CTIO IR Spectrometer Commissioned (1Sep94) (from CTIO, NOAO Newsletter No. 39, 1 September 1994) The CTIO IR Spectrometer was successfully recommissioned with a 256 X 256 InSb array during the last two weeks of July. Engineering runs were carried out on both the 4-m and 1.5-m telescopes, as well as observations of Comet Shoemaker-Levy 9 impacting Jupiter (on the 1.5-m telescope). Scientific results are described elsewhere; in this article we concentrate on describing the instrument performance. Because of the Newsletter deadlines, the performance described here is based primarily on the earlier run, which was on the 1.5-m telescope. The IRS upgrade project was carried out jointly by staff at KPNO and CTIO. It is our expectation that most (though not necessarily all) major CTIO infrared instrument projects will be carried out in this way, with technical staff in Tucson contributing a proportion of the resources needed to carry out the project. In this particular case, the mechanical modifications of the IRS were done in Chile, with Andres Montane as the mechanical engineer. The modifications were needed to mount the new array and to reduce light leaks and scattered light inside the instrument. We were also able to enlarge the usable slit slightly. Fabrication of the WILDFIRE electronics for the new array was carried out in Tucson under the supervision of Jerry Heim. The electronics also provide control of the IRS stepper motors. The instrument was shipped to Tucson for installation of the new array. Manuel Lazo (from the Tololo electronics staff) travelled to Tucson to assist with the installation. Richard Elston was the CTIO scientist in charge overall; he also was heavily involved in the details of the installation and commissioning. The additional software needed for WILDFIRE to run the instrument, drive the motors, and communicate with the telescope was provided by Nick Buchholz. Both Buchholz and Heim travelled to Chile for the commissioning. In addition to the people listed above, many others contributed to the success of the upgrade. We would like to mention particularly Rolando Rogers, from CTIO ETS, who helped with final commissioning, the La Serena mechanical shop, whose ability to make random bits of black-painted metal at the last minute was critical, Paul McIntyre and Rich Land of the Tucson IR group, who worked on construction and testing of the new electronics, and last but not least both Clark Enterline and Hernan Bustos, whose ability to deal with the bureaucracies of two governments helped the instrument on its travels. Instrument Description Past users will already be familiar with most aspects of the instrument (and those who are not may also find helpful information in the article by DePoy et al. 1990: PASP 102,1433). The IRS is a cryogenic spectrometer with a beam size of approximately 62 mm. It is operated at f/30, and therefore can be used on the 1.5-m and 4-m telescopes only. Two gratings can be installed in the instrument at any one time, allowing the user a certain flexibility. A filter wheel holding up to eight blocking filters allows for order selection. The main internal change from the previous incarnation of the instrument is that the detector is now a 256 X 256 SBRC InSb array, replacing the old 58 X 62 array. The new array has smaller pixels (30 um vs. 76 um), which leads to a corresponding decrease in pixel size on the sky (since the optics have not been changed). One other change that has been made is a slight increase in physical slit length. As a result, the total usable slit length in pixels is much greater, about 50 pixels. The pixel scales are 0.94 arcsec/pixel on the 1.5-m telescope, and 0.32 arcsec/pixel on the 4-m telescope. Anamorphic demagnification increases these values in the dispersion direction for the higher resolution gratings, by 30% or so. The smaller pixels provide a much better match to typical image quality on the telescopes, while reducing overall background. The old array electronics have been largely replaced as well. The new WILDFIRE electronics interface to the telescope Sparcstations; this system should be familiar already to users of the KPNO IR instruments. The IRS is mounted on a dichroic box, which allows viewing of the main field with one of the CTIO CCD TV's while observing. This simplifies acquisition and guiding, provided the object is visible or there are visible reference stars nearby. On the 4-m, the automatic guider can be used to acquire guide stars over a larger field; the guide probe can also be set to move automatically to compensate for small telescope offsets (for example, moving an object along the slit or moving off to sky). This capability does not exist on the 1.5-m telescope. Acquisition of objects can also be done using the instrument in imaging mode, either using zero order of the grating or a mirror mounted on the grating table. (At this time the optical quality of the mirror is unacceptable with the new, smaller pixels; we hope to replace it soon. The image quality in the grating zero orders is good, and sensitivity is adequate for all but the very faintest objects.) One can view through the slit, or through a large hole (diameter approximately 45 pixels) for initial acquisition. Currently Available Gratings Although only two gratings can be installed in the IRS at any given time, CTIO has a larger number of gratings available. Installation and removal of gratings is a process that requires warming up and cooling down the instrument. As this procedure takes roughly two days, users must specify the grating or gratings they will use at the time they request telescope time; changes in grating configurations once the telescopes have been scheduled may not be possible. All of the gratings can be used in the "I" band (0.9-1.1 um) although the 12 l/mm grating has a very limited free spectral range by that point, as it is being used in 6th order. Grating Resolution (pixels) Grating Blaze^a l/mm (um) J(1.2um) H(1.6um) K(2.2um) L(3.5um) M(4.7um) 632 2.4 9800 5370 8370 N. A.^b N. A.^b 210 4.2 3860 5350 4830 3760 5240 75 4.5 1800 1800 1650 1320 1760 75 1.9 900 600 825 1320^d N. A.^c 12 6.7 365^e 390^e 400^e N. A.^c N. A.