ASPE Topical Meeting, Charlotte NC 3/7/2011
Mid-Spatial Frequency Matters Recent Advances & Accomplishments in Manufacturing High-precision Aspheres by
Tony Hull1 Keith Carrigan1 Adam Magruder1 Shayna Khatri1 Ankit Patel1 Brian Catanzaro2
1L-3
IOS Division Tinsley Facility Richmond, California
2CFE
Services San Diego, California
L-3 Communications ISSS Integrated Optical Systems (IOS) Division Tinsley 85 years
Brashear
SSG
130 years
34 years
Heritage nearly 250 cumulative years in precision optics
OVERVIEW MID-SPATIAL FREQUENCY (MSF) SOME CAUSES OF MSF ERROR GEOMETRY & DIFFRACTION WHY CONTROL MSF POWER SPECTRAL DENSITY CONTROL OF MSF SPECIFICATION OF MSF CONCLUSION
OVERVIEW
Spitzer: Cryogenic Be
Kepler: aspheric corrector
HST: COSTAR, WFC 2/3, COS, STIS / NICMOS, ACS
Perspective from Key/Flagship NASA Missions
JWST cryo-null off-axis Be aspheres
Perspective of diverse aspheric requirements
NIBF 0.7-3.0mm beam
Stress Mirror Polishing
1nm optical system
PSD for Tinsley 0.75m OAP (S) (in-process result 12/08)
nm^2*mm^2 = Angstroms^2*cm^2
1.00E+05 1.00E+04 1.00E+03 1.00E+02 1.00E+01 FullAp 1.00E+00
Ref (TDM)
1.00E-01 1.00E-02 1.00E-03 1.00E-04 0.001
0.01
0.1
1
Spatial Frequency (1/mm)
PSD control on OAPs
SiC systems (6 spaceborne)
NIF 3ω Wedge focus
Beryllium
Diverse technical approaches
Special Facilities Meter class precision interferometry
9 identical 1.5m CCOS polishing machines
Freeform SPDT 3 identical Leitz CMMs (1.6m x 2.4m)
MRF to 2.5 meters
Subaru 8.3m PM to 13nm RMS
What problem are we solving? • Challenging aspheric surfaces: – What makes them difficult?
• Fast f/ratios • Off-axis • Extreme lightweighting for inertial or thermal reasons • Performance at cold temperatures or in presence of thermal sources or sinks • Metrology and sampling
Examples • To be illustrative, examples of quilting and zonal effects have been selected from the literature, and in some instances, in-process data. • Illustrations showing Mid-Spatial errors should not be assumed to be inevitable. • Methods available at IOS routinely control these errors to exceptional smoothness.
MID-SPATIAL FREQUENCY (MSF)
Slope requirements & Surface Requirements RMS Surface related to Slope 10ur RMS
3ur RMS
1ur RMS
0.3ur RMS
100nr RMS
30nr RMS
10nr RMS
3nr RMS
1.0E+00 1.0E-01 Diffraction
1.0E-03 1.0E-04 1.0E-05 1.0E-06 1000
100
10 L(mm)
1
0.1
RMS(um)
1.0E-02
Limited @633nm
Si atom dia.
The first Zernike terms usually are not be sufficient to describe a surface’s MSF Character!
In-Process: 36 Zernikes Removed, but meaningful structure remains 6nm RMS
13
9/9/2008
Filter >100mm 60 cy/ap
36 nmP-V 0.5 nm RMS
MSF domain is often application specific or cause specific Frequencies • Low frequencies: about 4 cycles/aperture where the first Zernike terms fail to describe • All the way to microroughness that defines hemispherical scattering • Application specific – Planet finding ~3 to 40 cy/ap – Synchrotron grazing incidence 1mm and shorter
Root Causes • Zonal errors where aspheric departure is large • Sub-structure printthrough • Operational thermal boundary conditions being different than factory conditions
MSF Errors • Phase Control: surface error patterning over a number of cycles per aperture – Geometric Beamwalk issues in sensitive gauges and where wavefront correction is applied at a reimaged conjugate pupil – Retrace errors in interferometry – Wavelength dependent diffraction.
