2.1

Applications and market potential

The desire to avoid self-shading effects in telescopes, head mounted - / head-up displays or in lithographic systems forces breaks in symmetry and does call for the application of freeform elements in Schiefspiegler systems. Monolithic optics with several freeform shaped surfaces are also entering the market for example as freeform prisms or monolithic telescopes. Oblique or rear projection systems with tilted object or image planes do benefit from freeform elements as well as highly asymmetric scan systems or anamorphic imaging applications. Additionally complex optical systems with decentered or tilted components like beam splitter plates, filter or cover glasses and tilted lenses are suitable for being improved with freeform surfaces. Another field of attraction are adaptive optics, in particular variable refractive optics for wavefront manipulation with moving freeform plates as shown in Figure 1.

Figure 1

Alvarez-type freeform elements enable variable refractive optics

In comparison to pressure variable or electronical controlled fluid lenses or liquid crystal lenses the refractive power of Alvarez-lenses does not vary under temperature influence, does not degenerate over time and allows universal wavefront manipulation with also high refractive power, because every optical glass can be used. A pair of adjustable Alvarez-type freeforms provides variable optical power to the system by lateral movement and does work diffraction limited. Although the technical potential of these Alvarez-type freeforms is evaluated as high, the manufacturing efforts and costs still lower the broad use in optical systems.

2.2

Typical Specifications

There are various challenges being responsible for the high manufacturing efforts. To avoid wavefront errors, the manipulator plates need to be assembled close to each other. Both parts need to be thin. Figure 2 shows a representative drawing extract with typical properties. They also lead to further claims concerning stability in manufacturing process by tensions-free clamping techniques or the application of less process forces in order not to move or bend the parts. This causes less process throughput and additional tight tolerances in the lateral position of the freeform surface to the outer geometry. Since the parts need to be adjusted in all six degrees of freedom, exact references within the parts are required too. Both, the position and the outer geometry, need to be accurate in less than 50 μm.

Figure 2

a) representative sketch of freeform geometric tolerances; b) representative freeform surface with edge extrapolation

This is aggravated by small distances between clear aperture and the parts edges being forced by only little space provided by mechanical design. Since the applications are in high precision imaging, also tight tolerances for form, waviness and roughness need to be achieved. To ensure high contrast values, micro roughness needs to be less than 2 nm for a field of (150×150) μm² over the whole aperture. To separate deviations in form and waviness and their influences on the imaging quality, tolerances for the permitted surface deviation after deduction of adjustment degrees of freedom are split into spatial wavelengths ranging from 1 to > 16 mm with maximum rms ranging from 5 to 15 nm, so less than λ/10 (PV/WF). To make it even a bit more complicated, the refractive power of the glasses needs to be rather high, to keep the surface profiles as low as possible. Those are mostly uncommon glasses for manufacturing processes and do need specific process care and process know-how. Plastic optics do not yet provide the desired imaging quality. Typical surface amplitudes are < 2 mm PV over the whole aperture as shown in Figure 2 and minimal local radii > 40 mm.

3.1

Challenges and achievements

Typical process flow for an optical system from idea to product is shown in Figure 4 and divided into three main parts. The classical process chain for optics manufacturing is part of the whole component manufacturing step and next to the optics design and prior to system integration. This consecutive process sequences do depend on each other and need to be coordinated.

Figure 4

process sequences from idea to product with classical standard process chain for freeform manufacturing³

The classical freeform process chain is also derived from achievements in asphere manufacturing. After pre-processing and fixing the backside of the thin plates on a precision rod or zero point clamping system, a grinding process is applied, followed by polishing, smoothing and figure correction steps depending on final specs.

State of the art ultra-precise machining centers combined with in-house developments enable the production of Alvarez-freeform prototypes. UP-grinding machines with thermo-mechanical-stabilization are capable of shaping other freeforms in serial production down to 0.5 μm accuracy with sub-surface damages in the range of 10 – 20 μm depending on glass and tooling. Representative data is given in Figure 5. The residual form error after 2 to 5 iterative standard grinding steps with diamond grinding wheels is > 2 μm with remaining mid-spatial frequency structures in the range of 0.5 to 5 mm, which induce further iterative correction steps or do hardly fit to improvable wavelengths. In contrast there are no residual structures in the same range after one iterative loop of ultra-precision grinding and a remaining global form error is < 1 μm, which lead to a drop of machining, working and setup time.

