LEAPS-INNOV D3.5 - Systematic study of performance of different deterministic processing methods for X-ray reflective freeform optics

Deliverable no. D3.5 Page 1 of 41 Project information Project full title LEAPS pilot to foster open innovation for accelerator-based light sources in Europe Project acronym LEAPS-INNOV Grant agreement no. 101004728 Instrument Research and Innovation Action (RIA) Duration 01/04/2021 – 31/03/2025 Website https://www.leaps-innov.eu/ Deliverable information Deliverable no. 3.5 Deliverable title Systematic study of performance of different deterministic processing methods for X-ray reflective freeform optics Deliverable responsible ESRF Related WorkPackage/Task WP3 / Tasks 3.2 and 3.3 Type (e.g. Report; other) Report Author(s) S.G. Alcock, M.B. da Silva, I-T. Nistea, H. Wang, R. Shurvinton, P. Pradhan, K. Sawhney, R. Barrett, P. Bras, Ch. Morawe, F. Perrin, A. Vivo Dissemination level Public Document Version 3.1 Date 31 March 2025 Download page https://www.leaps-superflat.eu Document information Version no. Date Author(s) Comment 1.0 13/03/25 S.G. Alcock & Murilo Bazan da Silva 1.1 19/03/25 A. Vivo 1.2 20/03/25 R. Barrett 1.3 20/03/025 R. Shurvinton 2.0 26/03/25 R. Barrett 2.1 28/03/25 R. Barrett 3.0 30/03/25 S.G. Alcock 3.1 31/03/25 R. Barrett Final version Deliverable no. D3.5 Page 2 of 41 Abstract “SUPERFLAT” is Work Package 3 (WP3) of the LEAPS-INNOV project, which draws together ten LEAPS facilities1 with common needs for high-performance X-ray mirror and grating substrates. The aim of WP3 is to push European industry to the forefront in the manufacture of reflective X-ray optics. It focuses on three development activities incorporating the European optics industry: develop pilot processes for industrial production of moderate length, flat, X-ray mirrors; explore the basic limits of correction technologies for X-ray mirrors; and investigate novel metrology methods which are suitable for industrial environments. Deliverable 3.5 (which spans activities developed within Tasks 3.2 and 3.3 of WP3) reports on a series of experimental studies to utilise high-quality optical metrology techniques to investigate currently available deterministic processing methods and validate their potential to produce complex, freeform, surface topographies for X-ray optics. Prior to applying deterministic correction techniques to the Xray mirrors, a “round-robin” metrology inter-comparison was invaluable to determine and crosscalibrate the performance of various microand Fizeau-interferometers at optical labs at synchrotron facilities and manufacturers. Super-polished substrates, with micro-roughness < 50 picometers rms, were measured by several instruments. After careful calibration and averaging to remove sources of random and systematic errors, the results were in excellent agreement. Fiducial markers enabled the same regions of the optical surface to be interrogated by different instruments to an uncertainty < 10 µm. To test the limits of deterministic correction techniques, ion beam figuring and differential deposition were used to create a range of challenging surface features, including a spatially-varying chirp and linear addition or removal of material. The dimensions of such features were specifically chosen to extend beyond the typical requirements for X-ray mirrors, and to observe when errors induced by the surface processing techniques start to become significant. Optical metrology after deterministic correction found several interesting observations, including an accumulation of surface point defects with increasing removal / addition of material. For IBF the minimum correction period was successfully reduced to ~ 0.75 mm; and control of the polishing tool on the optical surface to a precision < 100 µm was demonstrated. For the differential deposition process the minimum correction period was evaluated as ~ 2 mm. Overall, this study pushed the measurement limits of micro-roughness to sub-50 pm with enhanced levels of reproducibility. It also demonstrated that ion beam figuring and differential deposition are suitable for creation of next-generation X-ray optics with freeform surface profiles for synchrotron and free electron laser facilities. Both techniques as developed could be readily adapted to an industrial production context. 1 Further information about Superflat consortium members is available at www.leaps-superflat.eu Deliverable no. D3.5 Page 3 of 41 Contents Project information ................................................................................................................................. 1 Deliverable information .......................................................................................................................... 1 Document information ............................................................................................................................ 1 Abstract ................................................................................................................................................... 2 Contents .................................................................................................................................................. 3 Glossary ................................................................................................................................................... 4 1 Introduction ..................................................................................................................................... 5 1.1 SUPERFLAT ............................................................................................................................. 5 1.2 Scope of Deliverable 3.5 ........................................................................................................ 6 1.3 Reproducibility of measurement ........................................................................................... 6 2 Optical metrology prior to deterministic processing ...................................................................... 7 2.1 Optics ..................................................................................................................................... 7 2.2 Optical metrology instruments .............................................................................................. 