Документ взят из кэша поисковой машины. Адрес оригинального документа : http://hea-www.harvard.edu/~elvis/40kevimaging1988.pdf Дата изменения: Thu Apr 14 01:07:32 2011 Дата индексирования: Mon Oct 1 21:49:28 2012 Кодировка: Поисковые слова: m 35

Grazing incidence imaging from 10 to 40 keV
Martin Elvis, Daniel G. Fabricant, and Paul Gorenstein

The prospects for imaging x rays at energies from 10 to 40 keV with grazing incidence optics are explored.

The scientific rationale and existing laboratory measurements are reviewed. Measurements of reflectivity
using possible mirror materials are described. Iridium-coated float glass gives an improved performance over gold by the factor predicted by theory but both had a lower absolute level. This may be due to a lower density

of the thin metal layer caused by the deposition method. The reflectivity of a sample of iridium-coated float
glass was measured at small grazing angles (25-5 min of arc) at energies of 8,17, and 26 keV. High reflectivity (>50%) was seen out to angles of 33,16, and 11 min of arc, respectively. These are close to the theoretical values. A design for a high energy imaging telescope of the Explorer class is described.

1.

Introduction

It is clear now that the advent of grazing incidence optics in x-ray astronomy has transformed the subject into one of the major disciplines of astrophysics. The Einstein Observatory (HEAO-2) covered the energy range from -0.1 to -3.5 keV and this is still the highest energy to which grazing incidence mirrors have been used in a mission. The more recent EXOSAT and ROSAT mirrors both cut off above -2 keV. X-ray
imaging up to -10 keV is expected only when AXAF

threshold of 1/75 Crab, some 1000 times higher than that of the Einstein Observatory, is set by the high background count rate. Table I lists some of the scientific prospects for the 10-40-keV energy band if sufficient sensitivity were available. They cover a wide

range of topics in many branches of astronomy. For x-ray astronomers the most obvious advantage of imaging at high energies is that the spectra of faint
objects can be measured up at energies where the diffuse x-ray background spectrum is not only well mea-

and the high throughput missions SPECTRA,
ASTRO-D, and XMM fly.
At higher energies (10-40 keV) true imaging is not

sured3 but also has a spectral feature (the exponential roll-off at'-35 keV). To find a spectral match between some candidate class of object (e.g., quasars) and the
x-ray background in this energy range would be a

currently being considered. Coded aperture designs (e.g., EXITE1 ) are being built and these do provide min of arc imaging and simultaneous background monitoring. However they do not realize the great background reduction, and hence sensitivity, that focusing provides. could provide.
2 sources

major discovery. It would make a much stronger case for their integrated emission being the origin of the xray background than any amount of indirect argument
from source counts or spectra at low energies, such as we have been forced to use up to now.

The lack of imaging is holding back The HEAO-A4 catalog of hard x-ray

the sensitivity of observations in this energy band in spite of the potential importance of the science that it is the largest to date and lists only some seventy sources in the 13-40-keV band, of which seven are

Seeing this problem in high energy x-ray instrumentation we decided to explore,the possibilities for true
x-ray imaging in the 10-40-keV range. The grazing angle at 30 keV is -10 min of arc. Bilderback and

extragalactic. This is in contrast to the thousands of sources accessible to an imaging telescope of smaller effective area. The cause is simply that the detection

Hubbard4 have demonstrated that uncoated float glass reflects well up to 40 keV, but only at tiny grazing
angles of -2 min of arc. At 10-min of arc grazing angle

the reflectivity they measured drops sharply beyond 10 keV. Bilderback and Hubbard5 also showed that highly polished optical flats coated with platinum (these were AXAF test flats) also reflected well up to
40 keV, in this case for grazing angles of 6.8 min of arc.

The authors are with Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, Massachusetts 02138. Received 15 September 1987. 0003-6935/88/081481-05$02.00/0. © 1988 Optical Society of America.

