Design of the Extreme Ultraviolet Explorer long-wavelengthgrazing incidence telescope optics

David S. Finley, Patrick Jelinsky, Stuart Bowyer, and Roger F. Malina

Designing optics for photometry in the long-wavelength portion of the EUV spectrum (400-900 A) posesdifferent problems from those arising for optics, operating shortward of 400 A. The available filter materialswhich transmit radiation longward of 400 A are also highly transparent at wavelengths shortward of 100 A.Conventional EUV optics, with grazing angles of S 100, have very high throughput in the EUV, which persiststo wavelengths shortward of 100 A. Use of such optics with the longer-wavelength EUV filters thus results inan unacceptably large soft x-ray leak. We have overcome this problem by developing a mirror design withlarger graze angles 220°, which has high throughput at wavelengths longer than 400 A but at the same timevery little throughput shortward of 100 A.

1. Introduction

Instrumentation for the Extreme Ultraviolet Ex-plorer satellites1 2 is currently being built by the SpaceAstrophysics Group at the University of California,Berkeley. Four grazing incidence telescopes will beincluded in the payload. Three of these will be scan-ner telescopes pointed in the spacecraft spin plane;during the first six months of the satellite's operationthese scanner instruments will carry out a photometricall-sky survey in different bandpasses from 80 to 900 A.The fourth, the Deep Survey/Spectrometer telescope3

will be pointed in the antisun direction normal to theother three telescopes. The Deep Survey/Spectrome-ter will concurrently perform a photometric deep sur-vey along half of the ecliptic plane during the surveyphase of the mission and will then be used for spectro-scopic observations during the second six months ofthe mission.

The design of mirrors for use in EUV photometry iscritically dependent on the transmission properties ofthe filters which are used to define the EUV band-passes. The combination of filters and mirrors mustbe selected so that high sensitivity is achieved fordetection of faint sources. Second, the combination ofmirror and filter must yield a well-defined bandpass,

The authors are with University of California, Berkeley, SpaceSciences Laboratory, Berkeley, California 94720.

Received 21 September 1987.0003-6935/88/081476-05$02.00/0.© 1988 Optical Society of America.

i.e., one which is restricted to the desired wavelengthrange.

A number of materials and material combinationswere developed and tested in our search for thosefilters which are optimal for performing an EUV sur-vey.47 Our filter development program included test-ing of samples of filter material combinations with anappropriate range of thicknesses, measurement oftheir transmission over a wide range of wavelengths,and calculation of the linear absorption coefficients.In addition, we studied the attenuation factors foroxide layers and any intermediate layers which arise ifdifferent materials interact. These filter parameters,determined experimentally, were used as the basis ofdetailed studies to maximize the sensitivity of the in-struments. The four filters currently baselined for theEUVE mission and their bandpasses are Lexan/boron

Lexan/B, 60-150 A), aluminum/boron (Al/B, 170-300A), titanium/antimony/titanium (Ti/Sb/Ti, 400-650A), and tin (Sn, 520-850 A).

The mirrors designed for use with the short-wave-length filters (Lexan/B and Al/B) are typical of mir-rors used in the EUV. Rather than the extremelyshallow graze angles typical of x-ray applications,these mirrors of Wolter-Schwarzschild Type I designhave 5 graze angles as shown in Fig. 1. Graze anglesof this magnitude maximize the throughput through-out the EUV. With a gold coating, the mirrors havehigh throughput down to wavelengths shortward of100 A. The total geometric collecting area of thesemirrors is 140 cm2. The effective area for these short-wavelength Type I scanner mirrors, calculated fromthe optical constants for gold,8 9 is shown in Fig. 2.

Given the desirability of using identical mirrors forall three scanners, we investigated the possibility of

1476 APPLIED OPTICS / Vol. 27, No. 8 / 15 April 1988

MOUNTINGSURFACE

27.9 CM.

Fig. 1. EUVE Type I scanner mirror schematic diagram (sectionview). Light enters from the left and is brought into focus on the

right.

