MODERN METHODS FOR PRODUCTION OF ASTRONOMICAL AND SPACE MIRRORS
odern high resolution optical and electronic systems of ground, air and space basing use different lens and mirror optics. Lens optics has diameter restrictions of about up to 1 m as the own lens deformations cause significant image distortions, and it’s obviously impossible to release them. Therefore for the lens of bigger diameter one uses the mirrors the surface shape of which can be operated from the back surface to create a reflected beam of diffraction quality.
Nowadays the sizes of monolithic mirrors have reached 8 meters, telescopes with composite mirrors up to 11 meters are created and extra-large telescopes with mirrors diameter of up to 40 meters are designed. Due to development of mirrors manufacturing and control technology more and more complex projects are created requiring solution of even more ambitious tasks. Monolithic thin and light-weight mirrors with diameter of up to 4–8 m with aperture 1:1 and less (an aperture is the ratio of the mirror focal length to the diameter), respectively, highly aspherical mirrors (the sphere deviation reaches several millimeters), extra thin mirrors (the mirror thickness ratio to the diameter is 1:50–1:100), light-weight mirrors (lightening is up to 90%), mirrors with extra-axial surface and with non-round external perimeter, the most complicated ones in manufacture and control in which all the technological optical achievements of telescope builders are shown to the full extent have already been manufactured. The required accuracy of aspherical surfaces deviation from the set form reaches 10–20 nm under the mean square deviation (RMS). Besides, the surface quality is to be kept within mirror operation timeframe reaching 50 years. It became possible to produce all this "optical" variety due to creation of appropriating processing technologies and control methods development on all the production phases. A special place in this process is taken by manufacturing of astronomical and space mirrors as exactly with them it is possible to construct the largest optical systems and to carry out global monitoring of the Earth and near-earth space, to look into distant areas of our Solar system and into the multi-billion past of our Universe.
For many years "Lytkarino Optical Glass Factory", JSC has carried out production of light-weight and thin mirrors of various configurations with diameter of up to 4000 mm from Astrocytall which is most frequently used for mirrors manufacturing though some works have recently been performed to produce mirrors from silicon carbide. Astrocytall has high strength characteristics and low thermal expansion coefficient (TEC) quite lower than other materials (Table 1) that is its basic advantage. Astrocytall СО-115М has the thermal expansion coefficient α = 0 ± 1.5 · 10–7 К–1, tensile strength 78 МPa, elastic modulus 92 hPa. On polished surface of the material it is possible to reach a roughness <1 nm (mean square deviation, RMS). Due to such unique characteristics it has successfully been used for manufacturing of ground-based and space mirrors for many years.
Astrocytall has high dimensional stability, i. e. it keeps the working surface shape with the set accuracy in the course of time in the operation process under different climatic conditions as proved within several decades of manufacturing and operation of optical devices .
Dimensional stability of five grades of glass ceramics with low thermal expansion coefficient used in production of astronomical and space mirrors was measured within a range from –40 °C up to 90 °C and introduced in an article of D. B. Hall . Materials Zerodur, Zerodur M, Astrocytall, Clearceram 55, Clearceram 63 were tested.
Detailed comparison of three various Astrocytall brands, two various Zerodur brands and one Zerodur M brand has shown that within a temperature range –40 °C – 90 °C the dimensional stability and uniformity characteristics of two Astrocytall brands are a bit better than of two other materials. The results prove some advantage of Astrocytall especially from the point of view of lowered isothermal deformability within a range from –40 °C up to 90 °C. Isothermal deformability observed at all the temperatures is mostly expressed at lowered temperatures.
Modern thermal expansion coefficient requirements for example for preparation of extra-large TMT (Thirty Meter Telescope) and E-ELT (European Extremely Large Telescope) telescopes with composite mirrors with diameters of 30 and 39 m establish an absolute value of thermal expansion coefficient at all the measured samples ≤100х10–9K-1, and an average value of thermal expansion coefficient of the sample pieces at operational temperature (–10 °C –20 °C) should be about 0±50х10–9K-1.