8 2.3 Fiducial alignment .................................................................................................................. 9 2.4 Micro-interferometry ........................................................................................................... 10 2.4.1 Micro-roughness .............................................................................................................. 10 2.4.2 Micro-interferometry: noise levels .................................................................................. 14 2.4.3 Micro-interferometry: reference subtraction ................................................................. 14 2.5 Fizeau interferometry .......................................................................................................... 15 2.6 Comparison between stitching micro-interferometry and Fizeau interferometry ............. 17 3 Deterministic correction of X-ray optics ....................................................................................... 18 3.1 Overview .............................................................................................................................. 18 3.2 Technical requirements ....................................................................................................... 19 3.3 Ion Beam Figuring (IBF) ........................................................................................................ 20 3.4 Differential Deposition (DD) ................................................................................................ 22 3.5 Magnetorheological finishing (MRF) .................................................................................... 23 3.6 Fluid jet polishing (FJP)......................................................................................................... 23 3.7 Elastic Emission Machining (EEM) ........................................................................................ 23 3.8 Plasma Chemical Vaporisation Machining (PCVM) .............................................................. 24 3.9 Scope of the deterministic correction tests ......................................................................... 24 Deliverable no. D3.5 Page 4 of 41 4 Optical metrology after deterministic processing ......................................................................... 25 4.1 IBF at Diamond: “wedge” removal ...................................................................................... 27 4.2 IBF at Diamond: spatial chirp ............................................................................................... 29 4.3 Differential Deposition at ESRF: “roof-top” deposition ....................................................... 31 4.4 Differential Deposition at ESRF: spatial chirp ...................................................................... 35 5 Conclusions and perspectives ....................................................................................................... 40 Glossary Acronym Definition LEAPS League of Accelerator based photon sources PCP Pre-commercial Procurement PSI Phase shifting interferometry (monochromatic illumination) MooNpics JRA Joint Research Activity - Horizon 2020: CALIPSOplus collaboration (grant agreement No. 730872 PyLOSt Python Large Optic Stitching Software: data processing and data analysis tool for surface metrology data. SUT Surface Under Test ROI Region of interest (on the sample’s surface) Deliverable no. D3.5 Page 5 of 41 1 Introduction 1.1 SUPERFLAT The quality of X-ray optics remains a limiting performance factor for many beamlines. This problem will be more prevalent for new and upgraded Free Electron Laser (FEL) and ultra-low emittance storage-ring sources. The challenge is to develop optical systems capable of providing X-ray beam characteristics to suit the experimental needs, whilst preserving the source brilliance and photon beam quality. This enables smaller, brighter, and more coherent photon beams to be utilised by a wide range of user communities for applications include nano-focussing, imaging, spectroscopy, and coherent scattering. However, there is a severe problem caused by the limited number of European suppliers who can routinely create such high-quality optical components. LEAPS envisions to overcome these difficulties by establishing strategic partnerships with appropriate industrialists. The “SUPERFLAT” work package (WP3) aims to improve European optical manufacturing capabilities to produce state-of-the-art quality substrates for X-ray mirrors, gratings, and multilayers for synchrotron and FEL beamlines. As an outcome of this project, it is hoped that LEAPS members (and facilities worldwide) will be able to more readily purchase high-performance reflective X-ray optics from European suppliers. This would eliminate the current monopoly situation of a single supplier in Japan and help mitigate the risk of supply-chain issues via a more efficient production cycle. A major component of WP3 is investigating deterministic finishing technologies and associated metrology methods and protocols. LEAPS members are using state-of-the-art metrology instruments, which will be further improved throughout this process, and draw on technologies and protocols developed independently and via the earlier “Horizon 2020” project CALIPSOplus. Such methods need to be suitable for implementation and replication in Industrial environments. WP3 includes three development tasks: • Task 3.1: PCP action to develop pilot processes for industrial production of moderate length, flat, Xray mirrors with figure errors < 1 nm rms and slope errors < 50 nrad rms, whilst retaining surface micro-roughness of ~ 0.1 nm rms. • Task 3.2: Exploration of basic limits of figure correction technologies for X-ray mirrors for more complex optical figures. • Task 3.3: Development of new metrology methods suitable for implementation in industrial environments, applicable to complex optical figures, including developing a universal data format for describing the measured surface topography of X-ray reflective optics. Deliverable no. D3.5 Page 6 of 41 1.2 Scope of Deliverable 3.5 Deliverable 3.5 (which relates to Tasks 3.2 and 3.3) reports on a series of experimental studies to utilise high-quality optical metrology techniques to characterise currently available deterministic processing methods (commercial and in-house developments), and validate their potential to produce complex, freeform, surface topographies. This complements the commercial production of flat optical surfaces, which forms the basis of the PCP for Task 3.1. 