In most modes TEXES is background photon noise limited. Since the background brightness and the array efficiency vary considerably with wavelength it can be difficult to calculate the instrument sensitivity at specific wavelengths, but some general rules can be given. For more precise predictions proposers may be able to use the ITC timecalc.xlsx or are encouraged to contact TEXES team members (lacy@astro.as.utexas.edu and tgreathouse@swri.edu). (To use the ITC fill in the white rows of the Object Parameters section. The colored rows will be calculated.)
For observers who have observed their objects with TEXES on the IRTF or on previous Gemini runs it is best to scale from the performance there. For observations of point sources the sensitivity improves in proportion to the telescope area, or a factor of 7 going to Gemini from the IRTF. For extended sources the surface brightness sensitivity is unchanged, but the spatial resolution improves in proportion to the telescope diameter, or a factor or 8/3 and the instantaneous solid angle coverage decreases by a factor of (8/3)^2.
For observers who cannot scale from IRTF observations the variation in NEFD in hi-med and hi-low modes caused by the instrument response and telescope background can be approximated by NEFD[in Jy, 1 sigma in 1 sec on-source, averaged over 1 spectral resolution element (2 pixels * (lambda/10um) longward of 11um, and 3 pixels shortward), and optimally extracted from the echellogram] = lambda(in micrometers)/7. To get the noise in the observed spectrum after correction for telluric transmission one needs to multiply by [(0.07 + telluric emissivity)/0.07]^0.5/(telluric transmission). This calculation may require a high-resolution telluric model, which can be obtained from the TEXES team. There are two additional factors to consider. First, the sensitivity degrades toward the ends of echelon spectral orders and at wavelengths longward of 11um the ends of orders fall off of the array and are not observed; for complete spectral coverage two settings may be required. And second, if the slit is too short to allow nodding a source along the slit, which most often occurs at short wavelengths and in hi-low mode, the telescope must nod onto blank sky, increasing the observing time by a factor of 2.
In med and low resolution modes the NEFD should in principle improve in proportion to the square root of the resolution width, or by factors of approximately 2.2 and 5 respectively, but in practice telluric division and array performance are typically worse in these modes so the improvements are typically not this large. It should be noted that the spectral coverage in med mode is the same as it is in hi-med, and a hi-med spectrum can be binned down to med resolution giving approximately the same NEFD as if originally observed in med mode. The main advantages of med mode for observations of moderately broad spectral features are that there is no fall-off in the sensitivity toward the ends of hi-res orders or gaps between observed orders, and that the slit is much longer for mapping observations.
For observations of extended sources the surface brightness sensitivity, or NESB, can be calculated from the above formulas assuming the beam to be diffraction limited. The diffraction-limited beam size is 0.028"*lambda(um), which is approximately 2 pixels along the slit at 10um. Extended sources are often observed in a mapping observing mode so to calculate the observing time one must multiply by the number of steps in the map. Typically, sources are mapped by stepping over a distance somewhat greater than the object width and using measurements off of the ends of the scan to sample the sky. In this case it is not necessary to nod the telescope at each point in the map, improving the observing efficiency. This observing mode is referred to as scan mode. Typically, the telescope step size in a scan is about half the slit width in a direction perpendicular to the slit length.
To calculate the total time for a project one must include overhead for spectral setup, source acquisition, delay time each time the telescope is nodded either along the slit or onto blank sky, and flat field measurement. Proposers should include in their time requests the following: 1) 5 min for each change in spectral setup, 2) 10 min to acquire a new source, 3) 33% of the on-source time for delays after telescope nods, and 4) 16% of the on-source time for flats. Under good sky conditions the last two factors can be decreased by nodding less often and including more nod pairs in a file, but it is best to use these factors in the time request. Scan-mode observations involve less overhead as there is no delay between scan steps, eliminating overhead 3 and setting overhead 4 to 1 min per file, but adding a scan setup time of 5 min. If comparison sources are used their observing and overhead times must also be included in the time request, but we have found that division by a telluric model (which can be done by the TEXES team in the data reduction) produces as good of results as division by a comparison source. The exception to this might be for observations of spectral features substantially broader than the spectral resolution, which can be affected by fringing.
We should note that for scan-mode observations of planets it is not unusual for the objects to be bright enough that the time per step to be limited by observing efficiency rather than S/N (and the argument for using Gemini is to achieve higher spatial resolution the can be achieved on the IRTF, rather than higher sensitivity). In this case the time between steps should be set to 2 sec, or perhaps longer on faint and small sources.