20 samples were subjected to accelerated tests to measure the thermal expansion coefficient to research changes of the thermal expansion coefficient in the course of time (40 cycles corresponding to 40 years), and sample pieces with diameters of 500 and 450 mm and thickness of 70 and 50 mm respectively with polished surface (RMS <20 nanometers) were used in the operation process to define changes in the surface shape, and the shape of the reflected wave front and mechanical parameters were defined for them. The following data were obtained based on the tests results:
thermal expansion coefficient changes within the range ±1 · 10–8 °C–1 (fig. 1), the measuring error is 5 · 10–9 °C–1, general tendency of keeping the thermal expansion coefficient is observed, on fig. 2 there is a chart of average value of thermal expansion coefficient based on samples of higher and lower layers and gradient of the thermal expansion efficient is observed in the range 5x10–9 at the depth of 250 mm of the sample pieces;
the optical surface shape does not change within the measuring error range RMS<6 nanometers.
Some tests under tensile deformations within a range from –10 МPа to +10 Mpa and under temperature influence within a range from –40 ˚C to +105 ˚С were carried out for the purpose of checking elasticity characteristics of Astrocytall СО-115М material.
For this purpose some operations on deformation of the sample with diameter of 350 mm and thickness of 10 mm were carried out by turns from the working and back surface. As a result of calculations with the applied method of final elements with ANSYS PO usage it’s been established that ±10 МPа tension arises in the tested material based on force application into the center equal to 333N while fixing at two opposite extreme points. Deformation tests results are specified in fig. 3а. On the chart there are surface shape deviations from the initial value (DRMS = RMSi – RMS0, where RMSi is the current value of the mean square deviation, RMS0 is the initial mean square deviation) for each cycle of the tests. The lines above and below show the tolerance range under modern requirements.
Tests with temperature mode changes were carried out in a thermal vacuum chamber. Cooling to –40 °C was performed with a speed not more than 20 °C per hour. The sample was kept at the temperature –40°±2 °C for an hour. Then the temperature in the chamber was raised up to 105±2 °C with a speed not more than 20 °C per hour. The sample was kept at the temperature 105 °C±2 °C for 1 hour. Then the temperature in the chamber was lowered up to normal one with a speed not more than 20 °C per hour. Such cycles were repeated for several times. A chart was made based on the control results, fig. 3б.
The tests results have showed that Astrocytall СО-115М material keeps its elasticity under tensile deformations within the range from –10 МPа to +10 МPа and under the temperature influence within the range from –40˚C to 105˚С and that the surface shape changes within 10 nm of the surface RMS (fig. 3).
A graphic example of long-term external influence upon astronomical mirrors from Astrocytall can be considered two light-weight mirrors with diameter of 1500 mm made in 1992 (55% weight reduction). The initial wave front error value was RMS(W) = 030λ. 17 years later after storage of the first mirror its error was RMS (W) = 0.042λ (December, 2009). The error of the second mirror after 18 years of storage was RMS (W) = 0.044λ (October, 2010). One can consider that the surface shape quality has remained practically unchanged within a long period of time.
Another example of deformation influence upon the Astrocytall sample piece was manufacturing of an extra-axial mirror segment with diameter of 1520 mm and thickness of 50 mm under the elastic deformation technique. The work was carried out within the experiment of manufacturing of the prototype model of the most remote mirror segment of E-ELT telescope. Deviation of this surface from the nearest sphere was 203 µm. For fast aspherisation of such a segment it’s convenient to bend the sample piece in a way to make the material removal sites bulged. Then the surface is polished with a full-size tool to obtain a spherical surface and the biggest material sampling will be on the sites which are elevated. After deformation sampling one obtains the aspherics that is to be brought in line with the specification requirements by means of program-operated polishing with a fine tool. The internal equivalent tensions arising out of such bending amount to 1.68 МPain the centre. The tensions are quite admissible for sample pieces from Astrocytall with the corresponding geometric parameters. Deformation of the sample piece is performed with the height regulated micrometric supports with the help of clamping mechanisms. The padding thickness is selected in such a way that its own deformations are small comparing with working deformation of the part. As seen from the comparative analysis (fig. 4.5) a prototype model shape obtained in reality looks like theoretically predicted result proving quite a reliable behavior of Astrocytall as for keeping deformation elasticity.
