Our instrument contains an echelon grating cross-dispersed by (or replaced by) an echelle grating. Therefore, we refer to it as the Echelon X Echelle Spectrograph, or EXES.
1. Optical Design
The EXES optics consist of three sections: focal reducing fore-optics, the echelon chamber, and the cross-dispersion/long-slit grating chamber. The fore-optics image the telescope secondary onto a cold pupil and reimage the telescope focal plane with 2:1 demagnification through a filter wheel onto a slit wheel. The echelon chamber contains the high resolution echelon grating and its collimator/camera mirror. The grating chamber contains an echelle and a first-order grating, their collimator/camera mirror, focal reduction lenses, and the detector array.
EXES can be used in four different modes. In hi-res cross-dispersed mode (R approx. 100,000), all three chambers are used, with the echelle grating serving as the cross-disperser for the echelon. In mid-res long-slit mode (R approx. 10,000), the echelon chamber is bypassed, and the echelle serves as the primary disperser. In lo-res long-slit mode (R approx. 2,000), a low-order grating mounted on the back of the echelle is used as the disperser. In acquisition-camera mode, the low-order grating is turned face-on to act as a low efficiency (~10%) mirror.
Several views of EXES, showing different parts, are in Figures 1-3. Figure 1 is a side view of the instrument in long-slit mode. An end view, looking toward the cryogen cans and the telescope, with most structural components deleted, is in Figure 2. A view looking down from the ceiling of the airplane and showing the echelon and cross-dispersion chambers is in Figure 3. The optics are described in more detail in the figure captions.
1.2 Optical Performance
Computer raytracing shows that the optical performance in all modes is quite acceptable, although not always diffraction limited. The dominant aberration is astigmatism, caused by the fore-optics ellipsoid and the focal reduction lenses. With planned slit widths and the expected SOFIA image quality, abberations have at most a small effect on the EXES spectral and spatial resolutions.
2. Performance Issues
The main concern in the fore-optics design is the entrance window/lens material. To work out to the 28 µm detector cuttoff we must use CsBr, CsI, or KRS5. None of these materials is very desirable; the Cs salts are strongly hygroscopic, whereas KRS5 has high reflection losses and cannot (to our knowledge) be anti-reflection coated. Our current idea is to use CsI with a cover that is removed only when the dewar is attached to the purged telescope cavity. The slit wheel is currently designed to have slits of five widths: 1 arcsec, 1.4 arcsec, 2 arcsec, 2.8 arcsec, and 4 arcsec, and three lengths for each width: short slits of 7 and 10 times the slit width and 3 arcmin long slits. It will also have an open position for imaging. However, we are studying slit mechanisms which would be continuously adjustable in width, length, and position in the focal plane. Adjustable slit width and length would allow slightly better optimization for each wavelength, but could complicate calibrations.
The echelon is essentially a 1 m long, 10 cm wide, R10 diffraction grating, with 7 mm groove spacing diamond-machined in a single piece of aluminum. Thermal stability of the echelon, both during machining and during cooldown, are a concern. It will be necessary for the grating to be held at constant temperature to ~1 C during machining, and the substrate will have to be heat treated and thermally cycled so that it does not distort during cooling.
2.3 Cross Dispersion Chamber
The choice and implementation of focal reduction lenses are a concern. The focal reduction options depend on the detector chosen and the tolerances for oversampling, aberrations, wavelength coverage, and number of feedthroughs. With 30 µm pixels, the most likely size, an f/10 => f/6 lens used for 5-14 µm, and an f/10 => f/3 lens pair used for 17-28 µm would give Nyquist sampling at 10 µm and 20 µm, respectively. However, the f/3 lens pair would have to be made of CsI, and so would suffer from surface reflections (7% per surface) and would require considerable care due to the hygroscopic nature of CsI. Aberrations in the f/3 lens pair cause the Strehl ratio to fall below 0.9 for the outer half of a 3 arcminute slit length.