Figure 5

MSF-structures measured with CMM after freeform a) standard grinding process compared to b) UP-grinding

As mentioned before an additional advantage of UP-machining of glass is the possible reduction of sub-surface damage induced by the grinding process. Therefore a parameter specific optimization and evaluation of resulting surface quality is necessary. 15 to 20 μm subsurface damage is good value for standard 4-axis grinding processes leading to more than 20 μm crucial pre-polishing material removal to secure passing visual inspection of scratch / dig. Due to precise, hydrostatic guides, dynamic drives and a rigid construction process improvements in microscale gets feasible. Freeform grinding with less than 15 μm can be achieved and enables the reduction of necessary pre-polish removal height and as a result less effort for corrective iterations. Modified tooling and research in grinding theory^4-6 allow additional reduction of sub-surface damage down to < 10 μm. Figure 6 gives an indication by microscopic pictures of ground surfaces for the transition from brittle to ductile grinding for high-index glasses on commercial UP-machinery with modified tooling and process parameter like infeed or cutting speed. This transition seems to be attractive for productive grinding but still suffers from tool wear and workpiece preparation.

Figure 6

microscopic evaluation of different grinding modes a) brittle b) mixed c) ductile

The pre-polish process based on bonnet polishing is driven by the wish to predict local material removal by taking local radii and pressure changes into account. Therefor a GPU-based simulation model for the polishing process is implemented in the post-processing infrastructure. After modelling a specific tool function, its structure and form with finite element methods, a calculation of the localized pressure distribution in the contact zones between the tool and the freeformed workpiece surface is possible as seen in Figure 7 a) is performed. By adapting Preston and other polishing hypotheses a localized removal rate map and based on this a global removal profile are calculated and verified with experimental data. This model suites to optimize polishing and correction processes regarding process parameter, tool concepts and surfaces towards a deterministic, predictable and cost-effective polishing material removal for freeform elements.

Figure 7

a) FEM tool modelling and calculation of localized removal rates on freeform surface by applying pre-calculated pressure distribution and polishing theories b) interferometric measurement of an early stage pre-polished substrate with MSF-structures

The required optical shape is calculated and designed mainly close to the clear aperture. The outer area has often no optical functions, is not determined and does often not even exist after edging to the final part size. But to polish and correct optical surfaces the applied tools need to overrun the edges depending on tool-size and process parameter. For an adequate polishing result steady edge extensions need to be extrapolated and added to the freeform surface without generating dynamic issues like jerks and jumps in acceleration, functional deviation or critical radii.

The point contact of the grinding wheel causes high frequent structures which can be passively smoothed within the subsequent polishing steps. Low frequent structures can be compensated by local correction. But the necessary removal of more than 20 μm by pre-polishing and a first correction step results in further mid-spatial structures in the range of 0.5 to 2 mm and form errors of < 1.5 μm PV as displayed in Figure 7 b). Those structures need to be addressed by further correction effort by smoothing and figure correction. The drawback of these approaches comes with additional loops to correct form deviations added by the tools. This means an increase of internal loops and as follows an increase of manufacturing time and consequential costs. To correct MSF on freeforms, processing methods with smooth, elastic and adaptive tooling are essential. Therefore different approaches for flexible, large aperture tool setups for the removal of MSF-errors are under development. A composition of tool rigidity, the flexibility of pad materials and the calculation of tool paths needs to be optimized towards selective smoothing of MSF-structures. A frequency selective smoothing effect for a test pattern composed of wide spatial frequency disturbance as seen in Figure 8 is realized for critical wavelengths of 1 - 5 mm with a decrease from 25 nm down to < 5 nm in spatial rms.

Figure 8

interferometric measurements of a) test pattern and b) test pattern after selective smoothing and c) results after spatial frequency analysis

To correct further non-rotationally symmetric deviations, to achieve form accuracies < 0.1 μm and to ensure tight micro roughness requirements, processes like MRF (magneto-rheological finishing) get finally applied. The achievable accuracy is practically limited by the measuring accuracy.

Furthermore a few comments on materials which do play a big role but being not addressed adequate in this explanations on freeform process chain. In particular soft and chemically unstable glass sorts may require special process treatment and additional development, which lead to higher costs. The optimal result concerning process costs and system performance depends on the material choice and a close cooperation between optical designer and manufacturer.

