4. Instruments I: EXES

4.1 EXES Instrument Overview

The Echelon-cross-Echelle Spectrograph (EXES) operates in the 4.5 ‒ 28.3 μm wavelength region, at high (R ≈ 50,000 ‒ 100,000), medium (R ≈ 5000 ‒ 20,000) and low (R ≈ 1000 ‒ 3000) spectral resolution. The instrument uses a 1024x1024 Si:As detector array. High resolution is provided by an echelon ‒ a coarsely-ruled, steeply-blazed aluminum reflection grating ‒ along with an echelle grating to cross-disperse the spectrum. The echelon can be bypassed so that the echelle acts as the sole dispersive element. This results in single order spectra at medium- or low-resolution depending on the incident angle.

4.1.1 Design

EXES is a liquid helium cooled instrument. The cryostat is approximately 24 inches in diameter and 72 inches long. There are two cryogen reservoirs, one for liquid nitrogen and one for liquid helium. These are at the forward end, as mounted on SOFIA, with the entrance window on the aft end toward the telescope. There are three layers of radiation shielding within EXES - a vapor cooled shield tied only to the cryogen fill tubes, one attached to the liquid nitrogen reservoir, and the third attached to the liquid helium reservoir. All optics except for the entrance window/lens are attached to the liquid helium level. Baffling tubes connected to the liquid nitrogen level reduce thermal emission impinging on the internal optics. Within the liquid helium level, the optics are all tied to a rigid optics box constructed out of aluminum, and the detector headerboard is isolated with G10 fiberglass and actively maintained at a uniform temperature.

4.1.2 Optics

The optics consist of an entrance window/lens, fore-optics, three wheels housing the slits, deckers and filters, an echelon chamber, and a cross-dispersion chamber. The entrance window/lens (2 inches diameter) forms an image of the SOFIA telescope secondary at the liquid helium cold stop within the fore-optics. The fore-optics, including the entrance window, changes the incoming f/19 beam to f/10. After coming to a focus, the beam expands through a pupil (at the cold stop) to an ellipsoidal mirror. The light is redirected off two flat mirrors to a focus at the slit plane.

As the beam comes to a focus, it passes through the slit/filter cassette. This consists of three wheels on a common axle containing (i) filters to isolate grating orders, (ii) deckers to determine the length of the slit, and (iii) slits of different widths. The filter wheel has 12 slots, and these will be loaded with specific filters for each cooldown cycle based on the planned observations. Broader Filters for use in the low-resolution configuration are included in 4 of the decker wheel slots. The decker wheel has a total of 11 features, which include continuously variable length slits, fixed length slits, pinholes, and an open position. The continuously variable slit length is provided by a cutout on the decker wheel that gets larger as a function of angle. The smallest size is about 4.5'' and the largest about 45''. The slit length depends on the wavelength and the instrument configuration. With that caveat, slit lengths can range from 1'' to 180'' on SOFIA.The slit wheel contains six slits. On SOFIA, EXES will typically use four of them (Table 4). There is also a wide 9.400 slit intended for flux calibration.

Table 4: EXES observing configurations, modes, slits and spectral resolutions

EXES observing configurations, modes, slits and spectral resolutions
Configuration Available Modes Available Slit Widthsa
Resolving Power b
High_Medium nod on slitc, nod off slit, map 1.4 112,000
    1.9 86,000
    2.4 67,000
    3.2 50,000
High_Low nod off slit, map 1.4 112,000
    1.9 86,000
    2.4 67,000
    3.2 50,000
Medium nod on slit, nod off slit, map 1.4, 1.9, 2.4, 3.2 5,000-20,000d
Low nod on slit, nod off slit, map 1.4, 1.9, 2.4, 3.2 1,000-3,000d

a 1.400 slit unavailable >12 μm, 1.900 slit unavailable >16 μm, 2.400 slit unavailable >21 μm
b Observers must check the most recent resolving powers as a function of slit width and wavelength at http://irastro.physics.ucdavis.edu/exes/etc/
c On-slit nodding not possible at all wavelengths. Observers must check this at http://irastro.physics.ucdavis.edu/exes/etc/
d Resolving power is a strong function of wavelength and slit width

After passing through the slit wheel, the beam hits a flip mirror mechanism, which is used to choose between instrument resolution configurations (Table 4) by either directing the beam into the echelon chamber (high-resolution) or into the cross-dispersion chamber (medium- and low-resolution). In the high-resolution configuration, the beam enters the echelon chamber and expands to an off-axis hyperboloid mirror that serves as both collimator and camera mirror for the echelon grating. The dispersed light, focused by the hyperboloid, bounces off a flat into the cross-disperson chamber.