Next important stage is selecting of a mirror model, either a monolithic mirror, a thin or a light-weight one. Let’s stop in more details on construction of a light-weight mirror as different variants of lightening construction are available here requiring special tensile calculations.
Mirror lightening is used both for the ground-based optical electronic devices and space objects to reduce the construction weight. But anyway it’s necessary to take into consideration deformations of light-weight mirror so that it keeps working surface shape within the set range during the operation process.
As a rule, mirrors are reduced in weight from the back side with trapezoid or triangle samples. The outer and inner edges of the mirror are fixed with stiffening plates. Mirrors lightening can be over 80% provided that the construction keeps the required stiffness.
After the selection of a mirror lightening variant, optimization of its geometrical sizes takes place with the purpose of reaching the construction maximum stiffness at its minimal mass. Variable parameters are the mirror height, thickness of undersurface layer before the lightening structure, number and thickness of ribs. While selecting the ribs thickness, one should take into consideration technological production capacities. The size optimization is carried out at calculation of the mirror model of the final element with application of a package of final element analysis.
For example, for a part with diameter of 1900 mm a reasonable variant as for stiffness is a triangle structure (Table 2). Application of a triangle lightening structure during production is the most complicated process. Based on the tests results it’s found out that the most reasonable choice will be a mirror with triangle structure with cell size 120 mm, stiffness coefficient KW=1.43 and lightening coefficient 82.7%.
Geometrical structure production accuracy reaches the value of 20 µm at diameter up to 4000 mm. Milling of the sample pieces is carried out on program controlled machines. An example of manufacturing of a mirror sample piece with diameter of 1200 mm with triangle structure can be seen in fig. 6.
Next important stage is forming working surface of the mirror. Moreover with improvement of shaping technology the designers of optical systems make more complicated requirements both for optics production complexity and for surface shape production accuracy. Nowadays time of the surface shape deviation from the sphere is counted in millimeters and not in microns like before (for example, extra-axial mirror of Magellan telescope with diameter of 8 m has asphericity about 17 mm). Mirror aperture is also significantly increased. The production technology of such mirrors has become more complicated due to this. The most complicated mirrors are the ones with extra-axial aspheric surfaces and random outer contour which can be manufactured only with application of special computer programmed techniques developing successfully at the present time.
Within the last two decades "Lytkarino Optical Glass Factory", JSC has developed computer-programmed techniques of large-scale astronomical and space optics processing that allow producing mirrors for large Russian and foreign telescopes. These are the main and secondary mirrors of the world hugest survey telescopes VST (VLT Survey Telescope, 2.6 m) and VISTA (Visible and Infrared Survey Telescope for Astronomy, 4 m) [3–8], installed in Paranal Observatory, Chile, a network of 17 telescopes LGOGT (Las Cumbres Observatory Global Telescope Network, USA) with diameter of main mirrors of 1 m and secondary ones of 345 mm [5,9], telescope TNT (Thai National telescope, USA, Australia, Thailand) with the main mirror with diameter of 2.4 m [9–10], mirrors of the telescope DOT ARIES (Devasthal Optical Telescope, DOT, for Aryabhatta Research Institute of Observational Sciences, ARIES, Belgium, India) with the main mirror with diameter of 3.7 m , mirrors of the telescopes SALT (Southern African Large Telescope – Large South African Telescope) and LAMOST (Large Sky Area Multi-Object Fibre Spectroscopic Telescope, large multipurpose spectroscope to observe extensive sky areas, China) [12,13]. Mirrors of these telescopes have significant asphericity and high aperture creating substantial difficulties not only within processing of these mirrors but also within their control. For successful production of such optical elements, an optics production process control is required on all the stages of processing starting with the preliminary polishing and finishing with the final fitting.