S

0

aoID

gW

100

80

60

40

20

0 200 400 600 800Wavelength ()

Fig. 2. Gold-plated Type I scanner mirror effective area as a func-tion of wavelength, computed from optical constants.

using the long-wavelength filters with Type I scannermirrors. Performing optimization studies for thatcombination resulted in the instrumental effective ar-eas shown in Fig. 3. As can be seen, the materials arehighly transmissive shortward of 100 A, where theeffective area of the gold-plated Type I scanner mirroris also quite substantial. The high short-wavelengththroughput is due to the fact that these filters are evenmore transparent in the soft x-ray region than in theirmain bandpasses. Therefore, if an object were to bedetected in one of the long-wavelength bandpasses, itwould be difficult to rule out the possibility that thedetection was due to soft x-ray flux. Our conclusionwas that satisfactory bandpasses could not be achievedwith the available long-wavelength filter materialswhen used in conjunction with mirrors which wereoptimized for short-wavelength response.

II. Alternative Mirror Design

After investigating a variety of possible solutions tothe soft x-ray leak problem, we conculded that the

1.5

Ef

>

4!

Wi

.5

0 0 20 I I I I 60I 5 0

0 200 400 600 800Wavelength (A)

Fig. 3. Type I scanner instrumental effective area vs wavelengthfor a titanium/antimony filter (solid curve). The primary bandpassis -400-650 A. Also plotted is the effective area for a tin filter

(dotted curve). The primary bandpass is -520-850 A.

optimal solution was to employ a different mirror forthe long-wavelength bandpasses. The replacementmirror was required to have the same physical inter-face and also fit in the same physical envelope as theother scanner mirrors. The additional requirementwas that it had to reject soft x-rays by a factor suffi-cient to eliminate possible confusion of EUV and softx-ray sources. An acceptable rejection factor was de-termined by examining the response of the instrumentto expected sources of EUV radiation, such as coronalsources and hot white dwarfs. A possible worst-casescenario was chosen as a nearby very hot DA whitedwarf with an effective temperature in the neighbor-hood of 60,000 K for which the interstellar hydrogenattenuation was sufficiently strong that the star wasjust detectable in the total bandpass (consisting of theprimary bandpass and soft x-ray leak). The criterionapplied was that, at the detectability threshold, nomore than 20% of the total counts received from anobject could be due to flux detected in the soft x-rayleak.

To achieve this goal, we found that it was necessaryto have

Aeff(500 A)> 30,

Aeff(100 A)

where Aeff is defined as the product of the mirrorgeometric area and reflectivity. This effective arearatio implies graze angles of >250. The mirror typesthat have acceptable imaging over a wide field of vieware the Wolter-Schwarzschild Type I and Type II con-figurations. However, Type I mirrors must be ruledout for this application. At graze angles of 25°, thetwo internal bounces would result in the light from themirror approaching the focus at angles which are near-ly normal to the optical axis and would not strike thedetector at acceptable angles.

The defining parameters of Wolter-SchwarzschildType II mirrors can be adjusted to produce mirrorswhich fit in the given envelope and have any desiredincidence angles. Consequently, the potential mirrorcandidates which could be used in conjunction withthe long-wavelength filters included the full range ofpossible Wolter-Schwarzschild Type II mirrors, fromextreme grazing to normal incidence. To produce the

15 April 1988 / Vol. 27, No. 8 / APPLIED OPTICS 1477

initial set of mirrors we imposed specific overall con-straints including a maximum outer diameter, a maxi-mum front-to-focus length, and a minimum separationbetween the rear of the mirror and the focal plane toallow room for the detector and associated mecha-nisms.

For each candidate mirror meeting these con-straints, we determined the location of the mirror sur-face end points and calculated the average effectiveareas at 100 and 500 A and the ratio of those areas.The following criteria were also considered in evaluat-ing the acceptability of candidate mirrors for furtherstudy:

(1) Aeff(500 A)/Aef(100 A) > 30.(2) It had to be possible to baffle the mirror to

exclude direct viewing of the sky by the detector.(3) Aeff(500 A) had to be within 70% of the maxi-

mum attainable for each geometric area studied.(4) The mirror had to be compatible with the struc-

tural elements designed for the shorter wavelengthscanner telescopes.

(5) The secondary surface inner radius had to be>3.2 cm.

