2.1.

Optical Design Requirements and Optimization

The design of the optics assembly must provide the necessary performance, when folded with the detector response, to achieve the top-level science requirements. These so-called Level-1 science requirements flow down to lower levels, with appropriate margins included to arrive at requirements for the optics themselves (Level-4).

For IXPE, early design optimizations and trade studies arrived at the final configuration of three modules, each with 24 concentrically-nested grazing-incidence mirror shells. This configuration was optimized to fit within the fairing of the initially-planned Pegasus launch vehicle (the launch vehicle was later changed to a SpaceX Falcon 9 but after the payload design was complete). Specifically, utilizing three independent mirror module assemblies instead of one larger mirror module permitted a shorter focal length for the given graze angles (dictated by the energy range of interest), which in turn permitted a shorter deployable boom to achieve the focal length on-orbit. Also, having three independent identical telescope systems (each with an MMA and a detector) gave some redundancy against hardware failure and permitted 120 deg detector clocking to mitigate against any systematic effects in the polarization-sensitive x-ray detectors once the relative orientation of the detectors was accounted for.

The final choice of 24 concentrically nested mirror shells per module was designed to give the necessary collecting area and energy response to meet science requirements, within the overall observatory mass budget, based on a mirror shell thickness that would permit an MMA angular-resolution requirement to be met with the assembly procedure planned (see Sec. 2.5). The mirror prescription chosen is the commonly-used Wolter-1 configuration⁴ in which reflections off a parabolic then a hyperbolic surface provide good imaging response off axis. Slightly unusual for an x-ray astronomy mirror system, there was no additional reflective coating on the inside of the mirror shells because the nickel/cobalt surface gave optimum reflectivity for the specific energy range of IXPE (2 to 8 keV). Thus, the additional steps of depositing an optical coating, such as iridium or gold, were avoided.

Table 1 details the parameters of the IXPE mirror module assemblies. Included in the table are the relevant requirements and the actual as-built-and-tested values. A discussion of the slight shortfall in the effective area and angular resolution is included in the calibration section, in which results are discussed in some detail.

Table 1

IXPE Mirror Module Assembly parameters.

2.2.

Mechanical Design

The principal challenge for MMA mechanical design is surviving launch loads while maintaining angular resolution. An accurate assessment of these loads is very important for avoiding overdesigning (or subsequently overtesting) the hardware. Typically, the launch provider gives the expected loads at the observatory interface, which would be propagated to the mirror modules. Clearly, this is an iterative procedure to optimize the mechanical design.

A second requirement is that the mechanical design facilitates the assembly of the mirror module, permitting assembly from the innermost mirror shell outward while not distorting the mirror shell at its attachments point and providing adequate access to bond areas sufficient to withstand launch loads. This requirement is particularly challenging as each mirror module is heavily nested, with just $\sim 2\mathrm{mm}$ between successive shells.

Figure 2 shows the mechanical design of the IXPE MMA with the front end down—its orientation during alignment and assembly (see Sec. 2.5). In order of assembly, the principal components of the MMA are the front spider, central support tube, mirror shells (inner to outer), rear spider, outer housing, and fore and aft thermal shields.

Fig. 2

Schematic showing the mechanical design of an IXPE MMA.

The front spider is the primary structural element, through which the MMA will be mounted onto the mirror module support structure (MMSS) during payload integration at the spacecraft supplier (Ball Aerospace). The central support tube serves as an alignment reference and support structure for components at the rear end of the MMA. Each of the nine spokes of the front spider has a precisely fabricated and positioned comb glued to it. During MMA alignment and assembly, each mirror shell is inserted and glued into the corresponding slot between the tines of each comb, thus attaching the mirror shells to the spokes of the front spider. The spider and combs are fabricated from 17-4PH stainless steel, which has an expansion coefficient close to that of the nickel/cobalt mirror shells. This similarity of expansion coefficients somewhat relaxes thermal control requirements (see Sec. 2.3).