• For systems sensitive to diffraction, also Amplitude control (coating uniformity, optic uniformly polished out) • Both phase and amplitude lead to diffraction effects, throwing out of phase photons a few λ/D outside the core image. – Problematic for ultra contrast requirements, where a faint object is narrowly separated from a bright object. – A factor in radiometric accuracy
SOME CAUSES OF MSF ERRORS
• Pattern “quilting” print-through from a cellular lightweighted mirror • Zonal errors relating to change of surface vs. tool shape across the part • Cryo-quilting • Metrology errors or sampling mismatch • Random surface errors in frequency band But “real world” manifestation is often complex
Polishing a Lightweighted Mirror
t = FS thickness
B
= DIA inscribed circle
Removal Rate ~ P Load Tool
δ~ P B4 δ~ t3 δ~ f(E, ρ)
Optical smoothing produces print through dependent on cell size, facesheet thickness and material properties Estimate the deflection of a mirror due to polishing pressure, and the reaction of the substructure cells W v E C b t b4/t3 q Data entry v= E= C= b= t= q= min WL= Result W(p-v)= W(p-v)=
Following Pravin Mehta (3/6/87) Pressure unit translator deflection in p-v or rms depending on constant C used poisson's ratio of facesheet material natural in GPa C parameter young's modulus 0.1 psi = 6.90E-07 cell geometry parameter cells p-v rms 1 Msi = 6.90E+00 diameter of cell inscribed circle square 1.512E-02 4.480E-03 1 N/m2 = 1.00E-09 facesheet thickness triangular 1.800E-02 4.520E-03 1 g/mm2 = 9.81E-06 quilting parameter hexagonal 1.332E-02 3.620E-03 0.01 atmos = 1.01E-06 uniform pressure Reference values Be Al Zerodur ULE FS(Cor) FS(Her) Si 0.21 v= 0.08 0.33 0.24 0.176 0.167 0.067 0.28 410 GPa E= 287 69 90.6 67.7 73.2 67.7 131 1.800E-02 p-v and = 4.520E-03 rms GrE(70/x30) 0.1500 m SiC(HP) SiC(HIP) SiC(CVD) SiC(RB60) SiC(RB90) C/SiC 0.0100 m b4/t3= 506.25 m v= 0.21 6.90E-07 GPa E= 430 425 466 310 410 9.3 0.633 um 0.015 um
or
W(rms)=
0.004 um
0.023 waves or W(rms)= 0.006 waves at minimum wavelength note: W(p-v) of 0.033 waves will result in 1% loss from the core of the PSF per Wetherall
Zonal Errors ZP
ZS
AD=ZS-ZP
1.20
1.00
0.80
0.60
0.40
0.20
0.00 0
0.2
0.4
0.6
0.8
1
ESA Image
ESA Image
Cryo-Quilting and its Management • Definition – Cooling of a Mirror Substrate Produces a Pattern with Similarity to the Physical Structure/Design of the Mirror
• Exhibited in Most Materials – – – –
Beryllium ULE, Zerodur SiC CFRC
• Causes – Material Variation – Built in Stress Expressed as Modulus and Strain Changes
• Various Techniques to Address Quilting • Take-Away – Assume it Exists – Compare with Requirements – Address as Needed
Example – SiC
Note quilt matchs Small Pockets
• 1.35 m Demonstration Mirror • Measured at λ = 10.6 um B. Catanzaro “The ESA Herschel Telescope Tiger Team Metrology Review: Test Results” SPIE 7010 ,2008
Example – SiC from Polishing
• Polishing 3.5 m/RT pentaprism Metrology T. Korhonen “Polishing and testing of the 3.5 m SiC M1 mirror of the Herschel space observatory of ESA” SPIE 7102 ,2008
Cryo Quilting Examples
ULE AMSD Demo
CFRP Herschel Demo
Be SBMD Demo
SiC Herschel Demo
Spatial Frequency Analysis for Herschel
Backside of M1
Understanding the Structure • Physical Size of Cells, Support Structures • Should Relate to Metrology Sampling
Spatial Frequency Scale M1 Size
Petal Width
Cell Size
Cathedral Rib
Cryogenic Ambient
ν/2 1/4 m-1
1/m-1
1/128 mm-1 1/32 mm-1
1/8 mm-1
Petal
• Physical Structure: • Metrology Methods:
Metrology
Best Practices: Match Scale of Metrology to Physical Structure
Cell with Cathedral Ribs
Poor Distortion Mapping Distortion dominating MSF Error
PSF
793nm PV 52nm RMS
Strehl 0.912
Part type: Concave On-Axis Asphere
Distortion Mapping Corrected
PSF
115nm PV 7nm RMS
Strehl 0.990
Part type: Concave On-Axis Asphere
Distortion Effect on MSF 2D PSD 1.E+07 1.E+06 1.E+05 1.E+04
2D PSD (A2cm2)
1.E+03 1.E+02 1.E+01 1.E+00
Distortion
1.E-01
Distortion Corrected
1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 0.01
0.1
1
Spatial Frequency (1/cm)
10
GEOMETRY & DIFFRACTION
Concerns Geometry • Wavelength independent • Beam Walk • In a pupil reimaging system, MSF errors and mechanical registration stability are linked. • Gauges • Wide angle systems
Diffraction • Wavelength dependence • Phased grating effects alter the PSF and EE curves – And many MSF errors have a periodic component
• Affect near angle scatter and reduce high contrast imaging ability (exoplanets)
Caution if test wavelength and operational wavelength differ!