The cross-dispersion chamber is conceptually similar to the echelon chamber. The light expands from the input to an off-axis paraboloid that again serves as both collimator and camera mirror. The collimated beam is sent to the cross-dispersion grating which disperses the light in the plane of the grating. The camera mirror sends the light to our detector. When operating in single-order, long-slit spectral configurations -- our medium and low resolution science configurations -- the light never enters the high-resolution echelon chamber.

There is a wheel in front of the detector, which provides a lens for imaging the pupil through the instrument, and a dark slide for isolating the detector. The wheel would also be available for including transmissive optics to adjust the plate scale on the detector, if desired.

4.1.3 Detector

The detector is a Raytheon Vision Systems Si:As array with 1024x1024 pixels. The detector material is bonded to a SB 375 multiplexer. The array is mounted in a separate enclosure to reduce scattered light. The headerboard is thermally isolated from the rest of the optics box to permit active temperature control of the array. The photon fluxes in the low-resolution configuration ("Low") will be significantly above the level intended for the array. This prevents observations at longer wavelengths and/or with wider slits. When photon fluxes allow, only a subsection of the array will be clocked out in this configuration for a faster read-out (as well as in any imaging configuration). It is expected that a quarter to half the array will be utilized in these configurations, so the effective slit length is about 60''. These limitations to the Low configuration are not reflected in the online ETC yet, and GIs are encouraged to contact the instrument team for more information.

4.1.4 Performance and Sensitivities

EXES delivered performance appears consistent with expectations over the flight series so far. There are some variations from observation-to-observation, but we believe the values quoted here are fair estimates of what is typical. The angular resolution of EXES will match that achieved by the telescope. For the latest sensitivities, observers are recommended to consult the online Exposure Time Calculator (ETC) at http://irastro.physics.ucdavis.edu/exes/etc/. The ETC also provides the slit length as a function of wavelength and instrument configuration (and therefore whether on-slit nodding is possible or not), as well as the wavelength coverage in a single setting and echelon orders that can be targeted.

The wavelength coverage ranges from 4.5-28.3 μm. There are three resolution regimes -- high, medium and low -- with the exact resolving power depending on wavelength, grating angle and slit width. Generally, the resolution will be higher at shorter wavelengths in each regime. The high-resolution configurations will use the echelon grating and will achieve R = 50,000-100,000. If the cross disperser echelle angle is 35-65°, the configuration is called High_Medium and if 10-25° it is called High_Low. For these high-resolution configurations, there is non-continuous spectral coverage in high-resolution configuration for λ > 19 μm, but the central wavelength can be tuned so that lines of interest do not fall in the gaps. The Medium configuration will use high angles on the echelle grating to achieve R = 5,000-20,000, and the Low configuration will use low angles to achieve R = 1,000-3,000.

The High_Medium configuration slits are 4.5″-45″ long, and the High_Low slits are 1-12″ long. The shorter slits in High_Low allow for more orders to be packed onto the array, thus increasing the instantaneous wavelength coverage, while maintaining the high spectral resolution (see Fig. 4-6 for an example). In the Medium and Low configurations the slit lengths vary from 25″ to 180″ depending on the number of rows to be read out.

The sensitivity of the instrument is shown in Figures 4-1 through 4-4 for the High_Medium, Medium, and Low configurations for both point sources and extended sources. The Noise Equivalent Flux Density for S/N of 10σ in a clock-time (Note that the other instruments in this Handbook report sensitivities based on the total time on-source, not the clock-time. The latter includes the total time on-source + applicable overheads, excluding target acquisition and instrument set-up time.) of 900 seconds is plotted as a function of wavelength. These values have been calculated for a point source assuming image quality between 2" and 4" (FWHM) and the narrowest of the available 1.4" to 3.2" slits, both of which vary with wavelength, and take into account estimated instrument efficiency. They assume an altitude of 41,000 feet, 40° elevation, and 7 μm precipitable water vapor.


Minimum detectable point source flux

4.2 Planning EXES Observations

All EXES configurations and modes are released for observations. Observers are always encouraged to contact the instrument team for the latest performance results, however, in particular for the Low configuration, which suffers from saturation effects.

The proposer needs to supply the central wavelength, the spectroscopic configuration, the slit width, and the observing mode for each observation (Table 4). These parameters define the default instrument set-up. Each central wavelength specified should count as a separate observation. In addition, the proposer should estimate the clock time necessary to reach the desired S/N.


Minimum detectable point source flux

The calculation may be based on Figure 4-1 or 4-2 for point sources and on Figure 4-3 and 4-4 for extended sources, noting that the minimum detectable flux (S/N) / √(texp). However, it is recommended that the online ETC at http://irastro.physics.ucdavis.edu/exes/etc/ is checked as well for the latest updates.