The rear spider has 18 spokes, each with a metal comb (see Fig. 3). After the 24 shells are glued to the front spider, the rear spider is attached to the central support tube, and its metal combs are aligned such that each mirror shell floats within the corresponding slot between the tines of the rear-spider combs (with a typical gap of 0.22 mm on either side). Unlike the front-spider combs, which hold the mirror shells in place, the rear-spider combs do not touch the shells and merely limit excursions of the shells under launch loads to preclude shell-to-shell collisions. In this way the mirror shells are not over-constrained, as they would be if they were rigidly attached at both ends. To prevent possible marring of the mirror-shell surface and to provide protection against shock and vibration impacts, the rear comb tines have heat-shrink sleeves applied to cushion impacts.

Fig. 3

Rear spider of the MMA.

2.2.1.

Engineering unit

Early in the program, an engineering unit (EU) of the MMA was fabricated. This was a copy of a flight unit, but populated with only 6 of 24 mirror shells plus 3 mass simulators to duplicate the mass and dynamic response of the missing 18 mirror shells (Fig. 4). The six EU shells, the innermost three and the outermost three, were fabricated using flight mandrels that were then refurbished for the flight mirror fabrication.

Fig. 4

Schematic of the MMA engineering unit.

The added value of the EU to the program was considerable. It enabled the whole fabrication and assembly program to be verified at an early stage of the project, and handling fixtures and other ground support equipment to be tested. Further, it permitted validation of the optical design through x-ray testing and validation of the mechanical design through environmental testing. In fact, x-ray testing was performed before and after each mechanical and thermal test to confirm that the optical performance had not changed.

During vibration testing, an early failure led to tighter controls on humidity during the bonding process when it was discovered that the epoxy used for shell bonding was sensitive to high humidity during the curing phase. Also, it was found, using high-speed cameras, that the thin mirror shells could become unstable under vibration and behave non-linearly unless tightly constrained. This led to a change in design, in which the number of rear spider spokes was doubled from an original 9 to 18. After a re-build, subsequent testing, interleaved with x-ray tests after each environmental test, confirmed that the new design could handle the test loads without any degradation.

The EU also permitted additional shock tests to establish limits on the number and magnitude of shocks that the flight units could withstand (from multiple firings of the launch locks holding the payload in the stowed configuration). Finally, the EU provided a test system to verify procedures and hardware for MMA calibrations.

2.3.

Thermal Design

Thermal requirements for the MMAs are derived from a STOP (structural, thermal, optical performance) analysis of the effects of overall temperature and temperature gradients on the imaging performance of the optics. For this, the MMA finite-element mechanical model is used, temperature gradients are applied, and the resulting model is inputted to a ray-trace program for an assessment of the imaging performance. Temperature gradients along the MMA axis and across the MMA diameter as well as changes in the overall operating temperature of the module were investigated. The most critical requirement was found to be circumferential gradient around individual shells (i.e., with one side of a shell having a different temperature than the other side), and a maximum of 2°C was set to maintain imaging performance. Figure 5 (left) shows figure deviations on the inside of an individual mirror shell distorted by this temperature difference. Figure 5 (right) gives the net imaging distortion [in terms of half-power diameter (HPD)] per shell under this condition. Other thermal requirements established were an axial gradient of less than 5°C per mirror shell and an operating temperature of $20\pm 5°\mathrm{C}$ for each of the mirror shells (acknowledging that the inner shells will run slightly cooler than the outer ones), the latter is a fairly loose requirement aided by matching the thermal expansion coefficient of the main spider and combs to that of the mirror. The combined effect for the complete mirror module, if all of these requirements are at their limit, is a distortion of $\sim 7\mathrm{arcsec}$ HPD, which adds in quadrature to the native MMA resolution, increasing it by less than 1 arcsec.

Fig. 5

Mirror shell distorted by a circumferential temperature gradient of 2°C (left); resulting x-ray angular resolution (HPD) as a function of shell number ($24=\text{outer shell}$).

