The optical design of FLAMINGOS is composed of mostly spherical all refractive optics and is derived from the successful multi-object optical spectrometers MOS at the CFHT (Morbey, Allington-Smith et al. ) and LDSS2 at WHT. I adapted these designs for use between 0.9 and 2.5 microns by using mostly near-IR materials with low indices of refraction. Glasses with low indices of refraction can be very effectively anti-reflection coated, giving losses of less than 2% per element over the wavelength range from 0.9 to 2.5 microns. Figure 1 shows the design for the FLAMINGOS optics. Figure 2 shows the expected transmission of the AR coated FLAMINGOS optics.

The key requirements for the collimator were to accept an input beam slower than f/7 and provide a pupil image with a diameter of 50-mm with good image quality for efficient baffling of the telescope thermal emission: i.e. a blur<2% of the pupil diameter. Additionally, we require a long collimated space for grisms, filters and the Lyot stops and sufficient space between the first surface of the collimator and the focal plane to allow the multi-slit mechanism.

The camera is optically fast (f/2.9 when the collimator is illuminated by an input beam of f/7) and must provide D(80)<36 microns over the full field (52 mm diameter at the array) without changing focus from 0.9 to 2.4mm. Charles Harmer of NOAO has carried out an extensive design study which has lead to the current FLAMINGOS design. The final design has a three element collimator and a 6 element camera.

A grism will be used as the dispersive element. I have chosen grisms since they offer both high efficiency and a rather compact optical design. For MOS spectroscopy we plan to disperse spectra across 1430 pixels of the 2048 pixels of the array. This will allow the MOS slits to be displaced ±15% of the FOV in the dispersion direction with out loss of spectral coverage. Two grisms will be supplied; one will place the spectral region from 0.9 to 1.8 microns across 1430 pixels of the array in first order (R~1000 with a 2 pixel slit). A second grism will disperse from 1.2 to 2.4 microns across 1430 pixels of the array (R~1000). R=1000 provides a resolution of about 300km/sec which is similar to the velocity dispersion of galaxies. FLAMINGOS provides continuous coverage from the CCD cut-off to the end of the H-band. Thus, this grism will be ideal for general survey work such as redshift surveys of distant galaxies. We also plan to try and provide a grism which will disperse a single window across 1430 pixels (R~2400) in the K and J bands. The R=2400 grism will supply sufficient resolution to study individual atomic lines and molecular bands in stars. This grating will thus allow sufficient resolution to classify stars and to make crude estimates of their abundances in the region of the near-IR opacity minimum. Additionally, R=2400 will allow the dynamics of galaxies to be studied.

Appropriate master rulings for the R~1000 gratings exist at the Richardson Grating Laboratory. Given the high cost of having a grism ruled directly into the prism face, we will construct the grisms by having a replica of the grating applied to the face of the prism. The absorption bands due to the organic molecules in the replicating resin produce only small amounts of absorption in the H band as long as the resin layer is kept thin. Such replica grisms have been commonly used in near-IR spectrometers and represent a low risk and low cost solution to the problem of near-IR dispersers (Cryogenic Optical Bench at NOAO, OSIRIS and MOSAIC at OSU and Redeye at CFHT among others). For the higher resolution R~2400 grism we are investigating direct ruling on KRS5 by Zeiss optical as has been done successfully for the NASA IRTF-NSF Camera. We are also studying the possibility of using a Volume-Phase Holographic grating manufactured by Kaiser Optical.