Gran Telescopio Canarias optics manufacture : Final ReportRoland GEYL, Marc CAYREL, Michel TARREAU

SAGEM Aerospace & Defence - REOSC High Performance OpticsAvenue de la Tour Maury - 91280 Saint Pierre du Perray, France

roland.geyl@sagem.com

ABSTRACTThis paper is intended to establish, after delivery of the last batch of segments, the final progress report of themanufacturing and testing of the Gran Telescopio Canarias optics,

Keywords : GTC, large optics, aspheric, active optics, polishing, testing, lightweight, Beryllium, SiC.

1. INTRODUCTION

The REOSC High Performance Optics team of SAGEM has been selected by GRANTECAN SA for the opticalmanufacturing of the Gran Telescopio Canarias (GTC) segmented 10,4-m Zerodur primary mirror and the 1,18-mlightweight Beryllium secondary mirror. The so-called master segment has been produced in September 2003 and wasused to test the first set of 6 primary mirror segments in March 2004. The last batch of segment was delivered inDecember 2005 and the Secondary Mirror in March 2006. This paper aiming to be the synthesis of the project willtherefore briefly present the project and its optics specifications. Developments conducted to set up our productionfacility within our giant optics laboratory of Saint Pierre du Perray will be presented. The final performance results ofboth primary mirror segments and secondary mirror, of unprecedented quality through the world, will be reported.

2. THE GRAN TELESCOPIO CANARIAS PROJECT

SPAIN, in partnership with Mexico and the University of Florida (USA), hasdecided to set up a state of the art large segmented aperture telescope : the GranTelescopio Canarias (GTC) in the Canaries islands, on top the Roque delMuchachos observatory, La Palma island. The GTC project office has been set upas an independent body with the status of a non-profit enterprise calledGRANTECAN S.A. owned in share between the Canarian Regional Governmentand the Spanish National Government.

The telescope is inspired from the Keck project but is designed for improvedperformances. The near 10,4-m equivalent diameter (11,4-m from edge to edge)primary mirror is split in 36 hexagonal segments as shown on the sketch at right.Each segment contour is inscribed in an 1.87-m circle.

The substrate of the segments is Zerodur material from Schott in the form of athin meniscus of 80 mm thickness. Each segment is axially supported on a 36point whiffle tree support system as shown on the CAD view at left. Six of thesesupport points are active in order to ensure the capability to generate some loworder optical surface deformation thanks to bending moments induced in themirror.

The primary mirror area is paved in a six-fold symmetry with six different typesof segments. Six additional segments bring the total number of segments to bemanufactured to 42 and are dedicated to be used to ease the periodicmaintenance and act as spares for the project.

GRANTECAN SA selected SAGEM-REOSC for the polishing of the 42 primary mirror segments and to integrate themon their whiffle tree support system. In addition, SAGEM-REOSC got the contract for the GTC beryllium secondarymirror design and manufacture.

3. THE GTC PRIMARY MIRROR SEGMENTS SPECIFICATIONS

The optical figuring specifications of the GTC primary mirror segments take into account the presence of the activesupport, the effect of atmospheric turbulence but are also driven by the goal of high quality imagery despite thesegmented nature of the pupil of the telescope. The segments shall be submitted to final optical performanceverification while integrated on their support with the possibility for SAGEM-REOSC to use this active support toremove some low spatial frequency figuring errors, but within a limit of ± 5 N.m max bending moment.

The high spatial frequency errors are specified by the Central Intensity Ratio (CIR) as per the ESO VLT project inorder to take into account an atmospheric free-air seeing of 0.4 arcsec at a wavelength of 500 nm corresponding to aFried parameter of 0,2516 meter. A min CIR performance level is requested for each segment. A mean CIR level isthen required for any set of 36 segments within the 42 to be produced. The contribution of local edge deformations(generally edge turn down) are also to be taken into account in the calculation of each segment CIR.

The segments are aspheric off-axis mirrors. Their relative position and orientation with respect to the vertex of theparent mirror has also to be achieved and known with a high accuracy. The main optical specifications of the primarymirror segments are summarized here :

Radius : R = 33.000 ± 10mm Inter-segment radius variation : ± 0.2 mm, 0.1 mm RMSConic constant : ε = -1,00225 ± 0,0005Active correction : < 5 N.m bending momentIndividual CIR : > 0,887 (λ = 500nm, r0 = 252mm) including edge effects Global CIR : > 0,90Relative position : ± 0,25 mm / ± 0,5 mradMicro roughness : < 20 Angstrom RMSCosmetic defects : scratch dig 80/60

The six types of segments represent six different aspheric surfaces to be polished withdifferent sag with respect to the best fit sphere. The figure below and at right show thesetypes with their aspheric departure contour level with 10,6 µm iso-level line spacing.

As explained below in a next paragraph, one of the segments of type 2 is used as a reference piece with respect to whichall other segments will we checked in term of radius of curvature. We call it the “Master Segment”.

