Enabling high-resolution extreme UV solar astronomy

It is the generally accepted view that heating of the solar corona is caused by magnetic reconnection in small-scale structures, i.e., with typical sizes less than 100km (0.15arcsec).¹ The first extreme UV (EUV) observations of such small-scale structures were conducted with the High Resolution Coronal Imager (Hi-C) telescope that was launched on a sounding rocket in July 2012.² Images from Hi-C had an angular pixel size of 0.1arcsec/pixel and revealed magnetic braiding in coronal loops. EUV images that were obtained simultaneously with the Atmospheric Imaging Assembly (AIA) of the Solar Dynamics Observatory (SDO)³—at 0.6arcsec/pixel—did not, however, resolve the same braiding.

EUV telescopes with angular resolutions of 0.1arcsec/pixel are now being developed at the Lebedev Physical Institute as part of the Small Satellites for Fundamental Space Research (MKA-FKI) program of the Russian Federal Space Agency.⁴ These high-resolution telescopes must be constructed from highly stable mechanical components and include large-area high-sensitivity CCD detectors. The most essential part of their design, however, is the precise large-aperture EUV mirrors they include.

As with most EUV telescopes that are used for solar corona observations, we implement a Ritchey-Chrétien (Cassegrain) optical layout with two multilayer mirrors in our design. Our multilayer mirrors (see Figure 1) consist of a multilayer coating sputtered on a substrate. The surface shape and roughness of this substrate defines the ‘imaging’ quality of the mirror itself. To achieve high optical quality and high efficiency of the multilayer coating, we require a precise surface shape (root mean square of about 4nm) and a low surface roughness (about 0.2–0.3nm).

Figure 1. Photograph of super-polished extreme UV mirrors with sputtered multilayer coatings. These mirrors have been developed for the Telescope and Spectrograph for Solar Investigations (TESIS) instrument onboard the Complex Orbital Observations Near-Earth of Activity of the Sun (CORONAS)-Photon satellite.

Standard optical manufacturing technologies provide surface roughnesses for substrates of about 1.5nm. Our requirement for super-polished substrates therefore involves issues of metrology in addition to the manufacture of the substrate itself. In dealing with angstrom-scale accuracy, it is convenient to characterize the substrate shape and roughness using surface deviations with different spatial frequencies or spatial size. Substrate surface deviations of low spatial frequency roughness (10⁻⁶ to 10⁻³μm⁻¹) distort the image shape, whereas deviations of medium spatial frequency roughness (MSFR)—10⁻³ to 1μm⁻¹—degrade the image contrast, and deviations of high spatial frequency roughness (HSFR)—1 to 10³μm⁻¹—reduce the reflectivity of the multilayer coating.⁵ We use a power spectral distribution function (defined as the autocorrelation function of the surface profile)—PSD_2D(v)—to express the amplitude of surface deviation with a given spatial frequency (v). For isotropic surfaces, we use a one-dimensional PSD(v) function. We also define the effective surface roughness (σ_e#), with spatial frequencies v_min to v_max, as: σ²_e#=∫_vmin^v_maxPSD(v)dv.

We used x-ray diffuse scattering (XRDS) and atomic force microscopy (AFM) to make measurements of MSFR and HSFR surface deviations. The spatial frequency of XRDS corresponds to the 10^{−2–1μm−1 range, and the spatial frequency of AFM corresponds to the 10−1–102μm−1 range. The results from a series of XRDS and AFM measurements that we conducted on different substrates are in excellent agreement for the regions where the two methods overlap (see Figure 2).}

Figure 2. Power spectral distribution (PSD) functions for fused silica substrates with effective surface roughness (σ_e#) values of (a) 1.02nm, (b) 0.71nm, (c) 0.54nm, and (d) 0.24nm. Measurements were made using x-ray diffuse scattering (XRDS) and atomic force microscopy (AFM).

We used traditional pitch polishing techniques, with successively smaller polishing grain sizes, to achieve σ_e# of about 1.5nm for our different substrates. We then exposed the substrates to chemical-mechanical polishing to reduce the PSD(v) by almost an order of magnitude and to produce a σ_e# of about 0.36nm. By changing the polishing regime and reducing the grain size below 1μm, we were able to decrease the PSD(v) even further (see Figure 3). We have tested several substrate materials, including fused silica (different grades), ultra-low-expansion glass, and Zerodur. We consistently produced the finest roughness (σ_e# of about 0.25nm) on a KU-1 (Russian-produced fused silica grade) fused silica substrate.⁶

Figure 3. PSD functions for a fused silica substrate. These AFM measurements were made after pitch polishing with a grain size of less than 1μm (upper curves) and chemical-mechanical polishing (lower curves).

We have tested several polishing techniques on different substrates to produce very small effective surface roughness values. We have also conducted XRDS and AFM measurements that confirm surface roughness down to about 0.25nm can be achieved. These techniques and materials can be used to manufacture precise super-polished mirrors for high-resolution EUV solar astronomy telescopes. We now plan to work on improving the precision of the substrate shapes with the use of ion-polishing under interferometer shape control.