1.3 Reproducibility of measurement Accuracy of measurement, a term often misused, even within the scientific community, is defined as closeness to the “true value”. However, in this study, it is difficult to determine accuracy since it requires test artefacts which have been measured using instruments with a rigorous calibration, traceable to primary standards. Such artefacts appropriate to the requirements of this study are not available. As a result, we have concentrated on the measurement of “reproducibility” as the appropriate metric for our study, defined as the level of agreement when different instruments measure the same property of a shared object. Due to a lack of an appropriate standard artefact, this is a common approach used in the field of surface metrology of X-ray optics. Since the rectangular field of view of a micro-interferometer typically ranges in width from a few millimetres to ~ 100 µm, only a very small fraction (< 0.1%) of the total surface area of a mirror is typically measured. Traditionally, micro-roughness is randomly sampled at a grid of locations along the length and width of the optic. This enables representative statistical properties of the clear aperture to be derived, including identification of trends in topographical defects caused by processing. However, to enable a more complete comparison between deterministic correction techniques, we extended this statistical approach to investigate if various measuring instruments can reliably interrogate the same “absolute” regions of interest on the optical surface. This would allow direct characterisation of the instrument-to-instrument reproducibility and provide an absolute comparison of the effects of deterministic processing. A second challenge is whether careful calibration of each instrument, and averaging of data, can sufficiently reduce sources of systematic and random measurement errors to reproducibly characterise optics with micro-roughness << 100 picometres rms. A final consideration is how to compare the output from instruments with different measurement parameters, including number of CCD pixels, field of view, and numerical aperture of interferometric objectives. These aspects were all critical steps towards measuring the surfaces with sufficient fidelity to draw conclusions upon the impact of the different deterministic processing techniques. Deliverable no. D3.5 Page 7 of 41 2 Optical metrology prior to deterministic processing 2.1 Optics 16 “super-smooth” (specified micro-roughness < 0.1 nm rms) silicon test samples (procured from Coastline Optics, USA) were provided to the project by ESRF. Each sample is a Si(100) single-crystal substrate with the following dimensions: length 50 mm x width 25 mm x thickness 7 mm (see Figure 1). These samples were chosen since their surface micro-roughness is representative of the requirements for the highest quality X-ray mirror substrates and will be very sensitive to any degradation of the roughness induced by the different processing methods under investigation. Fiducial markers were added to the optical surface (outside the clear aperture) of each sample using a laser writer at the ESRF (see Figure 2). Rotational asymmetry of fiducial cross positions uniquely defines End “A” and the name of each lane. As described below, the fiducial markers (Figure 3) were of great benefit to accurately align the regions of interest (ROI) for measurement and deterministic processing. Figure 1: Subset of the 16 super-smooth silicon test samples used for this project. Each sample was stored and shipped in an individual, protective 2” wafer holder box. Deliverable no. D3.5 Page 8 of 41 Figure 2: Diagram of the optical surface of each super-smooth silicon sample, including the location of the 7 fiducial cross markers and the two lines of interest. Figure 3: Zoomed photograph showing two of the fiducial marker crosses added to the optical surface using a laser writer at the ESRF. 2.2 Optical metrology instruments Prior to processing the samples using deterministic correction methods, each was first measured at Diamond and ESRF using Fizeau- & micro-interferometers (Figure 4). Each lab employed their standard acquisition and analysis protocols. Table 1 and Table 2 respectively show the relevant parameters of the microand Fizeau interferometers used in this study. Deliverable no. D3.5 Page 9 of 41 Table 1: Parameters of the micro-interferometers Diamond ESRF Bertin-Winlight Manufacturer Bruker WYKO Zygo Model Contour GTX NT9300 NewView Number of pixels 1200 x 1000 640 x 480 640 x 480 In-plane sampling 0.68 µm 2 µm 1.1 µm Magnifications 2.5X, 10X, 50X 2.5X, 5X, 50X 10X Measurement technique PSI PSI PSI Table 2: Parameters of the Fizeau interferometers Diamond ESRF Manufacturer Zygo Zygo Model Verifire HDX GPIAT+ Diameter of beam 152 mm 152 mm Number of pixels 3392 x 3392 1000 x 1000 In-plane sampling 45 µm From 158 µm to 32 µm Magnification 1X (fixed) Continuously variable from 1X to 5X Figure 4: (Left) Sample measured on the Fizeau interferometer (Zygo HDX) at the Optics Metrology Lab at Diamond. (Right) Sample measured on micro-interferometer (WYKO NT9300) at ESRF. 2.3 Fiducial alignment In a previous, pan-European metrology study of X-ray optics (MooNpics JRA), fiducial markers were shown to be highly beneficial to aid the alignment of measurement and analysis regions for the different instruments. Such marks were made using ion beam figuring, diamond-tipped scribes, or laser writers. Lessons learned were applied to this project. Fizeauand micro-interferometry scans were performed to investigate how reliably specific regions of the optical surface could be located. The XY lateral scaling of the Fizeau (pixel size and zoom factor) was carefully pre-calibrated using traceable dimensional artefacts. As shown in Figure 5, the fiducial crosses could be located within the XY plane of the optical surface to < 20 µm (< ½ of a CCD pixel of the Diamond HDX Fizeau interferometer). Deliverable no. D3.5 Page 16 of 41 Figure 14: Preliminary Fizeau interferometry of several optics resting directly on the tip/tilt stage revealed a repeated pattern of distortions induced in the optical surface (left column). After supporting each sample on a flat substrate (right column), such distortions were removed, highlighting the importance of carefully supporting optics. Figure 15 shows a representative comparison of 2D Fizeau interferometry of the clear aperture of Lane B for sample #1, as measured at Diamond (upper image) and ESRF (lower image). Figure 16 shows excellent reproducibility in the central, 1D line profiles extracted from the 2D data. Figure 15: Comparison between Fizeau interferometry at Diamond (upper plot) and ESRF (lower plot) for Lane B of sample #1 shows sub-100 pm agreement in the 2D height profile. Deliverable no. D3.5 Page 17 of 41 Figure 16: Central line profiles extracted from the the 2D Fizeau data shown in Figure 15. 