As has been noted, the first criterion allows for thereduction of the soft x-ray response to an acceptablefraction of the EUV response. The second criterionrequires the exclusion of light which is not actuallyfocused by the optical elements, thus reducing thestray light background at the detector. In the EUV,background suppression is especially important, sincethere is strong diffuse background at 304, 584, 834, and1216 A. In our investigations, we discovered that, ifeven a solid angle of only 10-2 sr is directly viewed, thesensitivity will be degraded by a factor of 3. The thirdcriterion eliminated the less sensitive mirrors fromfurther consideration. [The sensitivity varies linearlywith Aeff(500 A)]. The fourth criterion was necessaryto minimize the programmatic impact of having a dif-ferent mirror for the long-wavelength scanner tele-scope; it did not eliminate any of the best candidates.The final criterion was set in order that a magneticbroom (for charged particle rejection) could be de-signed which would fit in the available space.

Mirrors with the smallest focal lengths and largestAeff(500 A), selected by these criteria, were then raytraced to determine their imaging properties. Sincethe sources will be viewed in all parts of the field ofview in a survey mode, the sensitivity depends on theaverage pixel size over the entire field of view. Accord-ingly, the average size of the end-to-end instrumentresolution elements (pixels) over the field of view wascalculated, using an assumed aspect blur and detectorresolution. The mirrors do not have flat fields but areused with flat detectors. Therefore the average pixelsize is minimized when the detectors are moved for-ward from the on-axis focal plane. Sensitivity im-provements of factors of 1.5 or greater are achievableby such defocusing. Our sensitivity calculations in-cluded optimal defocusing of the detector. Theachievable instrument sensitivity with the optimumfilter thickness for each mirror was then calculated; it

2

0

-.

u0

Q

:s

40 60 80 100 120 140 160Focal Length (cm)

Fig. 4. Minimum detectable flux (MDF) of mirror candidates,including a Ti/Sb filter, plotted vs focal length. MDF is in photonscm-2 s- Al, assuming a flat incident photon spectrum. The trian-gle represents a normal incidence mirror. The square represents aType I scanner mirror. The filled circle represents a Type II scan-

ner mirror after optimization.

is expressed in terms of the minimum detectable flux(MDF) in photons cm 2 s-1 A-', assuming an incidentflat photon spectrum; the 5 detection above back-ground was selected as the threshold. Figure 4 showsthe resultant MDFs computed for each of the gold-coated mirrors with the Ti/Sb filters. For comparison,we have included the MDF for a typical normal inci-dence mirror, which is ii ,dium coated for optimal re-flectivity; its sensitivity is considerably worse thanthat of the mirrors which are closer to grazing inci-dence. The poor normal incidence performance is dueto the considerably smaller Aeff(500 A) which is achiev-able at normal incidence, even though the normal inci-dence mirror fills the available aperture.

As a final step, preliminary physical designs for themirrors with the greatest sensitivities were produced.Fortunately the most favorable physical design wasachieved for the mirror which also provided the high-est sensitivity. In the course of performing subse-quent detailed physical design, slight modificationswere made to the end points of the mirror surface usedin the initial search. An additional sensitivity gain of-10% was thereby achieved by increasing the effectivearea.

Secondary criteria applied in achieving the finalmirror design were that no direct light could reach thedetector, and ghost images must be eliminated. Theinstrument layout that achieved these goals is shown inFig. 5. The optical elements shown are the front aper-ture/field stop, the mirror, the rear baffle, and thedetector. The front element is sufficiently far fromthe mirror that it serves both as an aperture stop and afield stop.

The locations of the different optical elements weredetermined in an ordered progression. The elementswhich were predetermined were the front apertureouter edge and the detector. The size of the detectorand the focal length set the maximum acceptance an-gle for the instrument, 0max, defined by 6max = tan- 1

(Rdet/f), where Rdet is the detector radius and f is thefocal length. For these values, max= 2.0450. Giventhe front aperture outer radius, the location of thefront edge of the primary surface was set by the constraint

1478 APPLIED OPTICS / Vol. 27, No. 8 / 15 April 1988

. . _

.8 t I

.6 - * ' * l l l l l l l l l l l l

J

Front Aperture

Mirror

DetectorFig. 5. Instrument configuration showing optical elements. Raypaths shown are for the extreme rays which define baffle locations

and mirror surface end points.

