Ion Beam Figuring precision optics for synchrotron radiation sources L. PEVERINI, J. J. FERME & C. du JEU

Size: px

Start display at page:

Download "Ion Beam Figuring precision optics for synchrotron radiation sources L. PEVERINI, J. J. FERME & C. du JEU"

Rudolph Chandler
5 years ago
Views:

1 Ion Beam Figuring precision optics for synchrotron radiation sources L. PEVERINI, J. J. FERME & C. du JEU Thales SESO S.A.S., Aix en Provence, 13593, France

2 2 / Ion beam profiling (IBP) approaches for SR optics IBP schemes for 1D grazing incidence X-ray mirror optics (Silicon, SiO2 and Zerodur) ion beam d(x) v=v(x) d(x) Removal of material by sputtering v=v(x) v=v1(x) d(x) v=v2(x) Off-set scan Sample/spot moving Slow process High resolution x Single Blade Scan Sample/slit moving Slow process Interesting for coating Tight slope error specification along the longitudinal direction (< 5 nrad) Non-contact polishing method possibility to use in-situ metrology Gap Scan Sample Stationary very fast L. Peverini et al., Ion beam profiling of aspherical X-ray mirrors, NIM A 616 (21) (technology) Requires Broad Beams V.I.T.A. de Rooij-Lohmann et al., Roughness evolution of Si surfaces upon Ar ion erosion, Applied Surface Science 256 (21) (in-situ) x x

3 3 / Fabrication and testing Clean room environnent Phase Shifting Interferometry (PSI) R= ZYGO R TESTING Layout (HFM) objective flat/spherical Reference Phase shifting interferometry 1 shot measurements Reference system Grazing geometry (angle q~4-2 deg) simulate final SR use Gravity does not play a role (difficult with LTP instruments laser deflectometry very popular in the SR community) Double-pass reduces reference errors Useful technique in phase of integration adjustement of deformable mirrors TOP view Source IBF Typical processing scheme UHV 1-6 mbar 1-7 mbar L = 15 mm Ion beam Substrate scanning - offset scan miroir Direction scan X Mirror prepared by traditional polishing L = +/- 17 mm Z

4 4 / IBP convergence Convergence rely on: 1) Flux stability during scanning and several days (especially when using an iterative approach) 2) Removal rate determination accuracy 3) Reference system used for both metrology/ibp 4) Accurate metrology data manipulation to extract pure surface data (image and signal processing) 5) Possibility to account for systematic errors (fabrication+metrology) 6) Well defined operational sequences and well trained staff 2 15 P-P1 Surface Height [nm] Removal Function (theory) Removal Function (experimental) Profiling error Mirror coordinates [mm]

5 / µ-roughness and IBP µ-roughness preservation prooved for Removed thickness ranging 2 3 nm Regardeless

69 Before IBP After IBP 3.62 s =.11 nm s =.8 nm s =.27 nm s =.24 nm 1µm 1µm 1µm 1µm 7. 8.

5-3 PSD[log(nm 4 )] 6 5 4 3 2 1-1 -6 Roughness preservation for initially rougher substrates -5.5-5 -4.

5-3 k[log({1/nm})] k[log({1/nm})] Physics of ions intercation is more critical @ AFM spatial scales (p<1

5 5 / µ-roughness and IBP µ-roughness preservation prooved for Removed thickness ranging 2 3 nm Regardeless of initial morphology and roughness amplitude (periods 1-1 µm) Before IBP After IBP 2.69 Before IBP After IBP 3.62 s =.11 nm s =.8 nm s =.27 nm s =.24 nm 1µm 1µm 1µm 1µm PSD[log(nm 4 )] Smoothing observed for initially flat surfaces PSD[log(nm 4 )] Roughness preservation for initially rougher substrates k[log({1/nm})] k[log({1/nm})] Physics of ions intercation is more AFM spatial scales (p<1 µm) ion energy, angles etc. AFM studies E. Ziegler et al., Evolution of the surface finish of an X-ray mirror exposed to a low-energy ion beam, NIM A (21) F. Frost, B. Ziberi, A. Schindler, B. Rauschenbach, Appl. Phys. A 91 (28) 551 (nanopatterning)

6 / PSI data traitement for deterministic polishing the 1st requirement is the data repeatability: Turbulence, thermal gradients, CCD noise & laser coherence surface