2.6 Comparison between stitching micro-interferometry and Fizeau interferometry Many micro-interferometers incorporate encoded, 2D motion stages to translate the sample in the X and Y directions relative to the microscope objective. Such instruments also include control software to script and synchronise sample motion and interferometer data acquisition, thereby enabling programmable, sub-aperture stitching. After data collection, adjacent overlapping images are stitched together using dedicated software. For this work, the PyLOSt software was used. This was developed initially during the MooNpics project and has since been continuously updated and enhanced in functionality throughout the LEAPS-INNOV project. The upper panel in Figure 17 shows 2D selected regions of the clear aperture (~ 1.5 mm wide, running along the full 40 mm length of the optic) of Lane B of sample #8, as measured by Fizeauand stitching micro-interferometers at Diamond and ESRF. The lower panel in Figure 17 shows 1D line profiles extracted from each 2D map. The small variation in low-spatial frequency errors (only 100 to 200 pm peak-to-valley) can likely be attributed to a combination of differences in mounting, gravitation sag, stitching artefacts, and systematic instrument errors. Height errors of between 145 and 210 picometres rms, as measured by the various instruments, are state-of-the-art for flat mirrors, and demonstrate that traditional pitch-polishing can achieve exceptional results for flat optics -2 -1.5 -1 -0.5 0 0.5 1 -20 -10 010 20 nm mm ESRF_AT+ DLS_HDX Deliverable no. D3.5 Page 18 of 41 . Figure 17: 2D (upper panel) and 1D line profiles (lower panel) of Fizeau and stitching micro-interferometers. 3 Deterministic correction of X-ray optics 3.1 Overview After approaching the required curvature and micro-roughness using traditional stochastic polishing techniques, deterministic correction techniques are routinely used to fabricate state-of-the-art optics. Deterministic correction techniques use metrology data to guide the correction tool to selectively reduce optical surface errors or, for example, to introduce surface asphericity. Deliverable no. D3.5 Page 19 of 41 Subtractive correction methods, including ion beam figuring2,3 (IBF), magnetorheological finishing4 (MRF), fluid-jet polishing (FJP)5, plasma chemical vaporization machining (PCVM)6 , and elastic emission machining (EEM)7, selectively remove material from high points on the surface. Conversely, additive correction methods, such as differential deposition8,9,10 (DD), selectively deposit a deeper coating at the low points of the surface. A brief description of each correction method is provided in Sections 3.3 to 3.8. For modern synchrotron and free-electron X-ray sources, it is common for reflective X-ray optics specifications to require midand low-spatial frequency height errors less than a few nanometres (with high-spatial frequency micro-roughness < 0.3 nm rms) over aperture lengths up to ~ 1 m. For the most demanding applications, the required height errors can be < 1 nm and micro-roughness < 0.1 nm rms. Such height errors cannot be achieved reproducibly using conventional abrasive polishing. All pertinent processes must be capable of producing single crystal silicon optics which are, to date, by virtue of their physical properties, the best adapted to the specific constraints of applications with high power X-ray beams reflected in grazing incidence conditions. For correction of the optical surface following conventional polishing, iterative cycles of measurement driving the application of deterministic correction processes are commonly used. The removal or deposition rates for deterministic correction are typically of the order of nanometres per second. The correction tool function, typically 1 to 20 mm in size, is chosen to match to the in-plane dimensions of the polishing errors. 3.2 Technical requirements Effective correction of the optical surface using deterministic methods requires: a) Accurate metrology of the surface topography, including knowledge of the location of the polishing errors on the substrate. 2 “Surface Figuring Using Neutral Ion Beams.” Wilson, S. R., D. W. Reicher & J. R. McNeil., Proc SPIE 966:74–81 (1989) 3 “Ion beam figuring for X-ray mirrors: history, state-of-the-art and future prospects“, R. Shurvinton et al, J. Synch. Rad. 31, 655–669 (2024). 4 “History of Magnetorheological Finishing“, D.C. Harris, Proc. SPIE, Vol. 8016 80160N-1 (2011) 5 “Fluid Jet Polishing of Optical Surfaces”, O. W. Fähnle, H. van Brug, and H. J. Frankena. Appl. Opt. 37, 6771–73 (1998) 6 “The Study of Fabrication of the X-Ray Mirror by Numerically Controlled Plasma Chemical Vaporization Machining: Development of the Machine for the x-Ray Mirror Fabrication”, Y.Mori, K. Yamamura & Y. Sano, Rev. Sci. Inst. 71, 4620–26 (2000) 7 “Elastic Emission Machining.” Y. Mori, K. Yamauchi & K. Endo, Precision Eng. 9, 123-28 (1987) 8 “Highly Accurate Differential Deposition for X-Ray Reflective Optics”, S. Handa et al., Surf. Int. Anal. 40, 1019– 22 (2008) 9 ”X-ray mirrors with sub-nanometre figure errors obtained by differential deposition of thin WSi2 films”, P. Bras et al, J. Synch. Rad. 30, 708–716 (2023). 