that light entering the aperture at the maximum ac-ceptance angle should not be vignetted by the primary.The geometry of this mirror necessitated the place-ment of a baffle within the mirror to block light whichcould otherwise pass directly through the mirror with-out reflection directly onto the detector. The optimallocation for such a baffle was very near the front edgeof the secondary mirror surface. The design was sim-plified by simply extending the secondary front edgeforward to the appropriate location. This extensionhad the added benefit of reducing off-axis vignettingdue to the loss of photons at the front of the secondary.Given the location of the front edge of the secondary,the inner edge of the front aperture was then set so thatit did not obscure any light which would otherwisereach the primary at angles <0max. The rears of theprimary and secondary were then set to eliminate off-axis vignetting. Once the active areas of the mirrorsurfaces were determined, the physical ends of thesurfaces were chosen to be 3 mm beyond the activeedges, for all edges except the front edge of the second-ary. This 3-mm extension prevents rollover of theedges during polishing from affecting the mirror imag-ing performance. The rear baffle axial location waschosen to be the same as for the short-wavelengthscanner. The radii of the edges were then set to allowall light from the secondary to reach the detector. Theoff-axis performance of the telescope was verified byray tracing simulated point sources at a number of off-axis angles. The ray trace showed that direct light andghost images are excluded at all angles.

The adopted long-wavelength Type II scanner mir-ror assembly is shown in Fig. 6. The mirror has a focallength of 70 cm, an outer aperture of 20.107 cm, and aninner aperture of 16.571 cm. The resultant geometriccollecting area is 407.4 cm2. The primary length is-7.9 cm, and the secondary length is -7.5 cm. The

Fig. 6. EUVE Type II scanner mirror schematic diagram (sectionview). Light enters from the left and is brought into focus on the

right.

2

'6

Ma

Wt

9

:3

2

00 .5 1 1.5

Off-axis Angle (degrees)2

Fig. 7. Type II scanner mirror imaging performance. Intrinsic rmsblur radius, as a function of off-axis angle for a mirror alone, in plane

of best on-axis focus.

overall length is -10.8 cm. The average graze angleson the primary are -29°, while the secondary grazeangles average -21°.

The mirror imaging performance is shown in Fig. 7,which gives the intrinsic rms mirror blur as a functionof off-axis angle. Fabrication errors are not includedin the blur, but these are expected to be small; thefigures of the EUVE mirrors which have already beenfigured and polished are such that their resolution (i.e.,the on-axis full width at half-maximum of the pointspread function) is <3 sec of arc. (An imaging mea-sure which is more relevant for photometric observa-tions is the half-energy width, the diameter into whichhalf of the reflected energy from the mirror falls. Pri-mary or EUVE mirrors have achieved half-energywidths in the visible of <10 sec of arc.) Because theincident angles are much greater for the Type II mirrorthan for the Type I scanner mirrors, the intrinsic blurfor the Type II mirror is a factor of 3 smaller than forthe Type I scanner mirrors.

15 April 1988 / Vol. 27, No. 8 / APPLIED OPTICS 1479

80

8E

C)

60

40

20

0 I . . I . . . I . 1, . I . . . I 1 I

0 200 400 600 800Wavelength (A)Fig. 8. Gold-plated Type II scanner mirror effective area vs wave-length calculated from optical constants. Throughput at 100 A is

virtually negligible.

.5

0

a

3)

.4

.3

.2

.1

00 200 400 600 800

Wavelength (A)

Fig. 9. Type II scanner instrumental effective area vs wavelengthfor a titanium/antimony filter (solid curve), and for a tin filter(dotted curve); includes mirror reflectivity, filter transmission, and

detector QE.

In Fig. 8 the effective area of the mirror is shown.The mirror throughput < 100 A is highly attenuatedrelative to the throughput at longer wavelengths.This mirror was selected over all others because it hasthe highest possible throughput longward of 400 A[subject to the Aeff(500 A)/Aeff(100 A) constraint]. Itshould be noted, though, that the mirror selected hasless effective area than the short-wavelength scannermirror in the wavelength range >400 A. The decreasein effective area occurs because the increased grazeangles result in reflectivity losses which are not offsetby the gain in geometric collecting area.

