Soft X-ray Images of the Solar Corona with a Normal-Incidence Cassegrain Multilayer Telescope

High-resolution images of the sun in the soft x-ray to extreme ultraviolet(EUV) regime have been obtained with normal-incidence Cassegrain multilayer telescopes operated from a sounding rocket in space. The inherent energy-selective property of multilayer-coated optics allowed distinct groups of emission lines to be isolated in the solar corona and the transition region. The Cassegrain telescopes provided images in bands centered at 173 and 256 angstroms. The bandpass centered at 173 angstroms is dominated by emission from the ions Fe IX Fe X. This emission is from coronal plasma in the temperature range 0.8 x 106 to 1.4 x 106K. The images have angular resolution of about 1.0 to 1.5 arc seconds, and show no degradation because of x-ray scattering. Many features of coronal structure, including magnetically confined loops of hot plasma, coronal plumes, polar coronal holes, faint structures on the size scale of supergranulation and smaller, and features due to overlying cool prominences are visible in the images. The density structure of polar plumes, which are thought to contribute to the solar wind, has been derived from the observations out to 1.7 solar radii.

High-resolution images of the sun in the soft x-ray to extreme ultraviolet (EUV) regime have been obtained with normal-incidence Cassegrain multilayer telescopes operated from a sounding rocket in space. The inherent energy-selective property of multilayer-coated optics allowed distinct groups of emission lines to be isolated in the solar corona and the transition region. The Cassegrain telescopes provided images in bands centered at 173 and 256 angstroms. The bandpass centered at 173 angstroms is dominated by emission from the ions Fe IX and Fe X. This emission is from coronal plasma in the temperature range 0.8 x 106 to 1.4 x 106 K The images have angular resolution of about 1.0 to 1.5 arc seconds, and show no degradation because of x-ray scattering.
Many features ofcoronal structure, induding magnetically confined loops of hot plasma, coronal plumes, polar coronal holes, faint structures on the size scale of supergranulation and smaller, and features due to overlying cool prominences are visible in the images. The density structure of polar plumes, which are thought to contribute to the solar wind, has been derived from the observations out to 1.7 solar radii.

T HE STUDY OF THE SOLAR ATMOSPHERE IS COMPLICATED IN
that the sizes of the fundamental structures that control the important physical processes are below the resolution limit possible with instruments now available for all wavelengths. This problem is a consequence generally ofthe dominant influence ofthe solar magnetic field, which is characterized by structure on the scale of 70 kilometers (0.1 arc second as viewed from Earth) or less (1). The problem has been especially acute for soft x-ray observations because the available techniques could not provide observations with both high spatial and high spectral resolution simultaneously.
Most solar x-ray observations have been made with grazingincidence optical systems (2). These systems, which were first developed 40 years ago (3,4), have been used successfilly to observe the solar corona (5) and to study cosmic x-ray sources (6). Grazingincidence optics can provide high-quality images, but they have no inherent ability to resolve spectral features. Consequently, images obtained with grazing-incidence optics and thin filters are of low to moderate spectral resolution, and they cannot be used to isolate  (4). When soft x-rays strike a mirror at a grazing angle that is smaller than the critical angle (typically 0.5 to 3 degrees), they are efficiently reflected. The mirrors in the typical grazing-incidence telescope [the Wolter type I telescope (17)] present only a thin annulus for collecting incoming radiation. As a result, mirrors of large diameter and with long focal lengths are required to achieve reasonable collecting areas. Some improvement can be achieved by nesting one mirror inside another. The point response functions of grazing-incidence mirrors have broad wings (18), which make resohling small, faint structures difficult. As a result, radiation that is nonspecularly scattered by mirror surface finish imperfections and particulate and condensible contaminants also commonly degrades the performance of grazingincidence optics.
Resolution below an arc second may be difficult to obtain with optical systems composed of grazing-incidence optics alone. The use of normal-incidence optics allows many of the above problems to be eliminated. For a given aperture the collecting area and field of view of normal-incidence telescopes are significantly larger than those for grazing-incidence telescopes. Compound normal-incidence optical systems can be folded and thus are more compact. Normal-incidence optical surfaces are significantly easier and less costly to manufacture than grazing-incidence ones. The aberrations resulting from violations of the Abbe Sine condition that are associated with grazing-incidence optical (18) and spectroscopic (19) systems can be avoided. In addition, because normalincidence optics do not rely on grazing reflection angles to focus radiation, image degradation resulting from nonspecular scattering is much less severe.