10 “X-ray mirror figure correction using differential deposition”, C. Morawe et al, Proc. SPIE 13150 (2024) Deliverable no. D3.5 Page 20 of 41 b) Accurate positioning of the correction tool relative to the polishing errors. c) Characterisation of the size and shape of the removal / addition function of the correction tool. d) Accurate motion stages to dynamically move the correction tool along the length (and width) of the optic, following a specified trajectory and dwell pattern. e) Stable, well calibrated, rate of removal / addition by the correction tool. Systematic and / or random errors in the above parameters combine to deteriorate the final quality of the optical surface. Consideration also needs to be given to experimental factors which can limit the range, sensitivity, and quality of each processing technique or instrument, including: a) Length, width, and thickness of samples that can be accommodated and processed. b) Minimum size and shape of processing tool function (which determines the minimum size of in-plane spatial periods that you can be corrected). c) Rate of removal for coarse or fine correction (e.g. nm per minute) d) Calibration and long-term stability of tool size and correction rate. e) Positional repeatability to align the correction tool relative to errors on the optical surface. f) Maximum depth of removal / addition before problems start to occur (e.g. increased microroughness or non-uniformity of surface topography, or induced strain within thicker coating layers). 3.3 Ion Beam Figuring (IBF) Ion beam figuring was performed in the in-house IBF system at Diamond Light Source 11. The system is housed within a vacuum chamber ~ 1.6 x 0.9 x 1 m in dimension. The main components of the system are: KRI KDC100 gridded ion source, which generates a divergent beam of Ar+ ions; a pair of graphite aperture plates (one fixed-position, one movable), which mask the ion beam to the desired shape and size; and a 4-axis motion stage, upon which the workpiece is mounted during figuring. The system is also equipped with an onboard camera to aid in targeting. A schematic of the IBF system, along with labelled photographs, is shown in Figure 18. 11 “Ion beam figuring and optical metrology system for synchrotron x-ray mirrors“, M. Hand et al., Proc. of SPIE, Vol. 11109 111090A (2019) Deliverable no. D3.5 Page 21 of 41 Figure 18: Schematic of the IBF system at Diamond, with labelled photographs showing the key components. The motion stage has a travel range of 800 mm horizontally and 60mm vertically, allowing samples of up to 400 mm in length to be figured. The on-board CCD camera was used to align the samples with the ion beam by identifying the position of fiducial marks to ~ 10 µm in precision and ~ 50 µm repeatability. This also allows any parasitic rotation between the IBF motion coordinates and the measurement coordinates to be identified. A movable graphite plate, positioned in front of the ion source, defines the size and shape of the beam. Different apertures can be selected, including circular apertures between 1 and 10 mm in diameter, and rectangular apertures with widths between 0.75 and 10 mm. Larger beam sizes are suitable for coarse figuring, whereas smaller beams are necessary for fine figuring and the correction of higher-frequency surface errors. The beam removal function (BRF) of the ion beam for a given aperture can be characterised by etching and measuring the beam footprint onto a test sample. This also allows the position of the centre of the ion beam to be calibrated relative to the motion system / camera coordinates for precise targeting. The etching rate of the ion beam for silicon is ~ 1 to 1.25 nm/s, depending on the aperture size and the distance between the sample and the ion source. Deliverable no. D3.5 Page 22 of 41 Previous tests have demonstrated than the ion beam etching rate and BRF are highly stable. However, during testing it was found that over extended periods of operation (>2hrs), heating from the ion beam caused the effective beam position to drift by around 0.2 mm horizontally and up to 1 mm vertically. This can be mitigated by minimizing the amount of time the ion source is operational, and by adjusting the position of the aperture plate at regular intervals to compensate for the shift. Further solutions involving water cooling and thermal insulation are currently being investigated. 3.4 Differential Deposition (DD) In DD, a corrective thin film of variable thickness is deposited to compensate for the height errors of the optical surface. Although many different thin film deposition technologies exist, DD is most usually implemented using magnetron sputtering12 due to its high process stability, versatility, and uniformity. The DD process developed at the ESRF relies on a controlled-speed substrate motion in front of a static sputtering source13. Other groups8,14 have adopted the same approach. To achieve the required corrections, layers of variable thickness are deposited through beam-defining apertures of different openings. Choice of the beam-defining aperture offers the possibility to initially correct long period figure errors faster using a broad flux profile, and subsequently to refine the process with further iterations selecting narrower apertures with a consequent reduction in the deposition rate. Currently at the ESRF, the minimum aperture width is 1 mm giving an approximately Gaussian particle flux distribution of 1.4 mm FWHM at the substrate surface. Whilst, in principle, the process can be adapted to corrections of height errors in 2 dimensions on the optical surface, to date all known implementations of DD processes only include one motorised translation of the substrate which, with no masking in the sagittal direction, limits the optimal height correction to a single line along the tangential direction of the substrate surface. Due to the highly asymmetric aperture of most X-ray optics used at highly collimated synchrotron or FEL sources, this is not a serious handicap as, with appropriate masking in the sagittal direction, several tangential traces can be independently corrected. In practical applications, after correction, a functional optical coating (single or multilayer) will be applied to the surface to provide the required X-ray reflective properties. DD is currently a developmental technique which is not known to be applied in an industrial context. It is readily implemented in existing thin film deposition systems offering the versatility of combining the surface correction and optical coating processes in a single tool. Although not relevant for the silicon