6 6 / PSI data traitement for deterministic polishing the 1st requirement is the data repeatability: Turbulence, thermal gradients, CCD noise & laser coherence surface data are mixed to all these artifacts with different spatial/temporal dependencies Air bubbles and thermal gradients Data are rejected if this condition is not achieved

7 7 / TSESO metrology capabilities repeatability (difference between 2 consecutive measurements) typical 25-7 nrad Processing and values data confirmed by LTP customer measurements (next slides) Some mirrors have complex parabolic shape and multistripe design heavy image/signal processing requiring customized scripts adapted to each specific situation 1 m Automation and standardisation is necessary for smooth industry productivity but require a big effort and highly specialized staff accuracy (difference between 2 statistically independent acquisitions) typically 6-15 nrad High Frequency Noise 5 1 nrad

8 8 / Metrology validation for slope errors ~.5-.6 µrad slope(ltp)-slope(psi)<5 nrad (stdev) LTP

9 PSI/LTP data high frequency 9 / Metrology validation PSI vs LTP for slope error <.3 µrad LTP Excellent shape agreement stdev(ltp-psi)~.5 nm!!! Before final correction LTP High Frequency introduces a factor ~2 in the slope error estimation. 1)Correlated signal in LTP data (partial laser cohérence) 2)Shape obtained from LTP is smoothened naturally by the integral operator 3)PSI need to be filtered for slope error evaluation therefore part of the surface signal might be suppressed

1 / HF noise (speckles) a) Speckles are independent on the mirror size and the surface microtopography (in a morphological sense) b) The pattern observed is characterized by a «round» shapes

average environmental artifacts) e) 1D LTP data from 2 different customers presents very similar features «Cleaned» speckle pattern obtained using a lateral shift dx>x (speckles corr.

10 1 / HF noise (speckles) a) Speckles are independent on the mirror size and the surface microtopography (in a morphological sense) b) The pattern observed is characterized by a «round» shapes regardeless of the surface morphology and PSI geometry c) The speckle pattern is static d) The recording method can enhance this effect (especially using large statistical sampling necessary to average environmental artifacts) e) 1D LTP data from 2 different customers presents very similar features «Cleaned» speckle pattern obtained using a lateral shift dx>x (speckles corr. length) and low frequency FFT filtering Refs R. Cerbino, L. Peverini, M. A. C. Potenza, A. Robert, P. Bo secke, and M. Giglio, Nature Phys. 4, 238 (28) S.Berujon, E. Ziegler, R. Cerbino, and L. Peverini1,*PRL 18, (212)

11 11 / 1st APPLICATION short mirrors for nanofocusing Variable range of curvature t = 5 mm R = 2-8 m Specification Shape errors <.75 nm-rms Slope errors r.m.s < 5 nrad TORPEDO (parabolic) width profiles for elliptical shaping (variable inertia)

12 / Mirror length L=13 mm surface height [nm] 1-1 - 2-3 - 4 IBF corrections - shape errors High accuracy data (near-normal incidence PSI) Spatial resolution.24 mm (LTP FWHM~ 2.

12 12 / Mirror length L=13 mm surface height [nm] IBF corrections - shape errors High accuracy data (near-normal incidence PSI) Spatial resolution.24 mm (LTP FWHM~ 2.5 mm spot) PTV before = nm PTVafter2 after = 2.54 nm nm PTVbefore nm Shape r.m.s =.5 nm (specification.75 nm) before IBP (stripe 1) after IBP (stripe 1) mirror coordinate [mm] 6 Slope 4 2 Sl Sl Before correction stdev ( S l2) µrad After correction stdev ( S l1) µrad x1 1 x2 1 x3 1 x4 1 slope error r.m.s (before IBP)= 18 nrad slope error (after IBP)= 298 nrad

13 13 / 2 nd application: flat mirror 1.9 m long 1.9 m Specification: Slope errors < 3 nrad (L=1 m) Radius > 4 km Determination of optimal number of images to be averaged Theta=4 deg Environmental stabilisation The choice of the metrology method in order to extract the real surface : Pt Rh 3 mm 1)Three flat approach ( surface profile over 1 line only) 2) Replacement of reference optics with different statistical errors require a large number of optics to improve the statistics) 3) Mapping of the optical cavity - full surface available but time consuming and heavy when several iterations are needed critical environment stabilisation 15 mm

