Optics for the Queen Elizabeth 11 Telescope

G. J. Odgers

Details are given of the design considerations for the mounting, finishing, and alignment of theoptics for the Queen Elizabeth II telescope. Optical tolerances and test procedures are also discussed.

1. Introduction

Detailed engineering design studies for a large reflect-ing telescope to be built in Canada began in January1966. Previously a mirror of approximately 4-maperture had been ordered from Corning Glass Worksin October 1965 and completion of this mirror-theheaviest ever made-is expected in September of thisyear. An optical shop having a 26-m test tower will bebuilt at the University of British Columbia and themirror, together with the telescope secondary mirrors,will be ground and polished there; it is expected thatthe process of mirror finishing to the tolerances requiredwill take four to five years.

The technological impetus for the large telescopeproject was due to the development in the last fewyears of quartz mirrors of considerable size by both theGeneral Electric Company and Corning Glass Works inthe U.S.A. The 200-in. (5.08-m) mirror at Palomarand the Lick Observatory 120-in. (3.05-m) mirror weremade of Pyrex over thirty years ago and since then nolarge Pyrex mirrors have been constructed outside theSoviet Union, and no commercial company was willingto attempt one. Thus, the use of gaseous phases ofsilicon by Corning and the natural quartz fusing meth-ods of General Electric made it possible for astronomersto consider large mirrors again. This fact has alreadybeen taken advantage of by three groups, namely, theKitt Peak National Observatory (KPNO), Tucson,Arizona, The European Southern Observatory (ESO)-representing six European countries and whose tele-scope is to be erected in Chile-and our own. Sincequartz has less than one-fifth the coefficient of expan-sion of Pyrex, it is especially suitable for Canadianconditions where considerable temperature changes canbe expected.

The author is with the Dominion Astrophysical Observatory,R.R.7, Victoria, B.C.

Received 29 May 1967.

II. Optical Parameters

Advances in the art of figuring large mirror surfaceshave made short focal lengths possible-with conse-quent reduction in the cost of the telescope dome-andall three groups currently designing large telescopeshave chosen similar optical parameters, even thoughthese impose severe difficulties on the opticians and canonly be achieved under modern optical shop conditions.The primary mirror is to have an aperture ratio of f/2.8which implies a high degree of asphericity, and theprimary and secondary mirrors together form a Ritchey-Chr6tien* optical system of aperture ratio f/8 givinga field of good definition of approximately 45 min of arcat the Cassegrain focus, with no coma or sphericalaberration. Such an optical system requires thatprimary and secondary mirrors are very nearly hyper-boloidal in shape and the uncorrected field at the primefocus is very small; hence, prime focus correctors mustbe provided to make the usable field of good definitionat the prime focus approximately 10. The use of theRitchey-Chr6tien system depended largely on the possi-bility of successfully designing such correctors and greatattention was paid to this problem both in Europe andthe U.S.A. and several such systems have been pro-posed.

The Queen Elizabeth II telescope is to have a coud6focus at about f/30 with a horizontal coud6 spectro-graph using a 61-cm beam and allowing for the possi-bility of 61-cm gratings being available within a decade;this implies a distance of about 18.3 m between the slitof the coud6 spectrograph and the collimator.4

Ill. OpticsThe optical system is a pure Ritchey-Chrdtien system

and hence involves primary and secondary mirrors withlarge departures from sphericity, especially the f/8

* This aspheric telescope system is based on work of Schwarz-schild,' Chr6tien,2 and Ritchey.3 Much recent development hasbeen done by D. H. Schulte of the Kitt Peak National Observa-tory using the spot diagram method.

October 1967 / Vol. 6, No. 10 / APPLIED OPTICS 1635

secondary, since both mirrors are very nearly hyper-boloids. However, the eccentricity of the primarymirror is nearly unity, whereas that of the secondarymirror is greater than two; both mirrors will require theutmost perfection in optical finishing. The deviationsof the primary and secondary mirror surfaces from aspherical surface furnish a measure of the difficulty inachieving such surfaces; in the case of the primarymirror, the deviations amount to nearly 300 wave-lengths and for the secondary mirror nearly 150 wave-lengths. Since the mean wave aberrations of thefinished surfaces must be less than one-twentieth of awavelength, the precision required in optical finishingcan be appreciated.

