Future x-ray astronomy missions, such as the proposed International X-ray Observatory (IXO) or the Advanced Telescope for High Energy Astrophysics (ATHENA), will be enormous leaps forward compared to preceding x-ray observatories, with their point source sensitivities, spectroscopic ranges, and maximum observable light flux all an order of magnitude better than any previous telescope. This goal can be reached by coupling high angular and energy resolution with a very large effective area (more than 2.5m² at 1keV).

Obtaining such a wide effective area with a single telescope requires a very large grazing incidence x-ray optics system. Since x-rays can be reflected only at grazing incidence, x-ray mirrors are designed as tubes, called shells. In this configuration, the effective x-ray collecting area of a shell is very small: the surface area as seen from infinity along the optical axis passing through the center of the tube. To enlarge this collecting area, several tubes are nested. IXO and ATHENA's optical designs are based on tubes with an external diameter up to 3.8m and many thin shells assembled together, with a grazing incidence double reflection on separate parabolic and hyperbolic mirrors. This optical design is called Wolter-I and allows a perfect focalization of x-rays from sources in the center of the telescope's field of view.

For practical reasons, with such huge diameters it is impossible to nest together monolithic pseudo-cylindrical mirror shells as was done for the Newton-XMM and Chandra telescopes at the end of the 1990s. The use of segmented mirror shells is mandatory, and a hierarchical approach is required to integrate the subsystem into the x-ray optics.

Segmented mirrors can be assembled in x-ray optical units (XOUs), the XOU can be assembled in petals (azimuthal sections of the telescope), and petals can be assembled in a common structure to form the final overall optics (see Figure 1). The mirrors need a very high area-to-mass ratio to overcome the mass limit imposed by current rocket launchers. And the manufacturing process for mirrors should be scalable to high volume with affordable industrial costs.

Figure 1. The complete mirror assembly and the unitary element called x-ray optical unit (XOU) composed of slumped cylindrical glasses assembled together in Wolter-I configuration. This optical design fulfills the initial International X-ray Observatory specifications.

The National Institute for Astrophysics at the Brera Astronomical Observatory (INAF-OAB) is leading a feasibility study, supported by the European Space Agency (ESA), to investigate hot slumping of borosilicate glass sheets and an innovative assembly concept that uses glass-reinforcing ribs connecting pairs of consecutive foils stacked to construct the XOU. With an appropriate design, this approach will permit the fabrication of lightweight, low-cost XOUs to be assembled together to form complete multishell Wolter-I optics.¹ We are collaborating with other institutes, including the Max Planck Institute for Extraterrestrial Physics (Garching, Germany), and small companies like BCV Progetti (Milan, Italy), ADS International (Lecco, Italy), and Media Lario Technologies (Bosisio Parini, Italy).

Our new approach should be considered a backup technology for the ‘pore optics’ technology investigated so far by ESA in collaboration with Cosine.² US groups involved in IXO optics development are also studying the slumped-glass technology.³ Slumped glasses, and the rib structure for mirror integration, have also been successfully used to build optics for the Nuclear Spectroscopy Telescope Array.⁴ However, our present study includes a number of innovative technology elements and solutions different from those pursued in the US, both for the foil-forming process and for the integration method.

While achieving the required surface accuracy on the glass segments by means of a hot-slumping technique is challenging (see Figure 2), we have developed a unique approach for correctly integrating the mirror segments and co-aligning the mirror pairs assembled together in the XOUs. To position the glass segments in the stack without introducing deformations, we fix the glass onto a precisely machined mold by vacuum suction (see Figure 3). In this configuration the optical surface of the glass is kept in contact with the mold, and the mold itself is taken as a reference during the alignment of the plates in the stack. For this reason, this scheme is called mold integration.⁵ Using the known shape and orientation of the mold, each plate pair will be first aligned and then integrated onto the stack with glue (see Figure 4).

Figure 2. The process flow for the production of slumped glasses. From left to right: The forming mold, the oven, and the resulting cylindrical slumped glass.

Figure 3. From left to right: The mold integration scheme applied to single glass plate. Vacuum suction is used to force the glass plate against the forming mold, and glue is then distributed onto the ribs. The right-hand panel shows a glass plate once integrated into a rigid structure. The ribs are used both as spacers and to stiffen the XOU.

Figure 4. The relative alignment between the two integration molds (parabola and hyperbola) permits the accurate positioning of each couple of glass sheets in the stack with a common system reference that avoids cumulative errors.

The machining of the ribs does not need to precisely follow the curved rear of the glass plate's surface. On the contrary, the ribs will be simply tapered to a coarse conical profile. All the fine differences between rib and glass profiles at macro- and microscopic levels will be compensated by the glue filling the gap. The thickness of the glue layer is about 75μm. Not only is this thickness able to absorb the differences from the desired and real profiles of the ribs, but it also corresponds to the optimal values for toughness and adherence (shown by careful experimental tests). All the glass plates in the same stack are aligned with respect to a common reference system, and we can control the alignment of the segments with sub-arcsecond accuracy.

To achieve an angular resolution of 5 arcseconds half-energy width, the mirrors must be integrated with a precise optical shape. They must also be aligned in the XOU and mounted onto the overall structure in the proper way. We have developed a semi-robotic integration machine (see Figure 5).⁶ The integration machine will align two separate molds and then simultaneously place them in position for their integration into the XOU in accordance with the integration process and the corresponding error budget. The most demanding requirement is keeping the relative attitude between the two molds constant while curing, which lasts several hours. We achieve this with autocollimator feedback, working in differential mode: differential measures of the molds' angles are done in a short time and compared with prescribed values. Each aligned couple of plate pairs is then stacked into the XOU with a hexapod working with linear sensor feedback.

Figure 5. From left to right: The integration machine, the glass plate pairs locked on a porous forming mold ready to be glued to the stack, and our first complete x-ray module based on slumped glasses and assembled with our technology.

At the present stage of the project, the plates, thermally formed on cylindrical mandrels with the active pressure technique, do take on low-order errors. We have shown that these errors can be fixed using the proposed integration scheme. These results have inspired us to investigate a more effective process flow in which the slumping mold is cylindrical and common for a wide range of shells and radius of curvature, leading to a significant cost reduction for the slumping mandrel procurement.⁷ To achieve the final Wolter-I configuration, the cylindrical glass plates must be integrated by means of accurately shaped integration molds. An ad hoc alignment makes it possible to reduce the number of integration molds needed. We simply integrate a number of consecutive plates using the same molds.⁸ The construction of the prototypal XOU breadboard demonstrated the capabilities of the process.

In summary, we have developed a novel integration scheme that allows the accurate stacking of slumped glass segments into an x-ray module. A prototype integrated stack has been assembled with a dedicated integration machine. We plan to improve the optical performances of our systems by replacing the integration molds made of a porous material (Metapore) with ones based on BK7 glass that are being produced with higher accuracy.

We would like to thank the whole OAB team: S. Basso, O. Citterio, P. Conconi, M. Ghigo, G. Pareschi, L. Proserpio, G. Sironi, D. Spiga, B. Salmaso, G. Tagliaferri, A. Zambra. We are also grateful to M. Bavdaz and E. Wille of the ESA, M. Tintori, D. Gallieni, and P. Fumi of ADS International, and G. Parodi and F. Martelli of BCV Progetti for their support and valuable efforts.