The major challenge faced in designing this instrument is the sheer number of pixels required to image an approximately 2 arcminute field at the diffraction limit. For critical sampling, an array of 28,000 x 28,000 pixels is required. To match an assumed pixel size of 22 microns, the image plate scale will be 5.5 mm/arcsecond, while the detectors are 685 mm on a side. Assuming 4 K x 4 K detectors to be the standard, the resulting mosaic would contain 49 such detectors. The need for such a large number of detectors will likely require a dedicated foundry run both to produce the detectors and to drive down the unit cost from current values ($1.5M-$2M per detector).
Our preliminary design is a monolithic imager with an FOV of just under 1.5 arcminutes on a side; subsequent iterations will develop a concept for the full 2 arcminute field that appears achievable with the MCAO system described elsewhere. A schematic view of the optical design is shown in Figure 1.
Figure 2 displays the spot diagrams for the final image surface. The images are diffraction-limited for the wavelengths ranging from 1.2 to 2.2 microns and across the full field of 1.44 arcminutes.
The illumination of the pupil stop is shown in Figure 3. All of the light passes through a stop with a diameter of 39 mm. The physical size of the field for this concept is 485 mm on a side for the 1.44 arcminute FOV. This reduces the number of pixels required from a 28 K x 28 K to a 20 K x 20 K detector format. Such a detector would now require only 25 arrays, or cost a factor of two less than the 2 arcminute FOV imager. This represents a $35M to $50M in detector cost alone at current, nonfoundry-run detector prices.
A significant challenge with this concept lies with the sizes of the glass materials required. The CaF2 window has a full diameter of about 500 mm; it is currently difficult to acquire elements of about 300 mm in size. The LiF and BaF2 elements are large, but they are in the regime of currently available sizes. The ZnSe elements are currently available. Technology development may be required to secure the large optics needed for a monolithic imager. Further design effort should aim to explore concepts that mitigate these technical issues by finding optical materials that can be acquired with the necessary dimensions while not critically affecting the performance of the instrument.
An alternative to a monolithic imager is to make deployable image relays that feed into single detector imagers (see Section 4.7.2). A 4 K detector would image a 16 arcsecond FOV. Likewise, a less expensive 2 K detector would be able to image 8 arcseconds on a side. Perhaps multiple deployable arms could image onto different sections of a 4 K detector to allow excellent multi-plexing without the need to image unwanted regions of the sky.
Further efforts in defining the scientific requirements for such an instrument are needed in order to explore the tradeoffs between a monolithic and a deployable imager concept. We propose to carry out these trades during FY 2002.