A. Cassegrain Focus

If the Ritchey-Chr6tien condition can be satisfied,a large field will be provided, approximately 45 min ofarc at this focus, free from coma and spherical aberra-tion, but with small residual astigmatism. Fused silicadoublet lenses have been designed which when placeda short distance from the focus (approximately 18 cm)remove the astigmatism and provide a flat field about30.5 cm in diameter.

B. Prime FocusThe prototype of Ritchey-Chr6tien type telescopes is

the 213-cm at KPNO which demonstrated that thedesign was successful at Cassegrain and coud6 foci,but which had no prime focus function. Before thebasic optical design was adopted for very large tele-scopes having a prime focus cage, difficult opticalproblems had to be solved to see whether correctorscould be designed to give a usable field at this focus,since the uncorrected field of the primary is very small.Various solutions of this problem have been proposedby optical designers in Germany, France, England, andthe U.S., involving optical systems of different degrees ofcomplexity. A moderately simple system which seemslikely to produce good results has been designed byC. G. Wynne at the Imperial College, London.This involves only spherical surfaces for the correct-ing lenses with six air-glass surfaces, the back lensbeing a cemented triplet. Such a system producesimages of less than 0.25 see of arc over a field of 30 minof arc and less than 0.5 see of arc over nearly a 1 field.However, the triplet lens requires the use of dense flintin the center and hence there would be some loss oftransmission in the uv. These lenses are fairly small(approximately IS cm in diameter) and should notpresent any special difficulties of manufacture. A morecomplicated system has been designed by H. Khler atCarl Zeiss which uses three aspheric correctors of muchgreater size (approximately 60 cm in diameter) involv-ing surfaces of the sixth and eighth orders; to makesuch a corrector system would involve considerable diffi-culties in manufacture, and testing procedures wouldrequire making null systems of some complexity.This system would give very good results over a widefield; both systems are subject to a certain amount ofvignetting at the larger field angles.

However, it appears that a large Ritchey-Chretientelescope can be provided with a good field at the primefocus by means of fairly complex but attainable correc-tor systems. It might happen that several of thesewill be needed in the future to cover different wave-length ranges and with different transmission and fieldproperties.

C. Coud6 Focus

The coudd secondary mirror (approximately f/30system) must be figured so that it corrects the sphericalaberration of the primary mirror; this should notpresent any difficulty and will provide a small fieldof about 4 min of arc diam adequately correct forcoma and astigmatism at the coud6 focus.

The three flat mirrors in the coud6 system do notpresent new optical problems but need to be adequatelysupported; the small flat mirror requires to be driven ona nonequatorial mount to bring the light to the slit ofthe coud6 spectrograph.

D. Additional Cassegrain Mirrors

Secondary mirrors providing aperture ratios of f/12-f/15 at the Cassegrain focus could be added to thetelescope, if required, for particular observationssuch as photometry. Such a system could not meetthe Ritchey-Chr6tien condition since this has beendesigned for f/8, and the usable field would be ex-tremely small. However, because of the exchange-able upper end feature of the telescope, such mirrors donot affect the general design problem.

E. Meeting Optical RequirementsThe telescope mirrors must be figured to exacting

tolerances and we plan to do this in the optical shopunder construction at the University of British Colum-bia.

If the mean wave aberration of an optical surface iss, the loss of light in the image relative to a perfect(diffraction-limited) optical' system is given by (27r/X)2(52), where X is the wavelength of the light. Hence, ifwe require the light loss to be less than 10%, the meanwave aberration must be held to approximately X/20.The optical tolerances are better expressed in terms ofenergy; in this respect it is required that the primarymirror be figured so that 80% of the light energy fallswith a circle of diameter of 0.5 see of arc at the primefocus and all the light within a circle of 1.2 see of arcdiam.

