The European Space Agency is funding e2v’s development of an image sensor for the visible instrument in the Euclid space telescope. Euclid is a candidate mission for one of two launch opportunities in 2018. If selected, the mission aims to map the dark universe with two complementary methods: a galaxy red-shift survey and weak gravitational lensing using near infrared and visible instruments.
The visible instrument will be dedicated to the weak gravitational lensing method where the mission will survey 20,000 deg2 of the extragalactic sky with a resolution of 0.18 — 0.23 arcsec in the wavelength range of 550-920nm. The survey will measure the shapes of more than 2 billion galaxies to study the orientation and distortion of these by the gravitational field along the optical path. The results of the survey will reveal the dark matter distribution and cosmological structures from their gravitational effect on the images.
The precision with which the Euclid visible image sensors can measure the shapes of galaxies is of paramount importance to the success of the mission. In order to achieve this aim, Charge-Coupled Devices (CCDs) were selected over CMOS image sensors or infrared detectors operated in the visible range for their higher dynamic range, lower noise, superior uniformity and extensive heritage in space missions.
The baseline sensor for the mission is e2v’s CCD203-82, initially developed for NASA’s Solar Dynamics Observatory (SDO) Atmospheric Imaging Assembly (AIA) and Helioseismic and Magnetic Imager (HMI) instruments. With the launch of the SDO in February 2010, 6 sensors are currently in operation on the AIA and HMI instruments.
The CCD203-82 is a 4.1k × 4.1k pixel sensor with each pixel measuring 12 μm × 12 μm. The sensor can be read out through a single port, or a maximum of 4 ports to minimise the number of charge transfers and increase the frame rate. When assembled into the focal plane of 4 × 9 sensors, the Euclid visible instrument will have a large field of view that is sampled with sufficient frequency.
The CCD203-82 was designed to accommodate binning of multiple lines into the serial read-out register. This feature has two impacts on the performance of the sensor. The first is that the CCD output amplifier has to have a lower responsivity and higher noise than a circuit that is optimised for the capacity of the image pixel. The second is that the serial register width needs to be increased to handle the large binned signals.
The Euclid visible instrument requires the full resolution afforded by the image area and will not need to use the binning capability. The higher responsivity, lower noise amplifier can thus be used.
With a higher responsivity amplifier, the charge handling capacity of the serial register now far exceeds the requirements of the image area pixel. A narrower serial register can thus be used as this confines the charge to a smaller volume of silicon and improves the charge transfer efficiency, particularly later in the mission when the effects of displacement damage from energetic particles begin to take their toll. This change helps preserve the shapes of the galaxies during the serial transfer of the charge packets during read-out.
A third change to the CCD203-82 has been indentified for the Euclid CCD development. The instrument baseline, described in detail in M. Cropper et al, will use 4 output ports to read out each 2k × 2k quadrant. In this scheme the device is read out without transferring charge across the center of the device. A charge injection structure can thus be included in the center of the device to inject a line of charge that can be transferred through the device in over-scan rows when the images are read out. The injected charge can fill any empty traps in the device and could potentially improve the charge transfer efficiency.
The three improvements to the baseline CCD have been included in the design of a breadboard model for Euclid. Charge injection schemes will be studied in detail using the baseline operating conditions for the Euclid CCDs. It is envisaged that each image will be acquired with an integration time of at least 400 seconds at a temperature of - 120°C. The low temperature is specified to minimise dark current to such a level that the long exposure time enables faint sources to be detected.
Although there can be some advantage in optimising the read-out rates and temperature, the charge transfer efficiency is fundamentally limited by the different trap species generally present in silicon. However, these detectors will see a diffuse optical background that may serve to keep the traps filled. The combination of the diffuse optical background and charge injection described earlier may prove to limit the effects of radiation damage and extend the performance and potential duration of the mission.
The incorporation of the charge injection structure will sacrifice 4 rows of the image area, representing less than 0.1% of each device. If the improvement is considered to be less important than the loss of image area, the structure can be replaced by imaging pixels for the flight devices.
The devices will be back illuminated for the highest quantum efficiency and thinned sufficiently to be fully depleted, thereby eliminating the field-free regions that could lead to charge diffusion and loss of resolution. A low voltage process will be used to reduce power dissipation and as an added benefit, will decrease the threshold voltage shift experienced as a result of ionising radiation.
The lightweight silicon carbide package has been designed for 4-side buttability with the image area comprising more than 80% percent of the total footprint on the focal plane.
A temperature sensor has been incorporated in the side walls of the package to allow calibration of the science data without impacting on the focal plane assembly and integration.
The Euclid VIS instrument will benefit from an optimised sensor design that is built on extensive CCD heritage. The low noise amplifier, charge injection structure and serial register design change will lead to an improved signal to noise ratio and point spread function. This will improve the accuracy with which the instrument will measure gravitationally distorted galaxies.
M. Cropper et al., “VIS: the visible imager for Euclid”, Proc. SPIE 7731-55, , 2010.