Bilderback and Hubbard did not measure the image quality obtained in these studies. Their reflected beam was measured with a detector having a 10 opening angle. At small grazing angles a well-concentrated
OPTICS 15 April 1988 / Vol. 27, No. 8 / APPLIED 1481


TableI. Imaging Science 10-40 keV at

100

X-ray background
Quasars/AGN Galaxies

Measure spectra of candidate point
sources in region where XRB has a

feature
Multicomponent spectra AGN in cDs

s0

and clusters (distinguish from hot
gas) Color-color diagram (compare galactic sources)
Ir 8.1 keV Au 8.1 keV \

Binary content of Ellipticals Spectra of starbursts (binaries or hot
gas?)

Pulsars/SNR
Radio galaxies y-Rays Galactic sources

In SNR (distinguish from hot gas) spectra of synchrotron nebulae
Inverse Compton from radio lobes (M87); also detection in clusters Source identifications Low energy spectra, variability

10

Faint galaxy population (need >3keV sensitivity because of galactic NH), e.g., cataclysmic variablesonly 4 magnetic CVs known Cyclotron lines in faint sources
10, 20' 30' 40' 30' 1°

Sunyaev-Zeldovich effect

Present A-wavemeasurements predict
cluster temperatures of '15 keV, needs hard response to check

Fig. 1. Comparison of gold and iridium reflectivities at 8.1 keV (Cu-K): 0, 175-A layer of iridium on float glass; X, +, two examples of a 500-Agold layer on float glass.

11. Reflectivity Measurements

image is essential to realizing the background reducing
potential of imaging. For a typical 3-m focal length a 10-min of arc grazing angle gives a mirror diameter of 1.7 cm and a geometric area of -230 mm 2. If the effective area of the mirror is 20% of the geometric (see

We have modified the IPC test facility to allow mirror samples to be measured at small grazing angles and at energies up to 40 keV. The main element that
allows control of the grazing angle is a two-stage (rota-

tion plus 1-Dtranslation) platform with accurate readouts (to L1min of arc). The high energy x-ray source

below), to obtain a background reduction factor >10
the image spot must have a radius of <1.2 mm or -1 min of arc.

It is also clear that with such small mirror diameters the only ways to get significant collecting area at high
energies are either to go to very long focal lengths,

is based on a 50-kV power supply. Niobium, tin, and praseodymium are used as fluorescence targets to produce lines at '17.5 keV (Nb-Ka,f), -26.5 keV (SnKa,o), and -38 keV (P-Kolf3), respectively. The IPC was filled with xenon at 1 atm. The energy resolution
of the IPC was measured to be 12% at 22 keV using a

>100 m, or to have multiple, modular mirrors. The second alternative is more practical and we propose a method of achieving useful areas in a later section. With such small mirror diameters it is essential to use thin mirror materials, either float glass or preferably
foils6 if they can make a small enough image.

radioactive cadmium 109 source.
A. Comparison of Iridium and Gold as X-Ray Reflectors

We therefore decided to measure the reflectivity of coated float glass samples with an imaging detector as a first step to obtaining estimates of the reflectivity obtainable in a small beam size using real candidate mirror materials. Our first investigation was a comparison of gold and iridium as reflecting materials.
We did this since, while gold is commonly used in

Gold is normally used as the reflecting surface on x-ray mirrors because of its high bulk density (p = 19.3). The solid element with the highest bulk density
is iridium (p = 22.4) and so it should, in principle, have

grazing incidence optics, iridium has a potential for higher reflectivities because of its higher bulk density.
We followed this by measuring the reflectivity as a function of grazing angle of an iridium-coated sample at high energies.

a higher x-ray reflectivity. At high energies this is particularly important since it would allow 10%larger grazing angles to be used and hence 20%higher aperture utilization. No laboratory measurements of iridium x-ray reflectivity have been reported however. We have therefore compared the reflectivity of gold and iridium evaporatively coated samples of flat glass up to
energies of 8.1 keV (Cu-Ka). Figure 1 shows the

results of these tests.
Our two gold samples both had a thickness of 500 A.

To understand the implications of the encouraging results of the reflectivity measurements we investigated a mirror design that could form an Explorer class

payload. We find that a factor of 15 improvement in sensitivity over previous experiments is readily achievable.
1482 APPLIED OPTICS / Vol. 27, No. 8 / 15 April 1988

The iridium coating had a thickness of 175 A. Although this is thinner than the gold sample it should be adequate for x-ray reflection. At 30 keV the maximum l/e electric field depth in iridium is -11 A.7,8 This gives an estimate of the depth of the iridium layer needed for specular reflection.