A tower will be provided in the optical shop so thattests under carefully controlled temperature conditionscan be carried out at the center of curvature of theprimary mirror approximately 21 m above the mirrorsurface. The mirror will be ground to the nearestsphere in order to ensure the absence of astigmatismproduced in the grinding. At this stage the mirror sup-port system will be placed under the mirror and theradial and axial support system tested after which thecell will be removed and the figuring proceed towardthe approximate Ritchey-Chr6tien curve by zonal knife-

1636 APPLIED OPTICS / Vol. 6, No. 10 / October 1967

edge testing and with smoothness of figure measured bymeans of a null lens.6 At the final stages, the mirrorcell will be replaced on the machine and the finalfiguring take place; mirror metrology becomes very im-portant at this point and interferometric methods willbe used at the center of curvature to provide the ulti-mate accuracy required.

Because the Cassegrain and other secondary mirrorsare convex, they will be tested by means of the Hindlesphere-254-cm diam aluminum mirror-which itselfwill be figured using the large 381-cm polishing machinein the optical shop.7 For these mirrors, the possibilityexists that the final stages of figuring could be carriedout by means of ionic polishing methods which are cur-rently being developed and which could provide thehigh precision needed.

It is estimated that the time required to figure the254-cm Hindle sphere, the primary and secondarymirrors-these latter will be figured concurrently withthe primary-will be nearly five years with a furtherperiod of one year for testing in the telescope and finaladjustment of the mirror support system and someslight retouching of the mirrors.

The exact form of the surfaces for the primary andsecondary mirrors have been calculated for a 381-cmprimary mirror, a primary aperture ratio of f/2.S, and asecondary aperture ratio of f/8.0; these surfaces haveequations which can be written in the form: x = y2/2r +a,(y2/2r)2 + a2 (y2/2r)3 + ... , where is the radiusof curvature of the mirror so that r = 4fR and R is theradius of the primary mirror, x is the directed distanceparallel to the optical axis, and y the distance perpendic-ular to the optical axis. The mirror surface is givenwhen the coefficients a,, a2 , etc., are determined. Forexample, a, is approximately equal to 3.4 X 10-6 forthe primary, but since they depend on the exact value ofR they are not given here in detail; for a parabolic sur-face a, = a2 = a = ... = 0. It must be emphasizedstrongly that the final optical design can only takeplace when the mirror size is known, i.e., after comple-tion of the mirror by the Corning Glass Works; if themirror is 396 cm in diameter, r = 2219 cm, whereas ifthe mirror is 381 cm in diameter, r = 2134 cm, and theconstants in the above equation are considerablychanged. The actual size of the primary mirror deter-mines most of the other parameters of the telescopeas well; for instance, tube length, dome size, and mirrorsupport system all depend on it.

IV. Optical TolerancesIn general, the optical alignment shall be governed by

the seeing-limited image size of 0.5 see of arc obtainableat Mount Kobau, the observatory site, a desired opticalsetting accuracy within 4 10 see of arc on the sky, andthe decollimation limits that are established by theoptical instruments used at the foci (such as photometersand spectrographs).

The relative motion of the corrector lens systems orthe secondary mirrors relative to the primary mirror areexpressed by a rotation and a lateral translation. Thisrotation is considered about an axis perpendicular to the

axis of the telescope tube. The relative rotation shallnot exceed 6 see of arc and the lateral translation shallnot exceed 0.51 mm. These motions usually resultfrom relative motions of the primary mirror, or theupper end optics, or both.

The tolerance on the relative rotation of each end ofthe telescope tube about the tube axis are given by therequirements of prime focus photography. The rota-tion of the prime focus plate holder about the tube axisshall be less than 1 min of arc during an exposure. Thisrotation might result from (a) torsion in the tube and,(b) structural deflections of the tube support structure.