optics predominantly used in X-ray applications, DD, also has the potential to be applied to surfaces for which no effective subtractive correction methods may be available. At the ESRF, WSi2 has been determined to be particularly well suited as a material for the corrective layer. This is due to its ability to conserve the initial substrate surface roughness and to limit the film 12 “Magnetron sputtering – Milestones of 30 years”, G. Bräuer et al, Vacuum, 84, 1354–1359 (2010) 13 “Thickness control of large area x-ray multilayers“, C. Morawe & J-C. Peffen, Proc. SPIE, 7448 74480H (2009) 14 “Development of a one-dimensional differential deposition system for X-ray mirror figure correction” J. Kim et al, Precision. Eng., 71, 1-6 (2021) Deliverable no. D3.5 Page 23 of 41 stress to acceptable levels even with thicknesses of the deposited layer exceeding 400 nm. Typical deposition rates are ~ 0.3 - 0.5 nm/s. The substrates are translated in front of the particle source with freeform velocity profiles that are calculated with a deconvolution algorithm according to the target deposition thickness profile. 3.5 Magnetorheological finishing (MRF) In MRF, a fluid containing iron-rich magnetic particles and non-magnetic abrasives is continuously delivered to a rotating processing tool. A magnetic field applied in the contact area between the wheel and the optical surface increases the fluid viscosity, thus creating a soft polishing-like lap which can conform to the local surface of the workpiece. Material removal occurs by a classical abrasive mechanism but in a highly localised zone around the wheel apex where the fluid is ‘stiffened’. By modifying the dimensions of the wheel, it is possible to vary the effective tool size for the correction process. The principles of MRF were originally invented in Belarus and subsequently developed at the Centre for Optics Manufacturing at the University of Rochester (USA). The technique is now commercialised by QED Technologies (USA) who sell machining tools to optic manufacturers. These tools have found widespread application for the machining of visible light optics, but anecdotal feedback from manufacturers have indicated that the MRF process introduces significant mid-spatial frequency height errors which are subsequently difficult to remove. Nevertheless, publicity by QED indicates that under certain experimental conditions, a flat surface with a 2 nm peak-to-valley can be achieved. In a separate study, roughness of 0.1 nm rms was achieved, albeit at the very high spatial resolutions measured by Atomic Force Microscopy (AFM). 3.6 Fluid jet polishing (FJP) In this process, an abrasive slurry is passed through a nozzle at high-pressure (~ 5 - 10 bar) onto the optical surface. Typically, millimetre-diameter nozzles are positioned a few millimetres from the workpiece. Material removal is primarily through mechanical abrasion, but chemical mechanisms can be activated by modifying the slurry composition. The application of this technique to correct optical surfaces was originally reported in Delft (Netherlands)15. Although several groups have explored this method, its potential for figure correction at the sub-nm scale, whilst preserving surface roughness at the 0.1 nm rms level, remains untested. This is the development route currently being explored by one of the PCP contractors in Task 2 of the LEAPS-INNOV WP3. 3.7 Elastic Emission Machining (EEM) In initial implementations of this process7, ultra-fine ZrO2-based nano-particles are mixed with highpurity water to form a machining slurry. Using a dynamic approach akin to hydrodynamic lubrication, pristine powder particles are continually brought onto the optical surface via a rotating polymer sphere in proximity to the surface. Machining proceeds via ‘stripping’ of atoms from the optical surface via a chemical removal mechanism. As with the other deterministic correction techniques, the material removal is controlled by moving the tool (i.e. sphere) with a well-defined velocity profile over the 15 “Fluid Jet Polishing of Optical Surfaces.”O. W.Fähnle et al., App. Opt. 37, 6771–73 (1998) Deliverable no. D3.5 Page 24 of 41 optical surface. The technique is particularly effective for machining Si and can correct the surface in two-dimensions. The replacement of the rotating sphere with a small diameter (< 150 µm) nozzle allows the effect size of the correction tool to be reduced to ~100 µm16. Following its original development at Osaka University (Japan), the technologies have been transferred to JTEC Corporation who are the sole company exploiting the process for the manufacture of X-ray optics. The quality of optical surfaces produced by EEM is exceptional. Coupled with high-performance optical metrology, JTEC produce plane, spherical and aspherical Si optical surfaces with sub-nanometre figure errors and surface roughness < 0.1 nm rms. EEM technologies are not available to European optic manufacturers, which motivates the exploration of alternative deterministic correction methods capable of equivalent or enhanced performance. 3.8 Plasma Chemical Vaporisation Machining (PCVM) PCVM is a dry etching method using plasma at atmospheric pressure. A plasma of reactive atoms is created close to the optical surface and material is removed following the formation of volatile materials by reaction between the plasma and the surface atoms. The use of PCVM for machining Si is possible using a He:CF4:O processing gas combination. The process has been heavily developed by the same Osaka University group behind the development of EEM. Similar approaches, but with different names, have been developed by other groups e.g. Plasma Assisted Chemical Etching (PACE)17, Reactive Atom Plasma (RAP)18, and Plasma Jet Machining (PJM)19. The technique can offer extremely high material removal rates (MRR) which can attain 250 nm/s in extreme conditions. Nevertheless, the MRR is strongly temperature dependent and progressive heating of the substrate by the plasma can render the figure correction process difficult to control as the etching rate can vary significantly during machining. Also, the PCVM technique has not been shown to be capable of preserving surface roughness at the 0.1 nm rms level. It is believed that these shortcomings led the Osaka group to concentrate their efforts on the EEM technique for the correction of reflective X-ray optics. 