3. Mechanical and Cryogenic Design
The dewar design is shown in Figure 1 and Figure 3 . Its outer wall is an 0.5-m diameter, 1.5-m long cylinder (including its mounting ring). We estimate the mass to be 200 kg. The guts of the dewar are supported by fiberglass (G10) supports at each end, an arrangement we have used in several dewars and which provides quite stiff support with minimal heat flow. We expect 24-48 hour hold time for the LHe and ~24 hour hold time for the LN2.
There are five optical components that will be remotely driven while observing: the filter wheel, slit wheel, spectroscopic-mode-changing flats, cross-dispersion grating, and focal reduction lens ssembly. In addition, one or more tilt adjustments of the collimator/camera paraboloids may be made accessible from outside of the dewar.
The coupling of the instrument to the SOFIA SI mounting flange is relatively simple. (See Figure 1.) The aft end of the dewar is bolted to an 0.5-m diameter ring, which is bolted to a mounting plate. The mounting plate is then bolted to the SI flange.
Three mechanisms will be mounted on the telescope side of our mounting plate: the removable window seal, an ambient-temperature black body for flux and atmosphere calibration, and a mirror to view a gas cell for wavelength calibration. All can easily fit in the space between the dewar and the SOFIA gate valve.
The sensitivity of our instrument is dependent on the development of high quantum efficiency, low noise, impurity band detectors, and we require large two-dimensional arrays to make our long-slit and cross-dispersed modes worthwhile.
Two companies are working on mid-infrared doped silicon detector arrays: Boeing (formerly Rockwell) and Santa Barbara Research Corporation (Hughes SBRC). SBRC has made 256 X 256 Si:As IBC (BIB) arrays with 30 µm pixels optimized for low background use. These arrays have peak quantum efficiencies >50% (>25% between 5 and 26 µm, and falling to 10% at 28 µm), dark currents < 100 e -s-1 , read noise < 10 e -rms, and a 3 X 10 5e-charge capacity. They will work very well for any of our spectroscopic modes, and we take these parameters as our baseline for sensitivity calculations. Boeing is developing a 256 X 256 Si:As BIB array, with 50 µm pixels that is switchable between high background/low background use. At present, the demonstrated noise and dark current of this array are too high for use at our spectroscopic backgrounds, but we will follow their efforts to improve performance.
Our data acquisition and control software design will employ a modular (object-oriented) icon and menu driven structure very similar in appearance to that used at the NASA IRTF to run CSHELL. Observations may be run from manual set-ups using the icons and related menus, or by invoking text-based instrument control macros.
We will develop our quick-look data reduction and more detailed reduction packages in IDL. During observing the data will be saved to disk and piped to the quick-look program on the workstation. There will be quick-look routines for each on the four spectrograph modes which will display wavelength-calibrated spectra for the user. The first step of the data reduction will be a pipeline routine which carries out standard procedures of spike removal, chop and nod differencing, flat-fielding, distortion correction, wavelength calibration, and coaddition of repeated observations. This processing can be carried out during observations, so most users will leave the airplane with data they can understand.
6. Current Issues
We are currently addressing the following issues: Again, we welcome your comments and ideas.
Which spectral resolving power will provide the maximum scientific return?
In particular, we have great flexibility in the choice of a low resolution grating.
How can we best align the instrument with the telescope?
We might be able to translate the input lens to adjust the alignment. We will probably install a third lens option in front of the detector to image the cold stop and monitor our alignment.
What is the best way of heat treating and lightweighting the echelon?
What is the best size of the dewar top ring?
By increasing the top ring size, we increase the size of our optical surface and presumably increase the overall stability of the dewar to flexure. However, this will also increase the thickness of the flat end plates and the weight of the entire instrument.
What is the best mechanism for our slit wheel?
We are presently studying an adjustable system of two wheels: one with different width slits and the second with various deckers.
What is the optimum strength for internal dewar supports?
We cannot have the fiberglass tabs broken during a bumpy landing, but we also want to minimize heat flow into the dewar.
What is the likely future of detector arrays in terms of numbers of pixels and pixel size?
We want EXES to take advantage of future improvements.
What are the best choices for lens optics? In particular, is there a way of protecting and AR coating CsI?
Are coatings necessary for optical components made of diamond-machined aluminum?
We don't know the losses associated with diamond-machined aluminum in the 5 to 30 µm region for cold (4 K) mirrors.