Normal-incidence x-ray telescopes are made by coating (in vacuo) ultrasmooth surfaces with a series of altemating layers of two different materials, one a strong x-ray scatterer with high atomic number, and the second a weak x-ray scatterer or "spacer" with low atomic number. This synthetic structure acts as an x-ray Bragg diffractor. X-ray reflectivity is maximized when the condition for constructive interference is satisfied as indicated by the Bragg equation, nX = 2deff sinG, where n is the order of diffraction, is the angle of incidence with respect to the surface, deff is the effective layer-pair thickness (corrected for refraction effects), and A is the wavelength of the incident radiation. Although the reflectivity at each interface is small, when tens to hundreds oflayer pairs are used, excellent reflectivities (more than 50 percent near normal incidence) can be achieved. Methods for fabricating these thin-film synthetic microstructures (multilayers) were developed by Spiller (20) and by Barbee (21).
In contrast to naturally occurring Bragg crystals, multilayer structures can be deposited on curved substrates ofsufficient surface smoothness, which are required for high-performance optical systems. With proper selection of the materials and control of the thicknesses of the scatterer and spacer layers, multilayer structures can also be fabricated to have any layer-pair spacing between 15 and 500 A. Hence, conventional astronomical telescope configurations (22) with normal-incidence reflection optics coated with a multilayer of the appropriate layer-pair spacing can be made to operate in the soft x-ray-EUV wavelengths where in the past only grazingincidence or single-reflection normal-incidence optical systems could be used.
Because multilayer surfaces are reflective only in a certain bandpass, they can be used to isolate a particular region of the spectrum or a particular emission line or line multiplet between 30 and 1000 A (at normal incidence). Their effectiveness is limited at short wavelengtis by substrate smoothness and the atomic nature of matter and at long wavelengths by absorption in the multilayer structure. Compound systems containing both grazing-incidence and multilayer optics at non-normal incidence are required in order 30 SEPTEMBER I988 to operate at wavelengths less than 30 A (12). The spectral resolution of a multilayer is determined by the number oflayers that participate in the Bragg diffraction process, which in tum depends on the total number oflayers present and the amount of attenuation in each layer. For multilayer-coated optics, the spectral bandpass (AX/X) can be as large as 20 percent or as small as 1 percent. Normalincidence reflection efficiencies can be made as high as 70 percent with the proper choice of materials.
The first successful imaging experiments with normal-incidence multilayer optics were performed in the laboratory during the early 1980's (23). Subsequently, several groups of investigators have successfully fabricated multilayer-coated optics for the soft x-ray-EUV (24). Underwood et al. (25) have obtained a moderate resolution (about 10 arc seconds) image of a single solar active region with a single-reflection (Herochellian) multilayer telescope having a bandpass centered at 44 A (corresponding to Si XII radiation). Insrm entation. We obtained our normal-incidence images of the sun, using a Cassegrain telescope mounted on the Stanford-Marshall Space Flight Center (MSFC) rocket spectroheliograph instrument. This instrument contained nine optical systems: four systems, including the Cassegrain telescope, had normal-incidence optics exclusively, four had normal-incidence optics in conjunction with grazing-incidence optics, and one had grazing-incidence optics exclusively (12,26,27). The Cassegrain telescope had a bandpass centered at 173 A. The primary mirror was a concave spherical mirror with a diameter of6.4 cm and a radius ofcurvature of 1.2 m. The secondary mirror was a 2.5-cm-diameter convex spherical mirror with a radius of curvature of 0.5 m. This configuration yielded a focal length of 2.0 m for the combined f-33 system. We used spherical primary and secondary mirrors instead of conic mirrors because of the availability of spherical optics with superb surface quality; the resolution limit of the spherical mirrors [0.5 arc second root mean square (rms)] was comparable to that of the film used. A mesh-supported, 1720 A thick aluminum filter (28)  We measured the resolution of the Cassegrain telescope at both visible and in x-ray wavelengths (26,32) (Fig. 3). Tests at 5000 A demonstrated that the diffraction-limited resolution (defined as the ability to resolve two adjacent point sources on Kodak highresolution photographic plates) was 1.2 arc seconds. Because wellcollimated beams of 173 or 256 A radiation with the appropriate diameter were not available, we measured the x-ray resolution of the Cassegrain with 44.7 A radiation, available in the 300-meter-long MSFC x-ray test and calibration facility. We manufactured a special set of Cassegrain optics with a tungsten-carbon multilayer having the appropriate d-spacing (22.45 A) for these tests. The measured resolution at 44.7 A was 1.1 arc seconds.