To maintain temperature, the MMA is equipped with six heaters, arranged in two circumferential bands of three heaters each, located $\sim 30\mathrm{mm}$ from the ends of the MMA (see Fig. 6) and circumferentially wrapped in multilayer insulation (MLI). To minimize heat loss from the ends of the MMA, without losing low-energy x-ray response, the MMA apertures are covered with thermal shields, contributed by Nagoya University, Japan. These consist of a $1.4\text{-}\mu \mathrm{m}$-thick polyimide film coated with $\sim 50\mathrm{nm}$ of aluminum and are supported on a 97.5% transparent stainless-steel mesh (Fig. 6). The transmission of a pair of these shields, fore and aft, is 76% at 2 keV, 88% at 3 keV, 91% at 4 keV, and 94% at 8 keV.

Fig. 6

A thermal shield mounted on the rear of a flight mirror module assembly. Visible just below the shield is a band heater (one of three at each end) that maintains the MMA within its operating temperature range.

2.4.

Mirror Shell Fabrication Process

The IXPE mirror shells are fabricated from an electroformed-nickel replication (ENR) process. In this approach, a master or mandrel is produced for each shell diameter, and a thin mirror is replicated from it through electroforming. The mirror shell is separated from the mandrel by differential thermal contraction in a chilled water bath, in which the aluminum mandrel shrinks away from the nickel alloy shell. The ENR process has been used on many programs over the years—at MSFC for HERO,⁵ FOXSI,⁶ and ART-XC⁷ and elsewhere for such missions as XMM-Newton⁸ and eROSITA.⁹

The attraction of the ENR process is that it is a relatively inexpensive way to produce medium-angular-resolution optics and, once a mandrel has been fabricated, multiple copies can be easily electroformed, making it ideal for situations in which multiple, identical mirror modules are required. Also, for the typical Wolter-1 geometry, the parabolic and hyperbolic segments can be manufactured as a single piece, greatly simplifying the alignment.

The production of the mandrel involves machining an aluminum bar to near final shape and size [see Fig. 7(a)] and then coating it with electroless nickel (nickel-phosphorous), a durable material that can be accurately figured and polished [Fig. 7(b)]. Figuring is typically performed using single-point diamond turning [Fig. 7(c)] to achieve the micron-level accuracy necessary for 20-arcsec-resolution optics. After turning, conventional lap polishing is utilized [Fig. 7(d)] to achieve a surface roughness of 0.5 nm rms, which is necessary to minimize scattering at the x-ray wavelengths of interest.

Fig. 7

The ENR process: (a) mandrel blank fabrication; (b) electroless nickel coating; (c) diamond turning; (d) lap polishing; (e) mirror shell electroforming; and (f) released shell in handling fixture.

MSFC uses a nickel/cobalt ($\sim 80/20$) alloy for its mirror shells. This material has low ductility, which makes it less susceptible to distortion during electroforming, release, and subsequent handling. A typical IXPE shell of thickness $200\mu \mathrm{m}$ is electroformed in about 11 h [Fig. 7(e)]. After the shell is released in cold water, it is dried and put in a handling fixture to wait for installation [Fig. 7(f)].

2.5.

Assembly of Mirror Modules

Assembling a mirror module is challenging as thin mirror shells can be easily distorted at attachment points, leading to performance degradation. One way to reduce this effect is to float the mirror shell in a groove, into which low-shrinkage epoxy is injected, thereby preserving, for the most part, the shell’s natural shape. For past programs at MSFC, a mirror shell has typically been supported on pins that can be used to manipulate it and center it in the spider comb grooves for bonding.¹⁰ For IXPE, however, the mirror shells are too thin for this approach, and they distort at the support points. This distortion would then be incorporated into the nearby bonded joints, degrading angular resolution. To avoid this, the approach adopted for IXPE was to hang the mirror shells from a series of wires, so any distortions due to the mirror support are at the opposite end of the shell from the bonded joints.

Figure 8 shows the design (left panel) of the IXPE MMA alignment and assembly station, which has a graphical interface for software-controlled optimization of the alignment of each shell. To allow for parallel production of four IXPE MMAs (three flight and one flight spare), there were four alignment and assembly stations, located in an ISO-7 (Class-10k) cleanroom to help satisfy cleanliness requirements.