The CIR specification can be transformed, as an indication to near 30 nm RMS reflected wavefront residual figuringerror and a λ/4 edge turn down over the last 2 mm, if it is present along all the edges. These values correspond to veryhigh level of optical performances for capability of diffraction limited imagery in vacuum. In clear, the GTC telescopewill never be limited in optical resolution by its primary mirror segments.

The above specifications combined with the schedule constraints, lead to seriously approach the requirements for thecoming generation of Extremely Large Telescopes which are :

Accurate profile generation into a rough blank and associate metrologyRapid polishing without creation of irreversible edge effectAccuate figuring down to nanometer level for reaching the final specificationPrecise radius, conic constant, off axis distance, figure error and edge residual evaluation

The notion of speed and cost-effectiveness of the processes being developed is crucial for meeting the schedulerequirements which will be even more stringent for the ELT’s. In this sense SAGEM has developed a set of worldunique equipment and processes that have now been fully validated in real scale with the successful production of thefirst set of 6 segments for the Gran Telescopio Canarias.

The Master Segment was the path finder during the last two years of activity at SAGEM REOSC and was declared inSeptember 2003, finished and ready to hold its function of long term optical reference through the project.

4. MAIN MANUFACTURING STEPS

The production of the GTC segments has been organized within our giant mirror optical shop in a mode of serialproduction with several units progressively reducing the figure error from several hundreds of µm to the lastnanometers. In the same time the surface roughness and subsurface damage level has to be reduced from the roughmilling state provided by the glass ceramic manufacturer down to fine optical polish and low cosmetic defects. This isdone with new processes developed and equipment set up and validated specifically for the project.

An aspheric generator machine,produces, in a highly stable thermalenvironment, the ground surface ofeach segment to sub µm accuracy withlow sub-surface damage, ready to berapidly polished.

Fine lapping and polishing is donewith a battery of robot systems fittedwith a whole set of specific tools. Theyallow parallel production of smoothlypolished surfaces up to the last edge ina timely manner.

Final figuring to the exact shape isdone with Ion Beam Figuringtechnology (IBF) in a large chamberinstalled within the 8-m optical shopcapable to process components up to2.5-m in size..

The main difficulty of the whole development has been to continuously prevent the apparition of edge defects (turndown or turn up) that always lead to difficult and time consuming corrective runs, when not impossible to do so.

All these equipment represent major capital and human investments in large free form optics manufacturing.

5. MAIN METROLOGY STEPS

Associated to the various manufacturing steps described above, metrology means and procedures have been developedin order to be able to feed the optical machinery with accurate measurement. Again the considerations of productionrate have led to push towards rapidity of measurements without sacrifice in accuracy and density of the data. The sixdifferent types of segments complicate the tasks with an additional requirement for rapid reconfiguration of the testbenches from one segment type to any other.

A large 3D coordinatesmeasuring machine is thefirst tool for acquisition ofrough data of the generatedaspheric shape and bondedpads.

A 2D spherometer, extra-polated from the VLT one isused to generate sub µmaccurate, relatively dense,map of the optical surface,even at the lapping stagewhen specular reflexion isstill not reached.

He-Ne interferometry withCGH corrector is performedat center of curvature at thetop of the VLT, 29-m hightest tower.

Finally, the whole batch ofsegment is assembled on aglobal test bench in order toprecisely evaluate figure andinter-segment radius match.

At the top of the test tower we installed a sophisticated metrology systemcombining the individual segment test tools using carefully designed andmanufactured Computer Generated Holograms (CGH) and the global test meanbased on the use of an Offner type null lens.

Rapid switch from one configuration to theother is provided to the metrology engineerwho can immediately process the acquireddata to determine mirror performancestatus and generate the prescription for thenext figuring run.

Specific care has been taken anddemonstrated to our customer that pupildistortion is taken into account and

corrected for each segment and that the required 5 mm resolution at segmentlevel is obtained.

Also the WFE error budget through the null lens, and more generally, the CIRmeasurement accuracy for full pupil or edge effect has been carefully evaluated.

At the base of the test tower, mirror segment supports have been installed onthe 8-m table of the VLT mirror polishing machine. Up to four segments canbe placed simultaneously on the table for CGH testing.

In parallel, a global support structure for a full set of 6 segments, plus themaster segment, is also installed to conduct the final evaluation ofperformances, especially the inter-segment radius error of 0.1 mm. Eachsegment, installed on its 36 wiffle tree support device delivered by GTC, canbe moved along the 6 degrees of freedom in order to precisely place them attheir theoretical transverse position, to phase them with an accuracy betterthan 10 µm and orient them precisely to acquire individual or global fringepatterns through the individual CGH or Offner corrector lens.

The photo at left shows the arrangement viewed from the top of the tower.

6. THE MASTER SEGMENT

In September 2003 the master segment passed successfully the entire cycle of manufacturing and testing anddemonstrated that the various difficulties have been overcome. Its absolute certification has been performed todetermine its actual off axis distance and residual figure errors. This has been done according to an special procedurecombining several measurements.

A summary of its performances of the master segment is :Radius of curvature : 32 995,2 ± 1,5 mmConic Constant : -1.00197 ± 0,0001Surface figure errors : 17 nm RMSCIR (global WFE) : 0,946 (overall figure contribution)CIR (edge effect) : 0,955 (edge effect contribution)CIR (total) : 0,903 (product of the above values)Active support : not used for the Master Segment.