At each energy the measured iridium reflectivity exceeds the measured gold reflectivity, as predicted. However both the gold and iridium data fall below the curves of theoretical performance. Data on a similarly prepared 400-A thick iridium layer indicate a similar result.9 This suggests that our 175-A layer is thick
enough. These theoretical curves are based on calcu-

100

m

.

..

50

I'

..

s

.\

lations of the optical constants, as used in the Fresnel equations. They assume an infinitely smooth, flat surface and this approximation becomes important near the critical angle. The use of actual measured values is thus critical and is preferable to the theoretical values. Our data are currently limited to one sam10 Ij

268keV~

ple of iridium-coated glass. (The vapor deposition
process is expensive for iridium since it requires a large

-e

amount of raw material.) Variations from sample to sample may occur depending on the details of the coating process. This is true of our gold-coated samples. Figure 1 shows that our two samples of gold5'

coated glass differ by quite large factors at large grazing angles. The difference is -30% at 32 min of arc.

The causes of this changeability need investigation. One possible explanation is that the density of the
evaporatively deposited layer is not as high as for the

Fig. 2. Reflectivity of a 175-A layer of iridium on float glass at 8.1 keV ( Cu-K), 17.6 keV (X, Nb-K), and 26.8 keV (, Sn-K). Theoretical curves for the three energies are shown: that for 8.1 keV (dots) is based on measured optical constants; those for 17.6 keV

,10 20' 30'
1O I
i

o .x.1keV
k as ev

X176

1°

bulk material. A good fit to the observed iridium data
can be produced by allowing the density to be a free parameter. Doing this gives a value of 19.36 for the

(short dashes) and 26.8 keV (long dashes) are based on predicted optical constants (see text).

density of the iridium coating, compared with 22.4 for the bulk material. This fitted value is similar to the bulk density of gold and indeed the iridium data fall close to the theoretical curve for gold. Alternative deposition techniques, such as sputtering, need to be examined to see if they give higher densities.
B. Iridium Reflectivity at 17 and 26 keV

measured FWHM of the line in the IPC is 30%. The intrinsic resolution of the IPC at these energies is 12% FWHM (as measured from a radioactive cadmium 109
source) so that a blend of the two K lines should give a

FWHM of 26%, consistent with the measured value.
The effective energy we measure will lie between the two lines which for niobium are at 16.5 and 18.6 keV and for tin are at 25.1 and 28.4 keV. Varying the

Since iridium gave better reflectivity than gold by the expected amount, we adopted iridium as our base line reflector. We then measured the same 175-A
sample at higher energies, 17.5 and 26.5 keV. Mea-

energy of the incoming radiation to fit the measured
reflectivity at the -20% level gives energies of 15.6 and

25.2 keV for niobium and tin, respectively. These
energies are too low to be the actual effective energies

surements at 38 keV using a praseodymium target are planned. The results of these measurements are
shown in Fig. 2, together with the 8.1-keV measurements from Fig. 1 as a comparison.

The reflectivities are 30% at the critical angles of
17.5 min of arc (17 keV) and 11.4 min of arc (26 keV), respectively. Theoretical curves of reflectivity 7 8 are also shown in Fig. 1. They predict the shape of the

we are using so that the higher reflectivities we measure compared with the prediction must be due to the limitations of the simple theory used. Our reflectivity measurements agree well with the values found by Bilderback and Hubbard 5 for platinum. They found reflectivities at 17 keV of 88%and 70%at 6.9 and 14 min of arc and at 27 keV they found
86% reflectivity at 6.9 min of arc.

measured function well but slightly overpredict the reflectivity at small angles and underpredict it at large angles. The theoretical values were derived using predicted optical constants rather than measured values. Measured optical constants and accurate theoretical
predictions are not readily available above 10 keV. The mismatch at large angles could be due to an incorrect assignment of the line energy. At large an-