14 14 / Statistical methods 4 Bande - T OP - C - P1-P2-P3 4 Bande - T OP - C+ - P1-P2-P3 Ias 1 9 Ibs 1 9 Ics 1 9 Ias Ibs Ics IM Ids 1 9 Ie s 1 9 Ifs Ids Ie s Ifs x1 Slope x1 1 Slope Sl Sl Sl stdev ( Sl1) x1 1 stdev ( S l2) stdev ( S l3) Sl Sl Sl x1 1 stdev ( Sl4) stdev ( S l5) stdev ( S l6) The surface is probed using 3x3 statistically independent configurations Tilt and local radius fluctuation due to environment Slope errors seems to be more stable because are insensitive to tilt fluctuation of the wavefront Igs 1 9 Ih s 1 9 Ii s Igs Ih s Ii s Sl Sl Sl Bande - T OP C- ; P1-P2-P x1 1 Slope Strong height fluctuation observed on the edges of the substrate The effect is averaged out using statistics x1 1 stdev ( Sl4) stdev ( S l5) stdev ( S l6)

15 15 / Analysis on the usefull lenght 1m 15 Profiles IBF corrections - shape errors surface height [nm] mirror coordinate [mm] Stripe 1 Stripe 2 PTVbefore PTVafter PTVbefore PTVafter nm nm nm nm before IBP (stripe 1) after IBP (stripe 1) before (stripe 2) after (stripe 2) Sl Sl SLOPES Slope Stripe 1 Before correction stdev ( S l1) stdev ( S l3) µrad µrad Sl Sl x1 1 x2 1 x3 1 x4 1 Stripe 2 After correction stdev ( S l2) stdev ( S l4) µrad µrad

16 16 / Sliding 2 1 Windows mirror length.1 m 1 m Mirror length = 1 m Sliding windows -Mirror length=1 m Dg x1g Sliding windows -Mirror length=1 mm Mirror length =.1 m Dg height statistique rms [nm] PTV ~ 4 nm PTV ~ 1.5 nm height statistique rms [nm] Stdev ~ 1 nm sigma rms (sliding windows W=2 mm) sigma PV (sliding windows W=2 mm) 1 nm 4 nm window position [mm] Sliding windows Stdev ~.35 nm sigma rms (sliding windows W=2 mm) sigma PV (sliding windows W=2 mm) 4 nm.35 nm window position [mm] Sliding windows Scanning length slope statistique [µrad] slope statistique [µrad] slope rms (sliding windows W=2 mm) slope PV (sliding windows W=2 mm) Sliding windows of 2 mm each pixel window position [mm] slope rms (sliding windows W=2 mm) slope PV (sliding windows W=2 mm) window position [mm]

17 17 / Sliding Windows & shape error scaling Shape error scaling; L=1 m 2 3 Shape error scaling; L=.1 m 15 surface statistique [nm] PTV 1 sigma 5 <all windows> surface statistique [nm] 2 PTV sigma 1 <all windows> W windows size [mm] <PTV/stdev> W windows size [mm] <PTV/stdev> surface statistique [nm] 5 PTV sigma 4 deviation connected with sampling surface statistique [nm] 4.5 PTV sigma W windows size [mm] The ratio PTV/stdev scale linearly regardeless of the mirror size W windows size [mm]

18 18 / Conclusion a) The use of ion beam is prooved to preserve roughness and improve the surface errors in a deterministic fashion (slope errors <3 nrad, L=1 m) b) ion beam technology is today routinely used in our SR mirror production lines c) High frequency errors (speckles) need to estabilish the possibility to suppress this signal from PSI and LTP data to overcome the limit of PSI metrology d) Ultimate figuring performances are today metrology limited e) IBP requires tight operational procedures and well trained staff f) Traditional and deterministic ion polishing remain complementary techniques that needs to be matched g) Increase of difficulty between short and long mirrors is mainly linked to surface metrology

19 19 / Acknowledgements Agence Nationale de la Recherche (ANR) and the AXOC team for support and the R&D on nanofocusing optics The metrology labs in BESSY, SOLEIL & ESRF for useful discussion and sharing of LTP data R. Cerbino (University of Milan) for usefull discussion on speckles metrology and image processing The X-ray team and staff Thales SESO Eric Ziegler (ESRF) and Igor Kozhevnikov (Institute of Crystallography, Russia)

Focusing X-ray beams below 50 nm using bent multilayers. O. Hignette Optics group. European Synchrotron Radiation Facility (FRANCE) Outline

Focusing X-ray beams below 50 nm using bent multilayers O. Hignette Optics group European Synchrotron Radiation Facility (FRANCE) Outline Graded multilayers resolution limits 40 nanometers focusing Fabrication