schedule drivers Trace from source w/o polarization 400,000,000 rays in 4 days, numerical noise 0.4% With polarization, this computation is estimated to require 10 days Pupil Estimate using Forward Monte Carlo Ray Trace 150.00 MM Asy NIC, 47 mm shaper, w/nurbs relative irradiance 1.0000 0.5000 150.00 MM Total flux 0.26061E-09 Watts Max irradiance 0.26815E-11 Watts/CM^2 Min irradiance 0.00000E+00 Watts/CM^2 0.0000 14
Reticle Uniformity Predictions Based on Monte Carlo Methods also Drive Schedule Forward Monte Carlo Analysis Gaussian Model 7.0000 MM Point Source 0.0000 0.0000 120,000,000 rays in 12 hours. 28.000 MM 0.0000 Plasma Model 60,000,000 rays in 44 hours >6X increase in trace time when source complexities are modeled 7.0000 MM 28 000 MM.00002.00001 0.0000 Plasma Source 15
Simulation Methods to Enable EUVL Illumination System Optimization Complex source modeling Fly s eye surface modeling Discontinuous merit function evaluation Smart sampling Noise tolerant optimization 16
Complex Volumetric Source Models can be Developed Spatial Luminance Measured Data Spatial Luminance Using Inversion Technique Using an HID lamp as a test case, a match between the measured data and inversion based modeling has been demonstrated The accuracy of the simulations of the source are currently limited by the data for a potential source 17
NonUniform Rational B-Spline (NURBS) Model Reflective Fly s Eyes In EUV lithography, Normal incidence reflectance is ~70% (vs. 99.5% for DUV) Must minimize the number of surfaces (4 or 5 typical). Requires non-conventional surfaces to accomplish the same function as multiple spherical surfaces. Computer-aided modeling allows non-conventional surfaces NURBS surface shape often requires 2D grid 10,000 points is typical 10,000 variables is not practical for optimization. To accommodate optimization, New hybrid surfaces Small number of control parameters Allows designer to focus on the design problem, not programming the surface. 18
Textured Surface for Fly s Eyes In imaging systems, typical use is to pin the vertex location and allow the sag at the intersection diameter to vary Maintain paraxial properties In illumination systems, it is more common to pin the intersection diameter and allow the surface-tovertex to vary Maintain collection efficiency With the spherical textured surface, changing the intersection diameter will automatically adjust the surface-to-vertex 3 lenslets. Same collection efficiency but widely different curvatures 19
Hybrid Surfaces for Reflective Fly s Eye NURBS + Polynomials For optimization with a practical variable set, combine a base NURBS with Zernike coefficients Much like the approach of adding asphere coefficients to a base conic surface Large reduction in number of required variables, yet still model highly complex surfaces. Z2 Z3 Z4 Z5 Z6 Z7 Z8 Z9 Z10 Z11 Z12 Z13 Z14 Z15 Z16 Z17 Z18 Z19 Z20 Z21 Z22 Z23 Z24 Z25 Z26 Z27 Z28 20
Rapid Estimator for Pupil Uniformity Discontinuities and dead zones are a challenge for an optimization merit function constructor New methods have been constructed for EUVL applications Pupil Non-Uniformity (Radiance 1 facet, 1 field) Rel. Rad. 1.0000 X= 0.000 X= 0.000 Y= 0.000 0.5000 0.055 fe031399_rev_c.len 22-Apr-2000 Scan -35 mm; Total irradiance: 0.0006 RMS/Avg=2.818; Cent(mr): x= -0.71 y= 0.00 0.0000 11:37:00 Y= 0.000 0.055 FE031399: Reverse Wa fer to Source PUPIL BOUNDARY PROJECTED ANGLE (OBJ. SPACE) ORA 22-Apr-2000 Limiting Surface 18 20 stop 10:54:29 FE031399: Reverse Wa fer to Source PUPIL BOUNDARY PROJECTED ANGLE (OBJ. SPACE) ORA 22-Apr-2000 Limiting Surface 18 20 stop 21
Convolution: Smart Smoothing Point Source Response Pinhole Camera Response Computed Convolution Trace Results Number Of Rays 2,000 => => 20,000 Point source results can be used to help predict extended source results Convolution of two low accuracy results can provide >100X reduction in the number of required rays 22 200,000 2,000,000
Smart Sampling is Effective for EUVL as a Merit Function Accelerator Random Smart 50,000 rays 300,000 rays Spread @ 50,000 rays is similar to random @ 300,000 rays 23
Optimization Interface for Fly s Eye Components (Prototype Environment) 24
Fly s Eye (FE1) Predicted Irradiance Distribution FE 031299: 0.6 Sig, Mapped Obs irradiance.00010 600.00 MM.00005 600.00 MM Total flux 0.12061E+00 Watts Max irradiance 0.96121E-04 Watts/CM^2 Min irradiance 0.00000E+00 Watts/CM^2 0.0000 25
.08400 dc Fly s Eye (FE2) Predicted Pupil Fill Pupil Uniformity Lens: fe031399_effsource_c 0.6 sigma Irradiance 17119. 8560.5 Symmetric Formed of many discrete points Approximates uniform irradiance distribution.08400 dc FE031399: Reticle to Source Field (mm): xob= 0.000 yob= 0.000 Centroid (mr): x= 0.10 y= 0.00 2.0000 26
Fly s Eye Relative Reticle Irradiance Modeled Data 27
Conclusions Fly s eye designs have emerged as the baseline technology Adequate performance predicted Subsystems have been prototyped successfully Fly s eye modeling, simulation, and optimization for EUVL is a rapidly developing field as EUVL illuminator design continues to be a schedule driver 28