Terrestrial Planet Finder Inner Working Angle to Detect Planets typically between 2 and several λ/D Must mask out the core star Therefore care about pupil features between ~2 and several 10s of cycles per aperture Although DMs can address MSF errors, the larger the MSF error on the PM, the more sensitive the telescope to beamwalk pupil shear
Coronagraph Schematic M1
M2
Fold
DM
OAP1
OAP2
OCC MASK OAP3 PO Mirror & f/80 Guide CAM Long CAM Med. CAM Short CAM
LYOT MASK OAP4 FOLD
WHY CONTROL MSF?
“Specifications: Figure and Finish are not Enough”
From… Robert E. Parks, College of Optical Sciences, University of Arizona… ABSTRACT Several examples are given of optics apparently specified only by figure and finish. Although these optics met the specifications they did not produce good images. The presumed reason for the poor performance was the lack of a specification for mid-spatial frequency roughness…
Wetherell: The Calculation of Image Quality: Error =0.07518 λ rms (pp265-266, AO&OE VIII, 1980)
In the past, mirror substrate was constrained to minimize print through resulting from optical smoothing
Print-through (nm rms surface)
• Print-through is not addressed in most Zernike representations • MSF errors spread significant energy beyond the Airy core • Small cells and thick ULE face-sheets were ZERODUR needed
MSF Print-through (0.3psi smoothing) Z(t=5)
Z(t=10)
U(t=7.5)
Z(t=7.5)
U(t=5)
10
1
Cell inscribed circle B (mm)
HST
U(t=10)
HST Surface MSF Phase Map (Cohen & Hull 2004)
John Krist (JPL) simulations of Hubble Images personal communication 9/2/2009
POWER SPECTRAL DENSITY (PSD) & MSF EXAMPLES
Case: Convex 600mm sphere Filter >100mm 60 cy/ap
113 nmP-V 9.8 nm RMS
36 nmP-V 0.5 nm RMS
18 nmP-V 1.8 nm RMS
8.2 nmP-V 0.6 nm RMS
2-D PSD for MSF vs LSF Figure Error MSF 10nm
LSF 10nm
TDM Specification (ref)
0.1
1
1.E+07
PSD(A^2*cm^2)
1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01 1.E+00 1.E-01 0.01
Frequency (1 / cm)
10
Mid Frequency Error Scatters Image Intensity 1
0.9
Fraction of Enclosed Energy
0.8
0.7
0.6
10nm LSF
0.5
10nm MSF 50nm MSF
0.4
50nm LSF
0.3
0.2
0.1
0 0
5
10
15
20
25
30
35
40
Radius from Centroid in (2 r)/(λ*f/#)
45
50
Backside Print-Through Example Prior to correction
Post CCOS correction
0.036λ RMS / 0.18λ PV
0.007λ RMS / 0.05λ PV
Part type: Light-weighted Silicon carbide with silicon cladding, Flat Mirror. 1.78mm thick ribs
Backside Print-Through Before and After Correction PSDs 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03
2D PSD (A2cm2)
1.E+02 1.E+01 1.E+00 1.E-01
Prior to Correction
1.E-02
After Correction
1.E-03 1.E-04 1.E-05 1.E-06 0.001 1.E-07
0.01
0.1
1.E-08 1.E-09
Spatial Frequency (1/cm)
1
10
TDM (ref) 1 E+09
After CCOS
After Large Tool Polishing
Large Tool Polishing 1µm P-V (200nm RMS) Print through evident
1 E+08 1 E+07
PSD (A^2*cm^2)
1 E+06 1 E+05 1 E+04 1 E+03 1 E+02 1 E+01 1 E+00
CCOS Correction 60nm P-V (6nm RMS)
1 E-01 1 E-02 0.01
0.1
Frequency (1/cm)
1
10
CONTROL OF MSF
Traditional Good On-axis Mirrors
Example of Print Through removal via IOS MRF Convergence rates for the lightweight mirrors MRF processing showed the same 70% minimum convergence rates that were achieved on solid optics.
