The ASCA Mirrors

--by Peter J. Serlemitsos, GSFC & Hideyo Kunieda, Nagoya University

ASCA is equipped with four identical conical foil X-ray mirrors. These are lightweight versions of similar mirrors flown earlier on the Broad Band X-ray Telescope experiment (BBXRT) aboard NASA's Astro-1 shuttle mission. In the limit of long focal length (small glancing angles), a conical geometry is an excellent approximation to the more precise Wolter type I geometry for grazing incidence mirrors. The conical simplification is crucial because it allows the mirrors to be constructed of thin foil. Nearly flat pieces of such foil are first shaped into flexible conical segments having approximately the required curvature, then dipped in an acrylic bath for a smoother surface, coated with a (aprox.)500 vacuum-deposited gold layer to enhance X-ray reflection, then finally placed into a housing, constrained by supports which are adjusted for optimum performance.

The foil mirror concept was first introduced at GSFC as a means of obtaining high throughput, broad band, inexpensive X-ray optics for spectroscopic studies. The development of these optics began in the late 1970's, culminating in the BBXRT mirrors. Considerable effort has been expended since then toward the ASCA design where goals were a drastic weight reduction along with improved spatial resolution. This effort was done in close collaboration with members of the physics department at Nagoya University in Japan. For both BBXRT and ASCA the modular unit was a quadrant.

Key design and performance parameters for the ASCA mirrors are tabulated in Table 1.

Table 1 XRT design Parameters

Mirror Substrate			0.125 mm aluminum foil 
Surface				10 m Acrylic Lacquer + 500  Au
Number of Nested Cones		120 (back-to-back layers)
Outer/Inner Diameter			345/120 mm 
Focal Length				3500 mm 
Grazing Angle Range			0.24 - 0.70 deg 
Geometric Area			558 cm2    (each) 
Field of View				24 arcmin (@1 keV)
					16 arcmin (@7 keV)
Number of Telescopes			4 
Mirror Weight				10 kg (each)

Performance Parameters 

Effective Area			~300 cm2 @1.5 keV (each)
					~150 cm2 @7  keV   (each)
Size of Blur (Encircled Energy)	~2 arcmin HPR (no energy dependence)
Reflector Surface Roughness		~3

Mirror performance in orbit requires, among other things, that the reflecting surfaces remain free of any condensable materials that may escape from the rest of the spacecraft and that there are no large thermal gradients across the mirror structure that would tend to defocus the mirror. To guard against such possible performance degradation, the mirrors were equipped with heating elements to elevate their mean temperature in orbit. Additionally, very thin (0.22 and 0.54 m for the SIS and GIS detectors respectively) aluminized mylar thermal covers were fastened over the entire mirror aperture.

Before launch, mirror performance had been deduced from a variety of sources including the following:

Numerous tests on small mirror segments performed over the many years of foil reflector development at the GSFC 50 M X-ray tube
Ground calibrations and flight data of the two BBXRT mirrors
Reflectivity tests on sample foils using monochromatic X-rays at a synchrotron facility
Detailed calibration of a fifth ASCA (flight spare) mirror at a White Sands 1000 ft vacuum tube facility
Lengthy calibrations of all mirror segments at the ISAS 30 M long X-ray beam line
Ray tracing at GSFC as well as Nagoya University.

Ray tracing is especially a powerful tool because, once brought into agreement with selected calibrations, it can be used to predict the performance for arbitrary source distributions and orientations. Ray tracing parameters that must be finely tuned to the actual performance in orbit include the scattering model, the azimuthal dependence of the PSF, the effective density of the vacuum-deposited gold and stray light (single reflections) including back scattering from adjacent reflectors. In addition, however, we have known since the BBXRT flight that optical constants for gold, which are used to derive reflection efficiencies, do not provide a good description of the response near gold absorption edges and must be appropriately modified.

The mostly energy independent point spread function includes a very characteristic azimuthal pattern (Maltese cross) caused by quadrant boundaries. Image spreading can be described in terms of the encircled energy function (EEF), which gives the fraction of imaged photons as a function of the radial distance from the image center. The blur is characterized by a sharp core and broad wings. Roughly 10% of the photons fall inside a region 0.3 arcmin radius, whereas another ~10% fall outside 6 arcmin radius. The portion of this spreading that can be attributed to the approximate geometry is roughly seven times smaller than that listed above. Instead, the observed blur size (especially the large wings in the response) is caused mostly by small angle scattering from foil surface defects too large (more than 100 m) to be smoothed over by the acrylic coating. The size of the blur remains roughly invariant for off-axis rays whereas the shape does change reflecting the azimuthal dependence of vignetting losses.

The mirrors were mounted on an extendable optical bench that was commanded to extend 1200 mm after launch to increase the mirror-to-detector distance to the nominal 3500 mm focal length. During integration at the spacecraft, the fields of view of the four telescopes were co-aligned to within less than one arcmin. The performance verification phase included a number of targets specially chosen to verify mirror performance in space. The analysis of relevant data is not as yet completed but the following observations have emerged thus far:

Mirror performance in space for all four telescopes is very close to that derived from pre-launch calibrations. This includes the PSF, the absolute efficiency and the stray light effects.
There is no evidence that there is any initial or gradual degradation or contamination of the reflecting surfaces.
The launch and/or thermal stresses in orbit have caused a certain degree of misalignment among the four fields of view, ranging in focal plane relative displacements as large as approximately 1 mm and relative optimum throughput misalignments approaching 3 arcmin.

Return to the Table of Contents Proceed to the next article