Minimum detectable extended source flux

EXES operates in a wavelength region, parts of which are accessible from ground based telescopes. Proposers should carefully check the atmospheric transmission (using ATRAN, for example) and make sure that the observations require, or would greatly benefit from, using SOFIA. The proposer should take into account the Doppler shift of the target(s) due to their motion relative to Earth. If proposers find that the atmospheric transmission at the wavelength of interest is lower than the local median (calculated over a range ± 0.0125 μm), then more time will be required to reach the desired S/N. Higher transmission would imply shorter required times. In general the, S/N scales as transmission/√((1 - transmission) + 0.3). Note that the online ETC includes the impact of the atmosphere at precise wavelength of interest and so may differ from the Figures. The ETC provides the clock-time required to achieve the desired S/N per resolution element on a continuum object at the specific wavelength of interest and then indicates what the expected S/N should be for the entire setting.

Proposers should specify the slit width, which sets the resolving power for each configuration (Table 4). Note that the narrowest slit (1.4") is only effective below 12 μm (above this wavelength no gain in resolving power is achieved, while flux is lost with respect to the wider slits). Similarly, the 1.9" slit can only be used below 16 μm, and the 2.4" slit below 21 μm.


Minimum detectable extended source flux

In configurations using the medium resolution grating (Medium and High_Medium), the single setting spectral coverage ranges between 0.03 μm at the shortest wavelengths, and 0.3 μm at longer wavelengths (Fig. 4-5). For the low resolution grating (the Low and High_Low configurations) this is 0.2-0.8 μm. Note that while High_Low and High_Medium have the same spectral resolution, the larger wavelength coverage of High_Low comes at the expense of a smaller slit length, which is illustrated in Figure 4-6.


The single setting spectral coverage as a function of wavelength

Figure 4-5: The single setting spectral coverage as a function of wavelength. Note that these values are the same for the High_Medium and Medium configurations, and for the High_Low and Low configurations.

Proposers should choose a single line of interest for each observation. Fine tuning of the bandpass to observe lines at the extreme edges of a single setting should be done in consultation with the EXES team to see if existing data indicates such tuning is possible.

The slit orientation on the sky depends upon the time when the target is observed, and therefore the position angle cannot be specified.


Comparison of raw 2D spectra of EXES in the High_Med and High_Low configurations

EXES will not use the secondary for chopping in any of its observations. There will be two scientific modes - Nod and Map mode.

Nod mode: In this mode, the telescope is moved to a new position in order to remove the sky background. For point sources observed with a sufficiently long slit, the telescope is moved such that the object remains on the slit. For sources larger than about a quarter of the slit length, the telescope is moved such that the object is not on the slit. The time between telescope motions will depend on the sky variation, the telescope settling time, and the integration time. The goal is to maximize the signal-to-noise per clock time. For observations of point sources, the detectable flux plots (Figs. 4-1, 4-2) and the ETC include assumptions regarding whether nodding off the slit is required due to short slit lengths. For nodded observations of extended objects, proposers should contact the EXES team to check if nodding on the slit is possible. If not, the observing time required should be doubled. Unless specific nod parameters are requested for such observations, the instrument team will define the nod amplitude, direction and frequency. The sensitivities for extended source observations shown in Figures 4-3 and 4-4 assume that the source is nodded off-slit and take into account the variation in spatial resolution with wavelength. The atmospheric and overhead factors for nodding are included. If the source is small relative to the slit length, then it may be possible to nod along the slit. In this case, the source brightness given in the figures is for a SNR of 10 in 450 seconds.

Map mode: In this mode, the telescope is moved sequentially such that a series of positions along a straight line on the sky (a "stripe") are observed to create a map. The sky background is taken from the first positions and, depending on the size of the map, from the last positions. In general, we anticipate the telescope motions to be half the slit width to create a well-sampled map.

Proposers should specify the number of steps in a map and the step size. Map steps are generally assumed to be perpendicular to the slit. The first three positions for taking data in a map must be blank sky. These could be the first three positions of the map or at a separate sky offset position specified by the user. It is recommended that additional blank sky positions are observed at the end of the map on the other side of the object. For all maps, the instrument software returns to the sky offset position for three final sky observations at the end of the observation.

Proposers should specify the required clock time based on the flux limit desired, using the values in Figures 4-3 and 4-4, including any assumptions regarding binning of map positions to yield the final required SNR. The SNR for a single map position can be estimated by assuming that the required time is similar to that for nodding an extended object on slit, i.e. 10σ in 450 seconds for a given source brightness. If any spatial binning is required -- at least a 2-step sum is recommended -- then the SNR will improve by the square root of the number of steps in the sum. The online ETC allows the user to specify the number of steps and bins according to the predicted image quality in producing a clock-time estimate.

4.2.1 Wavelength Calibration

Wavelength calibration with EXES will be performed by applying the grating equation to atmospheric lines observed in the source spectra. As long as there is a single telluric feature in the bandpass with depth of at least 5%, the wavelength calibration is expected to be accurate to ≅ 0.5 km s-1.