The focusing of the telescope is achieved by movingthe corrector in question. The minimum travel incre-ments that are required for any of these focusing mo-tions shall be equal to or less than 0.04 mm (40 L). Thissensitivity is not to be construed as an absolute dimen-sional tolerance. The focusing of the telescope ismonitored visually through the finding telescope andabsolute repeatability on the location of the secondariesor correctors is not necessary.

The distance along the light path from the focus tothe primary mirror shall be held virtually constant.The uniform temperature change causes a movementin the focus of the primary of 5% of the thermal expan-sion of the tube structure. The thermal gradients inthe quartz primary mirror cause a focus shift of about0.25 mm for a sudden 50 C change in the environment.Fortunately, most of this movement occurs during thefirst few hours after the dome shutters are opened, andby the time viewing commences the remaining changewill be only a fraction of the total. In order to cover therange of possible temperature variation, the focusingunit must be capable of a total motion of at least7.5 mm. This range is adequate to suit all secondaryupper ends or prime focus corrector systems that arelikely to be used.

The required setting accuracy of the telescope on thesky, a total of 20 see of arc, with an optical tolerance of10 see of arc, is the criterion that limits the allowablemirror translations and rotations. The decollimationallowed in the photometers and spectrographs of 1 minof arc and about 5 min of arc, respectively, makes in-strument decollimation uncritical in the design of thetelescope.

The allowable rotations and translations of the threeflat mirrors of the coud6 system are:

3rd coude mirror4th coud6 mirror5th coud6 mirror

5 see of arc6.4 see of arc

70 see of arc

0.81 mm0.81 mm0.81 mm.

In determining these standards, the effect of mirrormotions on the telescope setting accuracy and the in-strument decollimation is assumed to be equally sharedbetween the result of mirror rotations and translations.Also, it is assumed that each flat contributes equally tothe setting error and that these errors occur simultane-ously in a direction which gives maximum effect. It isseen from these assumptions that an increase in thetolerances for one mirror can be made when a correspond-

October 1967 / Vol. 6, No. 10 / APPLIED OPTICS 1637

ing reduction in the tolerances for another, or others,occurs.

These tolerances also assume a displacement of thesecondary mirror of 0.51 mm relative to the primarymirror in a direction which minimizes the above transla-tions and rotations. When no 0.51-mm relative dis-placement occurs, the allowable rotations and transla-tions shown in the previous table are multiplied by two,except for the rotation allowances on the fifth mirrorwhich remains at 70 see of arc. This relative lateraldisplacement is almost eliminated in practice by aproper proportioning of the upper end and lower endSerrurier trusses of the telescope tube.

Much of the difficulty in the mechanical design of alarge optical telescope arises from the necessity of align-ment of optical and mechanical axes to extremely smalltolerances and in the support of the mirrors themselves.Methods of support for mirrors which weigh nearlyeighteen tons and whose surfaces should not vary bymore than one millionth of an inch (0.025 ) in any

orientation of the telescope are under intensive in-vestigation at the present time. It can be shown thatinternal frictional forces must be made less than 0.1%if the system is to be satisfactory but a discussion ofhow such problems are solved is outside the scope of thisarticle. However, if even larger optical telescopes areto be built using solid quartz mirrors, the mirror supportproblem will have to be overcome.

References

1. K. Schwarzchild, Gttingen Universitat Steinwarte Astron.Mitt. (1905).

2. H. Chretien, Rev. Opt. 1, 232 (1922).3. A. G. Ingalls, Sci. Am. 147, 20 (1932).4. M. Born and E. Wolf, Principles of Optics (Pergamon Press,

Oxford, 1964), p. 245, 2nd ed.5. Ref. 4, p. 463.6. A. Off ner, Appl. Opt. 2, 153 (1963).7. J. H. Hindle, Monthly Notices Roy. Astron. Soc. 91, 592

(1931).