4.1

Laser-based process chains

Beside the classical process steps, laser based process chains or single process steps show advantages for optics manufacturing. The achievements over the last years are continuing in fields of ultrashort pulsed (USP) lasers and their application and widen the landscape of alternative technologies for optical component manufacturing and system integration. There are many potential USP-laser based processes in the long-term forecast: Athermal ablation for 3D-material removal^7-9 or local polishing^10-13 by selective thermal interaction are two promising technologies, which combined in one machine could disrupt optics manufacturing. Other approaches like edging by ablation or by filament generation up to several millimeter thick glasses¹⁴, direct laser micro- or nanostructuring^15-16, selective laser etching for micromachining 3D-parts^17-18 or additive manufacturing techniques like two-photon polymerization for structured or printed optics¹⁹ are already applied in industrial environment. USP-laser bonding^20-22 or in-volume marking²³ do provide new optical features and benefits for systems integration / alignment.

The USP-laser based ablation seems to be attractive for shaping freeform surfaces in a process chain like it is displayed in Figure 9. Processes such as grinding are limited in geometric diversity due to regiments in minimal tool sizes³. Machine builders build stiff and high precision and though cost-intensive machining centers to follow trends towards reliable ultra-precision grinding for serial production. As laser source manufactures came up with stable und high out_put power ultrashort pulsed lasers, contact- and wearfree as well as clean industrial machining of every material became a focus of broad attention. Sharp focusing of ultrashort laser pulses leads to very high intensities which enables nonlinear multiphoton absorption and plasma interaction no matter which material is irradiated. As consequence of physical processes matter gets ablated in the micrometer range by only very low thermal influence on material properties²⁴. A constant material removal as well as the possibility of macroscopic glass ablation for optics manufacturing is possible^25-29.

Figure 9

exemplary alternative process steps for freeform optics manufacturing

Primary laser parameter like pulse energy or fluence, wavelength as well as the applied pulse length and also geometric parameter like line or pulse overlap as well as material and environmental conditions do have to play a significant effect on the target figures roughness, induced sub-surface damage, form deviation and possible induced stress birefringence. After a certain amount of overscans, ablation regime runs into roughness saturation and constant ablation depths. With a so called multipass ablation process characteristic surface roughness rms < 1 μm, ablation depths > 0.1μm per overscan (surface passes), form deviations < 5 μm and sub-surface damages of less than 10 μm are attainable as exemplarily shown for fs- and ps-laser treated fused silica surfaces in Figure 10.

Figure 10

stepwise polishing of fs-/ps-laser treated fused silica samples for sub-surface damage analysis²⁹

4.2

Precision glass molding

Glass molding is an attractive method for mid- and high-volume manufacturing of spherical, aspherical as well as freeform optical elements. The whole classical process chain (grinding, polishing and correction steps) can be substituted and shortened as Figure 10 shows. Depending on desired quality two process regimes, non-isothermal and isothermal molding are available. First being applicable for high quantities, simple glasses but complex shapes, is limited to illumination optics due to temperature differences in process flow and thus less form accurate. Second regime (aka: precision glass molding) is suitable for middle quantities, delivers qualities for photo lenses, laser applications, micro- and IR-optics made from special low-Tg glasses. In a thermally controlled regime the glass-preform gets pressed into the preheated tool and then cooled according to predetermined cooling curves. A multistage tool and shrinkage optimization process in combination with finite element process modelling is essential to achieve qualities in form accuracies < 1 μm and roughness < 3 nm for aspherical and freeform surfaces. Due to long contact times and high temperatures between 500 – 700 °C, the life cycle times of the tungsten carbide tools need to be extended by sputtering bonding and separating layers on the tool shapes and a cleanroom environment. Ultra-precision manufacturing machines, like UP-grinders, are necessary for the tool making processes. The high effort and preliminary development, put into several months lasting prototyping phases, can only be justified with quantities in the middle range starting from several 100 replicates of one tool geometry^30-31. Actual subject for applied research, in the fields of process management, modelling and feedback into optics design, is the process-related drop and gradual change of refractive index in the range of 10^-3 – 10^-4, the residual stress birefringence after molding the glass as well as the tough aspect ratio of the desired freeform plates. With positive achievements regarding these topics, sub-μm form errors seem to be likely achievable.