The agreement of our reflectivity measurements in a
small (-3-min of arc) beam size shows that large angle

scattering due to poor surface roughness is not an
important factor even at wavelengths as short as 0.5 (27 keV).
Ill. High Energy Telescope: Baez Mirrors Multifocus Kirkpatrick-

A

gles the reflectivity is sensitive to the precise energy.
The effective energy of our measurements is not accu-

Since the frontal area of a high energy mirror is only -12 cm 2 for a focal length of the order of 3 m, it is

rately determined since the measured line is a blend of the K-a and K-: lines and the K-a line is more absorbed by the niobium or tin filters that are used to block the continuum radiation from the source. The

essential to combine many mirror modules to obtain a useful effective area. For reasonable aperture utilization efficiencies a 10 X 10 array of modules will give a
2 150-cm effective area. The detector area onto which
15 April 1988 / Vol. 27, No. 8 / APPLIED OPTICS 1483


the images are focused sums to 1 cm2 giving an imaging

(a)

INCIDENT X-RAYS

advantage of 150 at the lower (15-25-keV) range and
60 at higher energies (25-40 keV).

The individual construction of many such small mirrors would be difficult and time-consuming. We propose a telescope made of highly nested reflectors in a

35-cm square aperture.

Possible materials include

thin glass and metal foils. We consider both a conservative design using the rigidity of float glass and a design using metal foils. For the front mirror a series

of Kirkpatrick-Baez fan-shaped mirrdrs of 1.75-cm radius would be laid parallel to one another, forming a square frontal surface (Fig. 3). The rear mirror would
be an identical configuration (with appropriately altered grazing angles) turned through 90°. Viewed
(b)

from the front the glass plates form a criss-cross pattern. Each of the intersections of the front and rear
plates forms a single small mirror module. The diffi-

culties of making the 100 small mirrors are at least square rooted by this means. The availability of suitable float glass plates is demonstrated by the LAMAR success using even larger plates. The difficulties of

mounting metal foils in this arrangement are under investigation but do not seem to be insurmountable at this stage. This mirror module forms 100 separate images with
-5-min of arc fields of view at the focal plane and thus requires either 100 small detectors or, more simply, a large area position sensitive detector. A 30- X 30-cm

Fig. 3. Illustrations of the multifocus Kirkpatrick-Baez geometry: (a) isometric view showing crossed sets of mirrors. Each intersection of a front and a rear fan forms a separate telescope. Note: All the angles shown are greatly exaggerated. In practice since the grazing angles are never >20 min of arc, the reflectors are essentially parallel. (b) Top view of a 10 X 10 module instrument with an expanded view of one module showing the arrangement of the plates.
400

IPC has been constructed at SAO and functioned well.
A xenon-filled IPC was used for our high energy mea-

surements and behaved nominally with the expected
E improvement in energy resolution to 12% at 22 keV. In many regards a high energy IPC behind such a
300 (a)

mirror would be simpler than the Einstein IPC. It
would need no gas flow system since a thick entrance window could be used. (This would remove the response below 2-3 keV but this is not designed as a telescope for low energy use.) The crisscross pattern of the images would form a natural pattern to use for the window support structure (ribs) and so no effective area need be lost in this way. The same crisscross pattern of black bands would also provide a valuable
':

200

100

2

10

20 E(keV)

30

40

simultaneous monitor of the particle background in the same detector.
We have calculated the effective area achievable with such a design. Figure 4 shows the area vs energy

Fig. 4. Effective area of the proposed 3-m 35- X 35-cm instrument as a function of energy for (a) 0.1-mm foils, (b) 0.7-mm float glass.

for reasonable design parameters.

The reflectivity sources and other hard x-ray missions for a short
(3000-s) and a long (105-s) exposures. The foil reflec-

used for these calculations was that given by the simple theory of Sec. II. Our measurements suggest higher values at large grazing angles so that the effective area The shown in Fig. 4 is probably an underestimate. complete instrument will be described in more detail in a forthcoming publication.

tors are -30 times more sensitive than the HEAO-A4
survey in the 15-25-keV band. Since the HEAO-A4 survey detected seven active galactic nuclei (AGN)

this instrument should be able to detect more than
1000 AGN in a minimal 3000-s exposure. It is also 12 At

To estimate the sensitivity of this instrument we
have assumed background rates of 2.5 X 10-4 ct cm2

keV-1 through the 10-40-keV band. 2.5 times the HEAO-A2 experiment and with, e.g., those obtained on the Ginga have calculated the resulting sensitivity 12-25 and 25-40 keV with those of
sol
1484

This is some is comparable satellite. We in two bands, typical x-ray

times more sensitive than HEAO-A4 in the 25-40-keV range in the same short exposure. In a deep exposure
it is over 150 times more sensitive than HEAO-A4. this level the sky contains nearly 15,000 AGN.