We fabricated the multilayer mirrors, using previously described techniques (33). The multilayers on the mirrors in the 171 to 175 A telescope are molybdenum-silicon pairs with the spacing 36.8 A and 55.2 A. The effective layer-pair spacing doff at A = 171 A is about 85.5 A; this spacing is less than the sum of the physical thicknesses of each thin film because of refraction effects. Molybdenum-silicon was chosen because of its high reflectivity above the silicon L edge (A > 123 A), moderate spectral resolution (V/AX 15 for a single reflection and -25 for two reflections) at 171 A and excellent longterm stability (34). The effective 2d spacing and the mirror reflectivity were measured with a laser-generated plasma source and grating spectrometer (35). We were able to compare the measured and theoretical spectral responses (Fig. 4). The measured peak reflectiv-1784 ity of 0.35 was 66 percent of the predicted peak reflectivity of 0.53 (36). The super-polished spherical substrates used (37) had a surface smoothness near 3 A rms as judged by measurements on similar substrates (38).
The images were recorded on an EUV-sensitive, tabular grain (Tgrain) (39) emulsion that was specially prepared for us by the Eastman Kodak Company. This emulsion is similar to Kodak TMAX 100, but was prepared without a gelatin overcoat for optimum performance in the EUV. The 35-mm film had a Rem-Jet backing which prevented static discharges during film transport in the vacuum; the film was transported in conventional Canon T-70 cameras with no adverse effects. Previous observations in the EUV region have been carried out with Schumann emulsions (40), which are extremely sensitive to contaminants and abrasions and yield a maximum photographic diffuse density no greater than 1.6. Our Tgrain emulsion yielded densities greater than 3.0 (semispecular densities greater than 3.9), and the T-grain 100 film provided high sensitivity throughout the spectral region studied during this flight ( . These images complement the information that can be obtained (i) from images in the spectral region from 1 to 30 A (7), which is most effectively studied with grazing-incidence optical systems (42), and (ii) from the spectral region between 304 and about 600 A, (43) which is accessible to conventional single-reflection optics (16,44). Below 30 A, emission of the nonflaring solar atmosphere is dominated by hot (2 2.0 x 106 K), dense (about 108 ions per cubic centimeter) plasma confined in magnetic flux tubes (45). Between 304 A and about 600 A, solar emission is dominated by plasma from 2 x 105 K to 1 x 106 K [the transition region (46,47)], which is strongly confined by the highly structured magnetic field of the chromospheric network (Fig. 1C).
Changes in the structure of the atmosphere occur between 1 x 106 K and 1.5 x 106 K. In closed-field regions much of the gas at these temperatures is confined in loop structures (45). Systems of coronal loops are visible on the limb in the northwest, west, and east (Figs. 1, A and D, and 5), and are responsible for the emission visible from active regions 4872 and 4873 (see also Fig. 1E), at south 600, west 0°to 500. In open-field regions, where the coronal plasma is expanding into the solar wind, structures such as streamers and polar plumes (47) that are at the interface between the corona and the interplanetary medium, are visible in the 171 to 175 A images at 1 x 106 K to 1.5 x 106 K (Figs. 1, A and D, and 5). Streamers, representing the corona-solar wind interface, are visible virtually everywhere on the limb, including coronal plumes in the polar regions, which are otherwise deficient in emission. Coronal holes, representing regions that generally lack closed magnetic structures capable of confining hot plasma, are clearly visible at the poles.