To create an operating program for the working surface processing on the stage of polishing one needs the surface topography information. It can be obtained in three ways.
mechanical control with the help of 3-grid or 6-grid measuring machine, e. g., KIM-1400 of Russian production providing accuracy of deviations from the required aspheric surface of 2–3 µm ;
three-point lineal spherometer successfully applied at the present time can be used as a surface shape control technique for the parts of any sizes both convex and concave but axisymmetric;
for the parts of non-fixed shape on the polishing stage one can apply Infrared Interferometer [14, 15] in combination with a mirror corrector of the wave front. A mirror corrector for IR-range (λ = 10.6 µm) was used to control the processed surface shape at automated asphericity of the main mirrors of VST and VISTA  projects.
Both on grinding and on polishing stages and especially on the stage of the mirror final fitting it’s based either on nominal release on which it will be used in a telescope or on membranous and pneumatic technological release. Thus for example a TNT mirror with diameter of 2.4 m was processed on a nominal mechanic release, and VST, VISTA  mirrors on a membranous and pneumatic one where the membranes imitated nominal release and were located in the same places as for nominal release. Hence calculations of the mirror release on nominal and technological frame are carried out, the card of a difference in the surface shape is defined and processing and control are conducted taking this card into consideration. It is necessary to point out that nowadays mirrors are made in form of meniscuses of smaller thickness than a classical ratio between the thickness and diameter 1:10, 1:8. So thickness of VISTA mirror is 165 mm at diameter of 4100 mm, i. e. 1:25. It is due to the fact that mirrors have active release system, i. e. manageable surface shape within the observation process.
Lens and mirror-lens wave front correctors [10–11] are used to control concave aspheric surfaces on the polishing stage. They modify spherical or flat wave front into aspheric corresponding to the controllable surface. Constructions of correctors have become significantly more complicated as they have become highly apertured and highly aspheric. Both lens parameters and distances between the lenses must be kept with micron accuracy. But wave front correctors in form of diffraction optical elements or CGH-correctors (CGH-correctors-Computer Generated Hologram) with CGH mirror imitators [10–11, 16–17] have recently been used more and more successfully. CGH mirror imitators allow controlling also the wave front lens corrector. They allowed increasing accuracy significantly and extending the range of functional possibilities of optics control especially of extra-axial mirrors, and increasing adjustment accuracy. Upon complication of optics under production especially large and expensive mirrors it became necessary to provide cross check of optical surface with two correctors of different construction that allowed avoiding mistakes during mirror production and receiving the required values of apical radius and eccentricity of the working surface. Thus measurements of the processed surface radius are carried out with a laser tracker that allows receiving accuracy in radius measurements up to 0.1–0.3 mm at the length of up to 15–30 meters. Such control techniques have allowed receiving precise mirrors which are successfully used now in operating telescopes.
Nowadays there is improvement of shaping technology in a number of directions. But three techniques are the most effective for large-sized optics: computer management with a pitchy or synthetic polishing tool, ionic-beam polishing and magnetorheological polishing. And for the parts with diameter of over 2 meters only the first technique is applicable. All other techniques require further development.
Program managed machines of series "AD", "CD", "APD" [18,19], "KU168-ADM" are used for shaping.
Basic trends in polishing technology development have recently been directed at processing of highly apertured and highly aspheric mirrors or lens respectively thus obtaining a small-structure error which is defined with mean square slope of the beams reflected from the mirror (RMS Slope) measured in angular seconds.
The program complex contains a set of subprograms of different purposes providing carrying out of automated technological process, topography calculation of optical surface under the results of interferential control, calculation of material sampling with an estimated set of polishing tools, calculation of movement trajectory of the polishing tools along the part, calculation of correcting technological parameters, optimization of shaping process, displaying of operating programs for computer management of the polishing tool movement. Polishing tools of different diameters are used in one processing cycle in a technological shaping process with small tools. Combined use of a set of tools with local errors smoothing sessions allows consistent eliminating of errors from low-frequency up to high-frequency ones.