Fig. 8

Alignment and assembly of the IXPE mirror modules. Left image is a schematic representation of an MMA alignment and assembly station. Right image shows the bonding of an IXPE mirror shell.

The IXPE alignment and assembly system holds each successive shell by hanging it via a set of nine support wires: three wires can be adjusted by piezo actuators and the other six serve as mass off-loaders. A rotary table with three high-precision proximity sensors rotates while the sensors measure radial displacements of the external surface of the mirror shell near the intersection and near the two ends of the shell. Software acquires the displacement data as a function of rotation angle and fits those data to various curves, calculating parameters (shell center, tilt, and circularity) that aid in alignment and performance optimization. Custom software analyzes these calculated parameters and adjusts the three non-offloaded hanging points to obtain the best alignment and centering of the mirror shell. This computer-controlled process typically converges in about 20 min.

Once this is done, wire shims ranging in thickness from 0.14 to 0.18 mm are inserted between the outside surface of the shell and the opposing surfaces of the comb slots. See Fig. 2 for an illustration of how the shells are bonded to slots in the combs. The shims serve several purposes: to ensure that a nominal gap exists between the shell and the comb slots so that epoxy can effectively wick into the joints, to ensure that a nominal gap exists so that joints will achieve nominal strength, and to allow for radial adjustment of the shell shape at each spoke, which may be used to improve the shell angular resolution. When the performance is optimized, the operator injects epoxy into the front-spider comb slots (see Fig. 8; right), and the assembly is left to cure overnight.¹¹

There were multiple goals for the shell assembly process. One was to center the optical node of each shell within $30\mu \mathrm{m}$ of the MMA optical axis to conservatively satisfy a $50\mu \mathrm{m}$ requirement on node position determination for overall MMA alignment on the spacecraft. Another goal was for each shell optical axis tilt to be within 25 arcsec of the MMA optical axis to ensure that the shells would float in the center of the rear comb grooves. The third goal was to use the proximity sensor data that were used to align the shells to continually calculate a cumulative performance prediction for each MMA before each shell was bonded in place. If the alignment of a particular shell did not meet a 20 arcsec goal, conservatively set to ensure a 25 arcsec module performance, the cumulative performance calculation was used to decide whether to bond the shell in place or rework the alignment. Mean shell alignment parameters for shells 2 to 24 in each MMA are reported in Table 2. It can be seen from the table that centering, tilt, and circularity contribute a small amount to the overall module angular resolution, as all parameters in Table 2 contribute to the total error in quadrature. By far the largest component in the table is the calculated parameter Delta-Delta-R (DDR). This is a measure of the change in slopes of the parabolic and hyperbolic segments as a function of azimuth around the mirror shell. It is calculated from the Keyence circularity sensor data and is a measure of low-frequency distortions in the shell, which dominate imaging performance. This number, averaged over the module, provides the best indicator of shell performance, and its running average is constantly monitored during assembly to meet the 25 arcsec HPD performance requirement.

Table 2

Shell alignment summary.

The reported values are magnitudes and represent the alignment state of the shells an average of 15.8 h after bonding (to allow for any epoxy shrinkage), immediately before installation of the next larger shell. Shell 1 is excluded because, after shell 1 bonding, the MMAs are realigned on the alignment station to remove shell 1 alignment errors. These alignment parameter averages are shown to quantify shell alignment results and have not been weighted by shell effective area. The focal plane angular error due to each alignment parameter (arcsec HPD, shown in brackets) has been weighted by the shell geometric effective area to show the relative contribution of each parameter to the total MMA HPD.

An important component of the assembly process is measuring the position of the MMA optical node with respect to surface mount reflectors (SMRs), which are mounted around the front spider (see Fig. 9). These SMRs are later used to position each MMA precisely on the observatory to allow for coalignment of each telescope. The optical node is the central point of the optic, near the intersection plane between the hyperbolic and parabolic segments. The precise location of the SMRs with respect to the node is measured during module assembly. The resulting node-position information is passed to the spacecraft integrator (Ball Aerospace) for use during MMA integration into the observatory to ensure correct alignment.

Fig. 9

SMRs mounted on front spider.