Total machine time was 24 hours in four runs. pg. 53
3/7/2011
1D PSD of Optical Metrology Instruments
Instrument
Interferometer
AFM
PMM
1.E+05 1.E+04 1.E+03 1.E+02 1.E+01
PSD [nm^2*mm]
1.E+00 1.E-01 8.7um
1.E-02
2.2um 50x
1.E-03
10x Full Ap Optical
1.E-04
Sub Ap Optical
1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 0.0001
0.001
0.01
0.1
1
10
Spatal Frequency [1/mm]
54
3/7/2011
100
1000
10000
100000
Example S: 0.75m OAP Design: Double Arch Material: Fused Silica Parent FL: 3.8 m Mirror EFL: 5.1 m Aspheric departure > 2000µm • Turning angle = 60° • • • • •
60°
Reference for OAPs: TDM for TPF Cohen & Hull: Selection of a Mirror Technology for the 1.8m Terrestrial Planet Finder Demonstrator Mission, SPIE 5494, 2004
The Terrestrial Planet Finder (TPF) project realized that to demonstrate that a coronagraphic telescope concept could be used for terrestrial planet detection there needs to be a demonstration that a mirror of the required technology could be built. Most important
surface quality of the mirror over the spatial frequency range ~ 10 cm to 4 m. A ripple in the surface of the mirror, with a spatial scale in this range, would cause starlight to diffract onto the region where a planet may be located. In terms of an rms surface error the mirror would need to be better than 5 nm rms in this range
PSD (S) better than Critical Reference Standard PSD for Tinsley 0.75m OAP (S) (in-process result 12/08)
nm^2*mm^2 = Angstroms^2*cm^2
1.00E+05 1.00E+04 1.00E+03 1.00E+02 1.00E+01 FullAp 1.00E+00
Ref (TDM)
1.00E-01 1.00E-02 1.00E-03 1.00E-04 0.001
0.01
0.1
Spatial Frequency (1/mm)
1
Clear Aperture 690mm Diameter, Average of Measurement 1 & 2, 11-21-08
Vertex Side
Wavefront Error Normal incidence
Gravity
2008 vs 2009 Technical Performance 2008 vs 2009 Performance 1.E+05 1.E+04 1.E+03 1.E+02
PSD [nm^2*mm]
1.E+01 1.E+00 2008 Laser Fusion OAP 1.E-01
2009 Laser Fusion OAP
1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 0.0001
0.001
0.01
0.1
1
10
Spatial Frequency [1/mm]
3/7/2011
59
How do you characterize the full PSD? Overall RMS surface error < 0.30nm RMS! + 3 nm
M1 - 3 nm
1.00E+02
Coated M1
1.00E+01
PSD (nm^2*mm)
1.00E+00
M2
1.00E-01 1.00E-02 1.00E-03 1.00E-04 1.00E-05
1D Power Spectral Density
1.00E-06 1.00E-07 1.00E-08 0.01
0.1
1
10
100
1000
spatial frequency (1/mm)
10000
100000
1000000
MSF SPECIFICATION
Specifications to optical manufacturers • Often the true systems requirement can be expressed as a PSD • Most optics houses, even the most sophisticated, will have trouble with this. Questions of padding, orthogonality etc. enter. Buyoff can be ambiguous. But this is beginning to change! • Specifying structure function is another approach • For many optics, desired results are most easily procured by defining the measurement method and overlap and sampling for each surface scale length
CONCLUSION
MID-SPATIAL FREQUENCY (MSF) SOME CAUSES OF MSF ERROR GEOMETRY & DIFFRACTION WHY CONTROL MSF POWER SPECTRAL DENSITY CONTROL OF MSF SPECIFICATION OF MSF
• Often MSF requirements are underspecified • When it comes to high performing systems, especially with aspheric requirements, MSF usually Matters! Patterns are of special concern. • MSF errors depend on – – – – –
Aspheric departure Extent and nature of lightweighting Nature of optical finishing Spatial and wavelength sampling of the surface Operational thermal environment
• So be careful with extrapolations • Methods are available to address these errors and provide exceptionally smooth PSD curves
“Specifications: Figure and Finish are not enough”
From… Robert E. Parks, College of Optical Sciences, University of Arizona… ABSTRACT Several examples are given of optics apparently specified only by figure and finish. Although these optics met the specifications they did not produce good images. The presumed reason for the poor performance was the lack of a specification for mid-spatial frequency roughness…
Quiz: Can you explain reflection?
Contact…
Brian Catanzaro CFE Services San Diego, CA 92109 858 204 6299
[email protected]
Tony Hull Manager of Business Development Astronomy and Space L-3 Integrated Optical Systems (Tinsley, Brashear, SSG) 510 672 2499 cell 505 771 8566 land
[email protected]
RESERVE SLIDES & NOTES
69