An alternative way of exploring its sensitivity is to
ask what low energy sources would be detectable by

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


this instrument.

We use AGN as an example. A 0.05-

,Jy (-0.05 IPC ct s1-- 0.05 UFU) source is detectable
at >5o in 3000 s even if it has a power-law photon index of 2.0. For a 3C273 like spectrum (aph = 1.5), the source would be detected at over 1000. 3C273 itself is

flown to date and can obtain spectra of 0.05 UFU quasars out to 35 keV. This project was supported in part by Smithsonian Institution Research Funds. This material was presented as paper 830-47 at the Conference on Grazing Incidence Optics for Astronomical and Laboratory Applications, sponsored by SPIE, the International Society for Optical Engineering, 17-19 Aug. 1987, San Diego, CA.
References

almost 100 times brighter than these example sources. If the float glass reflectors are used instead the sensitivity is halved. A 0.1-,uJy source should be substitut-

ed in the quasar examples. The deep pointings would be -85 times more sensitive than HEAO-A4.
The design used for Fig. 4 produces a 1-mm image spot or -1 min of arc at a 3-m focal length. This is

fixed by the simplifying requirement that the plates be single flats. If the mirrors were made in three segments of 12 cm length each, the spot size could be

1. J. G. Grindlay, M. R. Garcia, R. I. Burg, and S. S. Murray, "The Energetic X-Ray Imaging Telescope Experiment (EXITE),"
IEEE Trans. Nucl. Sci. NS-33, 735 (1986). 2. A. Levine et al., "The HEAO 1 A-4 Catalog of High-Energy XRay Sources," Astrophys. J. 54, 581 (1984).

reduced to 0.33 mm. This would givean extra factor of -10 reduction in background. Mirror alignment becomes a significant problem at this level. This needs investigation.

3. F. E. Marshall et al., "The Diffuse X-Ray Background Spectrum
from 3 to 50 keV," Astrophys. J. 235, 4 (1980).

4. D. H. Bilderback and S. Hubbard, "X-Ray Mirror Reflectivities
from 3.8 to 50 keV (3.3 to .25 A). I: Float Glass," Nuclear

Instruments and Methods, Physics Research, 195,85 (1982). 5. D. H. Bilderback and S. Hubbard, "X-Ray Mirror Reflectivities
IV. Conclusion from 3.8 to 50 keV (3.3 to .25 A) II. Pt, Si and Other Materials,"

Nuclear Instruments and Methods, Physics Research, 195, 91
(1982).

Measurements at high energies using realistic candidate mirror materials and a small field-of-view detector have demonstrated that grazing incidence optics have potential for the building of astronomical telescopes. Simulations of a multifocus Kirkpatrick-Baez

6. P. J. Serlemitsos, "Broad Band X-Ray Telescope (BBXRT)," in X-Ray Astronomy in the 1980s, NASA Tech. Memo. 83848
(1981). 7. R. Giacconi, W. P. Reidy, G. S. Vaiana, L. P. Van Speybroeck, and T. F. Zehnpfennig, "Grazing Incidence Telescopes for X-Ray Astronomy," Space Sci. Rev. 9, 3 (1969). 8. B. Aschenbach, "X-Ray Telescopes," Rep. Prog. Phys. 48, 579 (1985).

mirror design with min of arc resolution shows that a significant effective area can be obtained with a modest instrument. A 35-cm square mirror/detector with
a 3-m focal length can easily achieve a factor of 30

sensitivity improvement over the best instruments

9. 0. Citterio, Astronomical Observatory of Brera Merate; private
communication (1986).

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

1485