Meeting Reports continued from page 1596

S. W. Rabideau, and NMR relaxation times of liquid water werereported by J. A. Glasel. The use of 170 as a probe indicatedthat the electronic structure of a water molecule in the liquid wassimilar to that of a molecule in the gas, whereas the use of D indi-cated that it was similar to that of a molecule in the solid.Glasel's measurements of the spin lattice relaxation time wereconsistent with single molecules that relax as classical spheres, andso provided little information about the fundamental processes inwater. J. C. Hindman reported the chemical shift of liquid waterat various temperatures relative to vapor and compared his re-sults with various models. Agreement could be obtained withmany, including the bent hydrogen-bond model.

A number of papers reported the effect of salt on various prop-erties of water. The results in general were even less susceptibleto a precise analysis than the properties of the pure liquid. Thefar ir spectrum was reported by D. A. Draegert, and low fre-quency Raman spectra by L. A. Blatz and P. Waldstein. Largeconcentrations of ions caused surprisingly little change in thespectrum; presumably the ions were to a large extent beingcarried along with the motion of the water molecules. T. T. Wallreported many measurements of the Raman spectrum in the O-Hstretching region and reached the interesting conclusion, closelyrelated to this, that the hydrogen motion of water of hydrationwas strongly coupled to those of other hydrogens, and did notseparate out.

E. R. Malinowski, P. S. Knapp, and R. 0. Waite showed howsuitably defined hydration numbers for salts could be derivedfrom the proton magnetic resonance chemical shift. G. Atkinsonmeasured ultrasonic absorption in water and salt solutions to 250MHz, and could rationalize the results on a two-state model.The tetrabutylammonium ion caused large relaxations, the exactorigin of which were perhaps not quite settled, but seemed to beconnected with association to ion pairs.

D. E. Irish and A. R. Davis reported on the spectra of aqueousnitrates. The splitting of the degenerate vibrations of the nitrateion was observed, and clearly showed the asymmetric environ-ment of the ion. Perhaps such measurements could probe thelocal electric fields in solutions.

Papers by R. A. Plane and H. L. Friedman were concerned

more with the structure of complex ions in solution than withwater itself.

A lively panel discussion was held on Tuesday evening, withthe main theme what we would like to know and how to measureit. One opinion was that we would perhaps know the molecularmotions in water by calculation in a decade or so. If this is to beaccomplished, we shall need to know a good deal more than we doabout the interaction of water molecules. Liquids have so farproved relatively unfruitful sources of this kind of knowledge.Solids are potentially more useful because of the limited configura-tions that can occur, and D. Eisenberg and M. Levine presented asymposium paper on the interaction of water molecules, emphasiz-ing the importance of the high pressure phases. Perhaps thegeneral consensus of the discussion was that techniques that tellabout the various correlation functions of the molecules in theliquid would be of most fundamental significance.

In scientific research, an important activity is the giving ofnames to concepts. Two that are frequently used in discussingaqueous solutions are structure making and structure breaking.When they were first invented by H. S. Frank they served to focusattention on the quite different things that different solutes do towater. It seemed at the symposium that perhaps they comeeasily to the tongue and perhaps now, having served their pur-pose, they tend more to obscure what is not known about solutionsthan to illuminate what is known.

It seems clear from the work reported at the Symposium thatthe real increase in our understanding of the molecular arrange-ments and motions in liquid water has not gone up in proportionto the effort expended over the past two or three decades. Ther-modynamic measurements, in conjunction with the appropriatetheory, have too little resolution as yet to be very valuable. Thiswill perhaps change when more accurate theories of the liquid areworked out, but none seems imminent. It appears likely that inthe near future our greatest information will come from the ab-sorption of light by nuclear translational motions, Raman andhyper-Raman scattering, elastic scattering of x rays, and inelasticscattering of neutrons. It seems however that much theory willbe required as well as many more well chosen experiments beforethe harvest is reaped. At the present time in most aspects ex-periment is well ahead of theory, as indeed it is for liquids in gen-eral.

1638 APPLIED OPTICS / Vol. 6, No. 10 / October 1967