Other features that are visible in the 171 to 175 A images are regions of low emission that coincide with large-scale loops of cool gas embedded in the hotter coronal gas that produces the 171 to 175 A emission. These loops of gas at chromospheric temperatures (6 x 103 K to 3 x 104 K typically) are called prominences (48), and can be observed in emission above the limb (Fig. 1, A and D).  6. Density structure of a typical polar plume (the plume is marked with an arrow in Fig. 5). The electron density is plotted as a function of distance along the plume and as a function of radial distance from the center of the sun {the radius of the sun is 6.96 x 105 km). The scatter of the data points toward the low-density edge is due to grain noise in the film. Distance along plume (1x105km) arc seconds, and to distances of 0.4 solar radii or less above the limb.
To demonstrate the power of multilayer optics for the study of the structure of the interface between the corona and the solar wind, we examined the structure of a typical polar plume (Fig. 5). We calculated plume densities assuming that the plume is isothermal with a temperature of about 1.1 x 106 K, a value consistent with previous analysis of Skylab observations. The mean electron density (VN ) in the plume in a resolution element (AxAy) can then be derived from the coronal excitation equation (7), which expresses the intensity in any emission line Xij as  6) are in good agreement with earlier EUV and soft x-ray results (52)(53)(54) and with earlier eclipse results (56) for less than 1.4 solar radii, and extend the EUV analysis ofplume structure to about 1.7 solar radii (57). The plumes near the edge of the polar coronal holes tilt outward toward lower latitudes as compared to the radial direction. The plume that is plotted in Fig. 6 is tilted about 200 away from the radial direction.
Future applications ofnormal-incidence x-ray optics. Our data demonstrate that normal-incidence multilayer x-ray telescopes can form high-resolution images with significantly lowered levels of xray scatter. Furthermore, these systems are not subject to acute optical aberration effects over reasonable fields of view. We anticipate that multilayer telescopes will provide solar x-ray images with spatial resolution approaching 0.1 arc second in the near future in well-defined spectral bands with excellent sensitivity. Additional capabilities that can be anticipated with multilayer optics, such as multilayer gratings, zone plates, and compound systems incorporating grazing-incidence primary mirrors and multilayer magnifying optics (27) or multilayer Rowland circle grating or "crystal" spectrometers (19) should extend the utility of multilayer optics to shorter wavelengths and provide astronomers with powerfiul high spectral 1786 resolution instruments that are stigmatic. These instruments should allow the study offundamental problems in solar physics such as (i) the structure, dynamics, and energy balance (including the heating source or sources) of coronal loops, (ii) the mechanism responsible for generating the solar wind, and (iii) the configurations in the corona which are necessary for the onset of flares.
The application of multilayer optics to nonsolar x-ray astronomy should prove extremely fruitful, as well. X-ray astronomy has revealed the existence ofhigh-temperature astrophysical phenomena that range from thousands of kilometers to megaparsecs (58)(59)(60).
Although not as bright from Earth as the sun, many cosmic x-ray sources have complex extended thermal and spatial structure. Among such sources are supernova remnants (61), the hot component of the interstellar medium in our galaxy (62) and in other nearby galaxies, and the hot intracluster gas bound in clusters of galaxies (60). The study ofthe thermal and density structure ofthese extended cosmic plasmas may be possible with the use of multilayer optics coupled to grazing-incidence optical primary mirrors such as the Advanced X-Ray Astrophysics Facility (63). In addition, spectrally resolved images of the halos (64) caused by interstellar scattering, which surround the image ofevery compact x-ray source, can be used to address fundamental questions concerning the composition and structure of the interstellar dust (65).
In summary, we have used normal-incidence multilayer optics to produce high spatial resolution full disk images ofthe sun in selected narrow wavebands in the soft x-ray-EUV. The photographic densities observed even in short exposures clearly reveal the high reflectivity and excellent waveband match of the primary and secondary multilayer mirrors in the Cassegrain telescope and also demonstrate the superiority of the new Eastman Kodak T-Grain emulsions over Schumann emulsions, which have long served as the photographic film of choice for soft x-ray-EUV observations (66). Multilayer optics can also lessen x-ray telescope system design requirements and yield improved image quality in an experimentally difficult spectral region. Multilayer optics do not exhibit the x-ray scattering characteristic of grazing-incidence telescopes. Consequently, small faint features can be clearly seen even though they are near bright loops or active regions. Because of these important technological developments, the images obtained during this flight have revealed features of the solar corona in greater detail at soft x-ray-EUV wavelengths. Clearly, multilayer optics will play a profound role in laboratory and astronomical x-ray imaging.