3.9 Scope of the deterministic correction tests Given the diversity of deterministic correction technologies, it was not possible to explore the performance of all methods. Based upon the available literature, much of which has been aimed at producing X-ray mirrors, PCVM was considered unlikely to achieve the necessary performance. Similarly, FJP is being studied elsewhere in the WP3 work-program, and results are not yet available from this task. EEM is a proprietary technique, only available from a single Japanese company, which is proven to be capable of satisfying the requirements for many high-performance X-ray reflective optics. 16 “Development of nanometer level accurate computer-controlled figuring with high spatial resolution and its application to hard X-ray focusing mirror“, H. Mimura et al., J. Jap. Soc. Prec. Eng. 76 338-342 (2010) 17 “Rapid, Noncontact Optical Figuring of Aspheric Surfaces with Plasma-Assisted Chemical Etching.” D. Bollinger et al, Proc. SPIE 1333:44–57 (1990) 18 “Rapid Optical Surface Figuring Using Reactive Atom Plasma.” M. Castelli et al. Prec Eng 36, 467–76 (2012) 19 “Plasma Jet Machining” T. Arnold et al., Vakuum in Forschung und Praxis 22, 10-16 (2010) Deliverable no. D3.5 Page 25 of 41 In this study, we chose to investigate the performance of two specific techniques, IBF and DD, which are well adapted to the correction of silicon surfaces. Moreover, these technologies could be readily transferred to manufacturers without concern for IP aspects. Diamond (IBF) and the ESRF (DD) have been developing expertise in these techniques over several years using in-house designed fabrication tools, with the aim of improve the quality of commercially-available X-ray reflective optics. IBF is also implemented by most of the European X-ray optic manufacturers, including those involved in the WP3 PCP. We intend to compare the performance of at least one of these correction tools with that at Diamond. However, to date, production constraints at manufacturers have meant that the processing of test samples has not been possible. However, one sample should become available to the consortium by April 2025. Attempts to access MRF processing tools for performance investigations at manufacturing sites which have experience in processing Si surfaces have also proven difficult to schedule. These tools are prioritised for production activities, and consequently performing pertinent tests would require specific machine setups. Tests are foreseen at a production site in France in May, which would allow performance evaluation before the end of the project. Whilst none of the planned tests at industrial sites are an essential element to this study, results from any subsequently processed samples will be incorporated in a future version of this report. 4 Optical metrology after deterministic processing After completion of the round-robin metrology exercise, samples were shipped to the various partners to apply deterministic corrections. Fabricators were instructed to create several “features” at specific regions on the optical surface, including: Lane 1 with a “wedge” removal (linearly-increasing depth of removal along length of optic); and Lane 2 with a spatially-varying chirp pattern. Partners were free to choose realistic parameters for the wedge and chirp profiles, or propose alternative features, based on their processing hardware. However, it was strongly encouraged that parameters should be challenging and reach a point at which each technique would begin to show signs of failure. The “sawtooth” wedge was specified to have a depth between 1 and 3 µm. Removal / deposition of several microns of material is rather extreme for typical correction of X-ray mirrors, but could be of technological relevance for converting a flat optic into a cylinder, or a cylinder into an ellipse. Figure 19 shows the location and depth profile of the idealised wedge for Lane 1. Deliverable no. D3.5 Page 32 of 41 Figure 25: Predicted thickness of the WSi2 coating to achieve the required profile (black curve) using two iterations of coarse (blue curve) and fine (red curve) differential deposition at ESRF. Figure 26: Fizeauand stitching-micro interferometry of “roof-top” profile created using differential deposition. Surface homogeneity on the rising and falling planes of the roof-top is shown in Figure 27 and Figure 28 respectively. The peak to valley height errors of both segments are < 10 nm, which when related to an average deposition thickness of 500 nm approximates to a process stability ~2%. This confirms DD is well suited as an iterative correction process. Deliverable no. D3.5 Page 33 of 41 Figure 27: Height error on the rising edge (16 x 1.9 mm aperture) of the “roof-top” profile (after removal of the bestfitting upward-inclined plane) created on sample #1 using Differential Deposition. Note, zero on the horizontal axis of this chart corresponds to -12 mm from the centre of the mirror. Deliverable no. D3.5 Page 34 of 41 Figure 28: Height error of the falling edge (16 x 1.9 mm aperture) of the “roof-top” profile (after removal of the best-fitting downward-inclined plane) created on sample #1 using Differential Deposition. Note, zero on the horizontal axis of this chart corresponds to +10 mm from the centre of the mirror. A range of surface defects are observed in the large area Fizeau measurements. Figure 29 charts the micro-roughness of the leading edge of the roof-top profile before (upper row) and after (lower row) differential deposition. After deposition, the micro-roughness Sq is substantially increased, by amplification of small surface defects or contaminants that were present before coating. Some areas also show much larger point defects (e.g. ROI4 after coating). The origin of these large defects is unclear, but on the final sample they were also observed in areas of the surface which had been masked during the deposition process, indicating that they may be particulate contamination from an unknown source during the characterisation process. Deliverable no. D3.5 Page 35 of 41 Figure 29: Micro-roughness of regions of interest 2 to 5 (indicated in upper schematic) on the leading edge of the rooftop profile, before (upper row) and after (lower row) differential deposition. 