In order to increase sampling capacity on the surfaces with high asphericity forced restraint of the tool from rotation around its longitudinal axis is used, i. e., flexible tools of certain rigidity so that the sampling structure corresponds to the one in meridional and sagittal sections on surface of the part. In this case it is possible to use a tool of quite a big size that gets adjusted to aspheric profile of variable radius on the whole surface of the tool.
In order to create material sampling on surfaces with significant curvature an APD-600-like machine with table slope with the part  is used to provide adjusting of the tool to the working surface of the part in normal in vertical position (fig. 10). It is possible to process the parts on it with diameter of up to 1000 mm.
Modelling of process of material sampling from the surface is carried out in such a way that the material is taken from the whole surface proportionally to the sizes of deviations of a profile from the deepest hole on the surface. I.e., where the maximal bump is, there the maximal sampling is taken. Such way of processing distribution within material sampling from surface allows obtaining smooth optical surfaces with minimal RMS Slope including also edge areas on the part.
On fig. 7.8 there are examples of optical part shaping with small tools under computer control.
A hexapod machine and laser tracker (fig. 9) are used to adjust and fit the mirrors location in space.
As example of results of large-scale astronomical mirror processing there are wave front interferograms of the main mirror of VISTA telescope with diameter of 4100 mm and asphericity of about 850 µm in fig.11: on the left there is highlighting of regular errors, distortive impairments created by the corrector, the own error of the wave front corrector received as based on the results of building a surface topographic map, and on the right there is a real interferogram for a little number of fields. It depicts remanent defocusing and zonal error equivalent to the remanent zonal error of the wave front lens corrector that’s deducted from the total wave front.
"Lytkarino Optical Glass Factory" has also designed ion-beam techniques of optics processing (IBP). A vacuum unit has been created for IBP of the surfaces of large-scale optical parts (‘Lutch (Beam) –2.5" unit, fig. 12) for the parts with diameter of up to 2.5 m. Similar units for composite mirror segments processing were created in France (SAGEM Company), the USA (ITT), they were successfully used to manufacture component parts of main mirrors of Keck I and Keck II (the USA), GRANTECAN (Spain) telescopes, etc.
Upon contact of a polishing tool to surface of the part there are significant errors in the edge area at its crossing over the part edge that’s why light area parameters are specified on the working surface. IBP technique does not have any such disadvantage and it is especially effective for processing of composite mirrors of the telescopes consisting as a rule of hexagonal segments incorporated into a single mirror in which it is required to make the whole surface with optical quality including edge areas of the part. Processing technology is on the stage of modernization and further development.
A laser tracker is used to measure apical radius of the optical surface allowing to measure sections between a wave front corrector and a mirror with accuracy in millimeter fractions, to adjust location of optical elements of the control chart with high accuracy.
In order to evaluate a degree of complexity of surfaces made by "Lytkarino Optical Glass Factory", JSC let’s use a chart (fig. 13) introduced by P. Dierickx in 1999 , supplement it with data of the last decade and specify on it location of the main and secondary mirrors of the key projects processed by "Lytkarino Optical Glass Factory", JSC. The chart shows a degree of complexity of the manufactured mirror depending on dy parameter and reached quality of processing of RMS (W):
dy = 8 N 3/k,
where N is an aperture (аfocal ratio) of optical surface, and k is a conic constant. Thus a conclusion follows that closer to the coordinate basic origin the point is located on the chart, the more complicated the manufactured surface is.
In figure 14 there is a chart of maximal asphericity gradient in micron/mm for the main (abscissa axis) and secondary (Y-axis) mirrors of telescopes: triangles are a number of foreign projects on manufacturing of telescope optics. The circles mark a number of projects executed by "Lytkarino Optical Glass Factory", JSC relating to manufacturing of optics for different telescopes, mainly within the recent time. As seen from the chart, a set of optics of VISTA (3.74; 4.02) telescope surpasses all the earlier main and secondary mirrors significantly in relation of complexity of production and asphericity gradient thus proving success of the technology developed.