The master segment appears then as a component of very high quality with very smooth residual figure errors as shownon the interferogram at right obtained synthetically from the phase map at left resulting from the processing of themany individual measurements involved in the metrology process. To our knowledge such a quality of an optics of thisshape and this nature has never been obtained through the world and constitute a major achievement of SAGEM-REOSC.

7. EDGE EFFECT EVALUATION

As explained above, edge effects are a serious contributor to light diffusion around the central sport of the imagedelivered by a segmented aperture telescope. This would make much more difficult the detection of faint objects, i.e.exoplanets, close to bright stars. Knowing that such type of discovery is one the main motivation to construct 10-m andmore telescope, it is therefore clear that great attention shall be given to such edge effect.

To evaluate it, SAGEM-REOSC has set up additional metrology means. This consists in a 200 mm diametersubaperture interferometer moved on 12 locations around each segment periphery. The obtained interferograms areprocessed to remove their low order content corresponding to the aspheric shape and figure segment errors evaluatedthrough the full aperture measurement. Some internal and external zones are removed due to gauge residual figureerrors. Finally, two edge portions are processed to determine an average edge profile for the segment. The average edgeeffect amounts to 50 nm wavefront error only up to a few mm from the edge. Worst cases remain below 200 nm. Fromthese data, calculation of the CIR degradationis done according to various simulations performed by GTC.

8. PERFORMANCE OF THE VARIOUS BATCHES DELIVERED

In December 2005, the last batch of segment was delivered fully within specification. A summary table of the figureerror, the most easily understandable performance criterion within the optical community, for the various batches ofsegments produced :

Batch # Date Figure error1 2004-03 16.0 nm RMS2 2004-07 13.9 nm RMS3 2004-11 13.5 nm RMS4 2005-03 12.5 nm RMS5 2005-06 10.5 nm RMS6 2005-09 11.0 nm RMS7 2005-12 12.0 nm RMS

The average performance over the full aperture is 13nm RMS, corresponding to a state of the artmonolithic primary of a 4 to 8-m class telescope.

The interferogram at right shows the outstandingperformance level reached by this primary mirror with no noticeable edgeeffect along the segment’s contour.

After the period of process development, the total optical area, amountingto 2.6 times the one of the James Webb Space Telescope (JWST), has beenpolished in less that 2 years.

The inter segment radius error is below 0.05 mm over the 33 metersbaseline radius of curvature.

Active forces were rarely used to improve segment performance, andmainly for the first batches, within the 5 N.m allowed range of torques.

Batch 1

The team and some segments

Reconstructed fringesfor batches 2-7 and 1

( in Actve Mode)

9. SECONDARY MIRROR

The Secondary Mirror design has been described in several other papers and wewill not present this here but only indicate the final performance results obtainedfor this mirror.

These are summarized as follow :

Item Spec Tol PerformanceRadius 3899.7 mm +/- 5 mm 3901.85 +/- 0.22 mmConic const -1.50484 +/- 0.001 -1.50420 +/- 0.0008WFE passive < 381 nm RMS NA 74 nm RMSCIR active > 0.976 NA 0.9835WFE active per CIR NA 8.9 nm RMSMass 46.4 kg +/- 2 kg 45.98 kg +/- 0.02 kg

Again, this mirror has reached a level of very high quality thanks to optimizedpolishing and figuring and metrology technology.

The resulting wavefront is well smooth as shown on the reconstructed interferogramcorresponding to the active mode performance, i.e. 8.9 nm RMS.

Edge effect has been subject to particular attention as eachmillimeter lost on its edge contour will mean loosing 10 mmon the primary mirror. Some zoomed interferogram like theone shown at left have been used to assess the mirror qualityup to the very edge.

The mirror has been carefully mounted on its support triangle that interfaces with thechopping mechanism. The photo below shows the mirror from its rear face, in thehand of Javier CASTRO leader of GTC optics team.

10. CONCLUSION

SAGEM-REOSC is very proud to have contributed to leverage the state of the art of optical manufacturing technologythrough its successful delivery of the 42 primary mirror segments and the lightweight secondary mirror to the GranTelescopio Canarias project. This is the result of major effort and dedication of all our team of technicians andengineers over quite 5 years and major capital investment in new and innovative optical manufacturing process,production machinery and metrology tools.

10 years after VLT and GEMINI 8-m monolithic mirrors, a significant step has been made with the first ultra highquality large segmented aspheric optics delivered for GRANTECAN. The lessons learned, and there are numerous, willbe more than useful to assess and prepare the next steps in giant telescopes of 30 to 50 meter primary mirror aperture.

SAGEM wants to thank its customer GRANTECAN SA for the confidence placed in the company and thank all theteam that contributed with courage and dedication to the success of this very challenging project. A special thank isaddressed to Javier CASTRO who contributed greatly to our success with its patience, acute mind and energy to push usbeyond our limit.

GTC M2 on its metrology plate

GTC M2 OPDZoom on one edge

Mirror integrated