When the mirror is used in conjunction with theselected long-wavelength filters described in Sec. I, theresulting bandpasses achieved are as shown in Fig. 9.The virtual absence of a significant soft x-ray leakshould be noted. For the Ti/Sb/Ti filter, the soft x-raygrasp is -0.1% of the total grasp and is -0.03%for theSn filter. [Here we define grasp as S Aeff(X)dX, whereAeff(X) is the instrumental effective area as a functionof wavelength.] The resultant sensitivity, comparedwith that obtained for these filters when used in con-junction with the Type I scanners, is a factor of 2 betterfor the Sn filter, and 10% better for the Ti/Sb filter.The soft x-ray sensitivity, which could compromise theobservations of EUV sources, is negligible for bothfilters when used with the Type II mirror.

I11. Conclusions

Given the available physical envelope for the EUVEinstruments, the Type I scanner mirrors describedabove are the optimum for an EUV sky survey atwavelengths as short as 100 A. However, such mirrorscannot be used to provide photometric bandpasses atwavelengths >400 A, because the available filter mate-rials are highly transparent to soft x-ray radiationshortward of -100 A. We have performed an exhaus-tive study of optics which are appropriate for use withthese long-wavelength EUV filter materials. Our con-clusion is that the optimum mirror configuration foruse in a photometric instrument operating in the long-wavelength end of the EUV spectrum (400-912 A) is aWolter-Schwarzschild Type II mirror which has inci-dent angles intermediate between grazing and normalincidence.

We thank Vincent Saba for his assistance in raytracing the EUVE mirror systems. We are grateful toJudith Goldsmith and Aileen Corelli for editorial assis-tance in preparing this paper. This work was per-formed under NASA grant NAS5-29298.

This material was presented as paper 830-15 at theConference on Grazing Incidence Optics for Astro-nomical and Laboratory Applications, sponsored bySPIE, the International Society for Optical Engineer-ing, 17-19 Aug. 1987, San Diego, CA.

References1. S. Bowyer, "The Extreme Ultraviolet Explorer," Adv. Space Res.

6, 153 (1986).2. R. F. Malina, S. Bowyer, M. Lampton, D. Finley, F. Paresce, G.

Penegor, and H. Heetderks, "Extreme Ultraviolet Explorer,"Opt. Eng. 21, 764 (1982).

3. M. C. Hettrick, S. Bowyer, R. F. Malina, C. Martin, and S.Mrowka, "Extreme Ultraviolet Explorer Spectrometer," Appl.Opt. 24, 1737 (1985).

4. S. Chakrabarti, S. Bowyer, F. Paresce, J. B. Franke, and A. B.Christensen, "Long Term Variability of Transmission of ThinIn-Sn and Sn-C Films for EUV Instrumentation," Appl. Opt. 21,3417 (1982).

5. P. Jelinsky, C. Martin, R. Kimble, S. Bowyer, and G. Steele,"Composite Thin-Foil Bandpass Filter for EUV Astronomy: Ti-tanium-Antimony-Titanium," Appl. Opt. 22, 1227 (1983).

6. S. Labov, S. Bowyer, and G. Steele, "Boron and Silicon: Filters forthe Extreme Ultraviolet," Appl. Opt. 24, 576 (1985).

7. J. V. Vallerga, 0. H. W. Siegmund, P. Jelinsky, and M. Hurwitz,"The Calibration of Thin Film Filters to be Used on the ExtremeUltraviolet Explorer Satellite," Proc. Soc. Photo-Opt. Instrum.Eng. 689, 138 (1986).

8. B. L. Henke, P. Lee, T. J. Tanaka, R. L. Shimabukuro, and B. K.Fujikawa, "Low Energy X-Ray Interaction Coefficients: Pho-toabsorption, Scattering, and Reflection," At. Data Nucl. DataTables 27, 1 (1982).

9. L. R. Canfield, G. Hass, and W. R. Hunter, "The Optical Proper-ties of Evaporated Gold in the Vacuum Ultraviolet from 300 A to2000 A," J. Phys. 25,124 (1964).

1480 APPLIED OPTICS / Vol. 27, No. 8 / 15 April 1988