4.4 Differential Deposition at ESRF: spatial chirp Figure 30 illustrates the design parameters and equation of the spatial chirp to be added to Lane 2 of sample #1 via differential deposition of WSi2. A constant amplitude of 10 nm was chosen, with a chirp period varying between 5.8 and 2.2 mm. Figure 30: Design profile of the 10 nm amplitude spatial chirp to be created using differential deposition, showing the target WSi2 coating thickness as a function of length along the sample. Deliverable no. D3.5 Page 36 of 41 Figure 31 shows the design profile (blue curve) and the best fit calculation (orange curve) of the chirp profile and the expected height errors. These predict negligible height errors at chirp periods above ~2.5 mm. At lower periods peak-to-peak height errors are not predicted to exceed ~1nm. Figure 31: (Upper chart) Ideal and best fit profile modelled using a 1 mm deposition aperture. (Lower chart) predicted profile error of the differential deposition chirp Figure 32 and Figure 33 respectively show the measured height profile (after removal of best fit plane) and height error profile (removal of best fit cylinder) of the chirp lane created using differential deposition. Aside from creating the chirp, deposition also caused a -10 km (convex) radius of curvature change in the optical surface. This may be induced by residual stresses in the deposited WSi2 films on the sample. Such effects are expected for thick films on thin-substrates. Experience with sputtercoating of X-ray optics indicates that with thicker substrates and thinner films, the stress induced curvature is usually negligible but, if necessary, mitigation strategies exist. Deliverable no. D3.5 Page 37 of 41 Figure 32: Interferometry of the chirp profile created using Differential Deposition (after removal of the sample shape before coating and a best-fit plane). Figure 33: After removal of the best fit cylinder from the data in Figure 32, excellent agreement is obtained between the various types of interferometer. Figure 34 shows the difference between the predicted profile and that measured by the ESRF microinterferometer. This analysis is preliminary but appears to show a small scaling error (< 10 %) in the amplitude of the chirp structure over the full frequency range. The origin of these errors is still under investigation. As expected, the amplitude errors become greater at high spatial frequencies. Although Deliverable no. D3.5 Page 38 of 41 the detailed structure does not follow the prediction of Figure 31, the increased amplitude at the highest frequencies is consistent. Figure 34: Profile error of the differential deposition chirp evaluated from the theoretical profile and the ESRF microinterferometer measurements Figure 35 shows the micro-roughness at 2.5X, 5X, and 50X before (upper row) and after (lower row) deposition of the chirp. If higher-order polynomials are subtracted from the micro-interferometer data (to remove the chirp profile and isolate the underlying micro-topography), features running along the length of the mirror are clearly visible (Figure 36). It is hypothesised that these linear defects may be caused by residual chemical contaminants from a solvent cleaning wipe of the surface prior to coating. Some of these features are already faintly visible on the uncoated surfaces but are revealed more strongly by the coating process. Other studies21 using differential deposition in an iterative correction process have shown that this roughness degradation is not generally observed and support this hypothesis. 21 “X-Ray Mirror Figure Correction Using Differential Deposition” C. Morawe et al. Proc. SPIE 13150:7–16 (2024) Deliverable no. D3.5 Page 39 of 41 Figure 35: Micro-roughness of the chirp lane before (upper row) and after differential deposition of the chirp profile (bottom row). The topographies are taken from ROI11 which corresponds to a valley in the chirp profil,e as shown in the schematic. The rms height values for the low magnifications of the chirp include a contribution from the underlying chirp curvature. Height scale =/- 0.5 nm Deliverable no. D3.5 Page 40 of 41 Figure 36: After subtraction of higher-order polynomials to mask the chirp structure, clear evidence of linear defects are observed, possibly caused by solvent cleaning of the surface prior to coating. 5 Conclusions and perspectives • Successful, first of its kind(?), round-robin metrology comparison of the absolute surface topography of multiple, super-smooth silicon samples with state-of-the-art micro-roughness (Sq < 0.5 Å rms). • After careful cross-calibration (to reduce systematic and random sources of measurement errors), exceptional agreement was observed in the measured surface form and microroughness between various types and models of Fizeauand micro-interferometers. • Fiducial markers enabled reproducible alignment of the optical surface to < 20 µm between different metrology instruments. • We demonstrate that IBF can accurately create free-from surface profiles on the nanometre scale, including spatial features ~ 0.75 mm in width, which are at the limit of what can be achieved commercially. Deliverable no. D3.5 Page 41 of 41 • Interesting features were observed for very deep IBF removals, including an increase of the number of point defects and overall micro-roughness levels. • Differential deposition also successfully created a spatial chirp and variable thickness “rooftop” profile, but changes in micro-roughness were observed. These are thought to be specific to these tests and related to either imperfect cleaning of the initial substrate surface and/or an unknown contamination source in the overall deposition measurement cycle. • Both the IBF implementation at Diamond, and the Differential Deposition process at the ESRF, appear capable of reproducibly correcting surfaces with the precision required for integration into an iterative deterministic correction cycle to produce curved X-ray optics with sub-nm height errors down to spatial periods of ~2 mm (DD) and ~ 0.7 mm (IBF). Both technologies should be readily transferable to an industrial context provided appropriately accurate metrology tools are available. • After the project, these highly-measured mirrors will be made available for other Facilities and will be a valuable resource to cross-calibrate their metrology instruments to improve performance.