FOCAL-PLANE ARRAYS: Space FPA will have close to a billion pixels
The Gaia satellite is a cornerstone mission of the European Space Agency (ESA); its goal is to map the positions, parallaxes, and proper motions of approximately one billion objects in the Milky Way to an unprecedented accuracy.
The Gaia satellite is a cornerstone mission of the European Space Agency (ESA); its goal is to map the positions, parallaxes, and proper motions of approximately one billion objects in the Milky Way to an unprecedented accuracy. Radial velocities of the brightest objects will be measured and photometric data taken to determine the composition of the objects. In addition, extrasolar planets and near-earth objects are expected to be discovered.
When Gaia launches in 2012, it will have the largest focal-plane assembly (FPA) ever flown in space.1, 2 The FPA will contain a mosaic of 106 large-area, high-performance CCD imaging sensors custom-designed, manufactured, and tested by e2v technologies. Once the CCDs are integrated into the FPA by EADS Astrium (the Gaia satellite prime contractor) in Toulouse, France, its light-sensitive area will be more than 0.28 m2 in size and contain nearly a billion pixels.
The result of Gaia's measurements will be the most detailed and precise map of the Milky Way ever produced; the aim is to measure the positions of stars to an accuracy of 10 to 300 microarcseconds. The Gaia satellite will be positioned at the second Lagrange point of the Earth-Sun system, about 1.5 million km from Earth. To maintain the correct attitude such that it is always facing away from the Sun, the Gaia satellite will spin on its axis. As a result, it will be possible for the whole sky to be mapped by the satellite's two telescopes. These telescopes will share a common focal plane.
|FIGURE 1. Each group of CCDs in the FPA has a distinct function (a). The FPA along with the two telescopes will be mounted on a SiC torus (b); 106 CCDs will be mounted on a SiC support structure (c). [(a & b) Courtesy of EADS Astrium; (c) Courtesy of Boostec]|
The 106 CCDs in Gaia are mounted on a silicon carbide (SiC) support structure manufactured by Boostec (see Fig. 1). Gaia contains three CCD variants, which are each optimized for different wavelengths within the 250 to 1000 nm range; these are positioned in the array according to their functions. Some CCDs are used in wavefront sensors (WFSs) and basic angle monitors (BAMs) to ensure good performance of the satellite as a whole, while others are used as sky mappers (SMs) to help determine which of the two telescopes is imaging a particular object.
Most of the CCDs in the array are used for scientific purposes. The largest imaging field is the astrometric field (AF), which contains 62 CCDs that will measure the positions of celestial objects. From repeated measurements of the positions of these objects, their proper motions can be determined. Two photometer fields, the red photometer (RP) and blue photometer (BP), measure the amount of light a star is emitting in the short- and long-wavelength portions of the spectrum. The three CCD variants are optimized for the AF, RP, and BP instruments. Another instrument, the radial-velocity spectrometer (RVS), is designed to measure the motion of objects toward or away from Gaia. The radial velocities and proper motions can then be combined to determine how the object is moving in the galaxy. The RVS instrument relies on the RP CCD variant.
The CCDs contain 4500 × 1966 pixels, each 10 × 30 μm in size, along with a single readout register and amplifier. Achieving this large image area requires the use of photolithographic stitching during CCD fabrication, with seven stitched sections in the horizontal direction and two in the vertical direction. Also included in the CCD is an antiblooming structure to prevent saturated pixels from spilling charge into their neighbors, and a charge-injection structure to reduce the degrading effects of long-term exposure to radiation.
|FIGURE 2. Data for the AF, RP, and BP variants of the CCD91-72 show high QE for each specific wavelength range (a). A flexible circuit connects each CCD while minimizing dead space between CCDs (b).|
Back-thinned CCDs have high QE
The CCDs are back-thinned to maximize their quantum efficiency (QE) using e2v's thinning process. In a back-thinned CCD, the light enters through the "back" surface, whereas for a conventional front-illuminated CCD, the light enters through the front and consequently must pass through the electrodes. Each CCD variant (AF, RP, and BP) has its QE optimized for a different wavelength range (see Fig. 2). This is achieved by optimizing the material thickness, initial silicon resistivity, and antireflection coatings for each variant.
Because the star images will scan across the FPA as Gaia spins on its axis, the CCDs will not operate as a conventional imager but rather will capture data in a time-delay and integration (TDI) mode. In this mode, the image field is clocked down the CCD columns at the same rate as the satellite is rotating so that the signal charge accumulates as it is clocked down the column. To keep very bright objects from saturating the CCD pixels before the image is read out of the CCD, the TDI length can be adjusted using one of 12 TDI gates. These are located at different positions in the column direction of the CCD so that when a given TDI gate is activated, charge from the rows above the position of that gate do not reach the readout register, effectively reducing the size of the CCD. In this way, both very bright and very dim objects can be measured.
From the outset, the particular version of CCD (called the CCD91-72) was designed to be placed in a FPA. The CCD package is designed to be four-side-buttable such that dead space between the chip edges and the edges of the package is kept to a minimum. On one edge, a flexible circuit provides the electrical connection to the device. Again, this is designed to reduce the dead space between CCDs in the FPA. The flexible circuit has a very thin footprint and allows electrical connections to be made directly underneath each device. The package is constructed of SiC, the same material as the FPA support structure. This material was chosen because it is lightweight and has thermal properties similar to silicon–an important consideration when the CCDs must be operated at -110ºC to minimize dark signal.
Each CCD in the FPA is driven by its own proximity electronics module (PEM). A row of PEMs receives power supplies from an interconnection module. Onboard computers are used as video-processing units to determine how each CCD is run, handling data receipt and transmission to and from Earth and performing onboard data processing.
To ensure that the CCDs meet the demanding specifications required for the Gaia satellite, they must each undergo extensive production tests at e2v, covering electrical, electro-optical, and mechanical performance.3 Further tests of the CCDs coupled with their PEM are performed after the CCDs have left e2v. Tests are also performed by the satellite prime contractor, EADS Astrium, as the devices and PEMs are integrated into the FPA.
Current status of Gaia
Gaia was first proposed in 1993 as a successor to ESA's Hipparcos satellite. e2v first began designing and manufacturing demonstrator CCDs for the mission in 2002. The final CCD design was evaluated for use on Gaia in 2006, and flight-model manufacture began in late 2007. Thus, the development of the CCDs for the FPA did not happen overnight but has instead been the result of more than eight years of close collaboration between e2v, ESA, and EADS Astrium. With just the final few CCDs to be delivered in the next couple of months, the FPA will soon start to take shape. Launch will mark the end of the Gaia design and manufacturing phase and usher in the planned five-year operational phase. The final Gaia catalog is planned to be released eight years after launch, and will spur many exciting discoveries in the years to follow.
1. R. Kohley et al., "Gaia–operational aspects and tests of Gaia flight-model CCDs," SPIE 7439A-14.
3. A. Walker et al., "The Gaia challenge: testing high-performance CCDs in large quantities," SPIE 7106-54.
Andrew Walker is a project manager in the High Performance Imaging Solutions Division at e2v technologies, 106 Waterhouse Lane, Chelmsford, Essex, CM1 2QU, UK; www.e2v.com; e-mail: firstname.lastname@example.org.