If atmospheric models show no telluric features within the EXES instantaneous band pass for a given observation then obtaining a good wavelength solution will require a few additional steps. Note that the absence of telluric features from SOFIA suggests the observation may be better done from the ground. First, the order sorting filter (OSF) is rotated so that a different order from the echelle is observed that includes a suitable telluric line. The grating equation can then be applied, providing wavelength calibration accuracy down to ≈ 1 km s-1. The process of rotating the OSF, observing blank sky, and rotating the OSF back to the original orientation should take less than 5 minutes of additional time. A demonstration of this procedure can be found in Harper et al. (2009, ApJ, 701, 1464).

4.2.2 Flux Calibration and Atmospheric Line Correction

For every EXES science observation, the EXES temperature-controlled black body source and a nearby blank sky field will be observed. From these, a calibration spectrum will be constructed that, after division over the science observations,  will correct for response variations, and provide flux calibration. In principle, division by the calibration spectrum would also correct for telluric absorption lines (see Lacy et al., PASP volume 114, issue 792, p. 153), but this is presently not the case because of the large difference between the blackbody and sky temperatures. The flux calibration is expected to be better than 20%, but the true accuracy is currently uncertain. Experiments focused on line profile information and those that can normalize the continuum level, or use past observations for setting the continuum, will likely be more successful. Projects requesting a telluric calibration object, in particular those observing lines near strong telluric features or those observing relatively broad lines, will need to include the observation time required in their proposal. Because of the difficulty of scheduling a given telluric calibrator with the science target in a given flight, the specific calibrator will need to be chosen at the time of flight planning in consultation between the program GI, the instrument PI and the SMO support scientist. For wavelengths below 8-10 μm this will most likely be a hot, bright star (e.g., Vega or Sirius) and at longer wavelengths an asteroid. Galilean moons will also be considered, provided they are well separated from Jupiter.

For the proposal, a separate observation entry should be entered via SPT with name "Cal_target", where "target" is the name of the associate science target (i.e. "IRC+10216" and "Cal_ICR+10216"), and given the coordinates RA:12:00:00, Dec:+90:00:00. The observing time for such a telluric standard observation will depend on the instrument configuration and wavelength observed, as well as on the signal-to-noise level needed.

Proposers must use the EXES ETC to estimate this, assuming a continuum brightness of 100 Jy below 10 μm and 150 Jy above 10 μm for the High_Medium and High_Low configurations. For the Medium configuration, a brightness of 50 Jy should be assumed, and for Low, 25 Jy at all wavelengths. Proposers are urged to limit the EXES clock times on the telluric standard at a given wavelength and instrument configuration to less than about 30 minutes. Further improvement of the removal of telluric absorption features may be achieved by employing models of the Earth's atmospheric transmission.

4.2.3 Overheads

The treatment of overheads for EXES differs from that of most other SOFIA instruments. Instead of on-source times, users are required to specify "wall clock" times in SPT and SSPOT, which is the on-source time plus all overheads except those related to acquisition and instrument set-up. The overheads include time on "empty" sky in the off-slit nodding and mapping modes, as well as read-out and other telescope and instrument inefficiencies. The ETC and the Figures in this manual give the clock times needed. The figures in this manual only give clock times. SPT and SSPOT will add an additional 15 minutes for peak-up, wavelength optimization, flux calibration, and flat field overheads in the High_Medium, Medium, and Low configurations, and 20 minutes in High_Low. In all configurations, an extra 3 minutes of peak-up time is needed for the narrow (1.4") slit.

If no peak-up is necessary (e.g., after a wavelength change on the same target, if the source is extended, or if the continuum emission is too weak), the overheads can be reduced by using the 'alternative overhead' option in SPT and the "No Peak-Up" option in SSPOT. Overheads can also be reduced if multiple sky positions are observed in the same wavelength setting. In this case, click the "No Wavelength Setup" button in SSPOT. Note however that the time on a given target on a single flight is limited to 90-180 minutes, so full overheads may be needed again once the sum of AOR times exceeds 90 minutes. Conversely, if a single observation takes more than 90 minutes, it may need to be split into multiple AORs, each with full overheads. Please consult the EXES and SOFIA staff in these cases.

4.3 EXES as a Principal Investigator Instrument

EXES is a Principal Investigator (PI) class instrument. The proposer is encouraged (but not required) to contact the Instrument PI, Dr. Matt Richter (richter@physics.ucdavis.edu), before preparing or submitting the proposal, since the PI has the most up to date information about the instrument capabilities. However, it is recommended that they do so, in order to get the most up to date information about instrument capabilities. The data collection and reduction will be done by the instrument team, and it is expected that data analysis and preparation of the results for publication will be done by the General Investigator(s) in collaboration with the Instrument Team.