^c a Blaze is given as for IRS configuration, not Littrow. b Grating cannot be used at this wavelength as ruling is too fine. c Background count rates are too high to permit use at this wavelength; array saturates in minimum read time. d Not recommended; grating is being used off-blaze. e Free spectral range at this resolution does not fill entire array. Other Parameters The blocking filters contained in the filter wheel are usually for the J, H, and K windows, plus one for the combined L and M windows, plus either an I blocker or a blocker for the combined I and J windows, which is used with the 632 l/mm and 75 l/mm 1.9 micron blaze gratings. Finally there is a Ks filter, which is used for acquisition in imaging mode. Several slits are available. These include slits of 0.35 mm width (2 pixels), 0.50 mm width (3 pixels), and wider slits (1 and 2 mm) that are useful for setting spectra zero-points. The usable slit length is almost exactly 50 pixels. In addition, there is a large hole intended for acquisition; its field and pixel scale are roughly the same as those on the old CTIO IR Imager. Measured Performance The second table lists measured sensitivities for gratings. The figures are for the 1.5-m telescope, using a wide slit. Signal levels for a slit matched to the resolution (2.0 arcsec) were 85% of these values on the night we measured performance; this value is probably typical for the 1.5-m. For gratings not measured, the sensitivity should scale approximately as resolution, to within about 30%. Sensitivities within individual bands also vary, in large part because in any given window the resolution in wavelength is constant, so the photon rate goes as lambda^-3. Sensitivity on the 4-m telescope should be roughly 6 times higher, without allowance for light losses at the slit. A 2-pixel slit is 0.7 arcsec; in typical seeing of 0.8 arcsec FHWM this should lead to light losses of somewhat less than 50% at the slit. Background levels are given in the following table, for wavelengths in the K, L, and M bands. At shorter wavelengths, the background is entirely due to airglow lines, so that the concept of "average" background is not particularly useful. At the lower resolutions, it is probably reasonable to use a count rate similar to that in the K band for estimating H-band exposure times. Count rates in the J-band will be several times smaller. For the highest resolutions, there is enough space between the strong airglow lines to consider ignoring them, although strong lines still occupy some 20% of the spectrum, and one should make sure in advance that key features don't coincide with strong airglow lines. Electrons/Sec/Pixel for 10th Mag Star^a Grating 1.08um 1.25um 1.65um 2.20um 3.50um 4.65um 210 l/mm 60 30 40 50 30 8 75 l/4.5um 130 175 165 170 70 25 a Gain is 15 electrons/ADU. Sensitivities are determined summing all pixels with signal perpendicular to the dispersion (usually three). Background Count Rates (e-/sec/pixel) Grating 2.20um 3.50um 4.65 um 210 l/mm 4 6 X 10^4 2 X 10^5 75 l/4.5 um 20 6 X 10^4 7 X 10^5 Please note that the rates given above are per pixel, whereas the count rates for an object are given summed along the slit, and thus involve several pixels (typically three). We also don't understand why the 3.5 um background with the 210 l/mm grating isn't lower; it's not clear whether this is real or a mistake of some sort in carrying out the tests. The minimum background is the internal background within the instrument, which is a combination of low-level light leaks and dark current. The internal background ranges from 5-10 e-/sec/pixel, and is mainly due to dark current from the array. Read noise for the array is roughly 30 e-; this is achieved with multiple reads of the array. Well size is about 120,000 e- with the normal array bias, but the array is becoming non-linear at lower count rates. We recommend keeping maximum counts/pixel below 5,000 ADU (75,000 e-) in order to keep linearity corrections reasonable. For high-background applications, the bias can be increased to give a larger well depth (240,000 e-) at the expense of higher dark current (roughly 100 e-/sec). Note that for the read noise and background levels quoted above, the integration time at which noise from accumulated internal background equals read noise is roughly 3 minutes. For most applications, the improvement in signal to noise over the previous array at the same resolution (not the same grating) will be a factor of 2-3. Spectral coverage is of course 4 times greater, so that for projects requiring coverage of more than one feature, gains in efficiency can be as much as a factor of 20. Since the new array does not suffer from the persistence problems that the old array had, and also appears much more stable, the reliability of results on faint objects is likely to be further enhanced. Sample Results The article on Comet Shoemaker-Levy 9 shows an example of a 3 um spectrum obtained with the IRS. In the figure below, we show some examples of long-slit spectra of the Galactic Center, where the slit was approximately centered on IRS 16. The scale along the slit is 0.9 arcsec/pixel, while the scale along the dispersion direction is about 30 km/s/pixel (the exact value differs from line to line). One easily sees the complex velocity structure in the hydrogen lines in the immediate vicinity of IRS 16; this structure is not seen in the H2 S(1) line, and is present only weakly in the [Fe II] 1.644 um line. [Figure not included] Planning Proposals If you are planning on submitting a proposal to use the IRS, the most important decision (once you have decided on the science) is your choice of gratings. The two factors that affect this are wavelength coverage and resolution. If you need data at or beyond 3 um, then you must use either the 75 l/mm, 4.5 um blaze grating or the 210 l/mm grating. If not, then you are probably better off using one of the gratings blazed for shorter wavelengths, as they have somewhat higher efficiency and don't have order overlap problems in the I and J windows. In particular, the 632 l/mm grating is probably preferable to the 210 l/mm grating for high resolution applications, unless the difference in spectral coverage turns out to be important. The 75 l/mm, 1.9 um blaze grating is probably preferable to the 12 l/mm grating unless you really need to cover a spectral window out to its extreme limits: for example, at K the 75 l/mm grating provides somewhat more than 0.3 um coverage, which would allow you to observe from the He I line at 2.058 um out past the first CO bandheads. Another possible low-resolution option is the cross-dispersed grating (read below). Coming Attraction: Cross-Dispersed Grating We are in the process of ordering a low-resolution grating that incorporates a prism cross-disperser. This combination could be installed as one of the two gratings in the instrument; the other choice would be another grating, run without cross-disperser. The cross-disperser would put several orders of the grating on the array, providing full spectral coverage from roughly 0.9 um to 2.5 um, at a resolution of about 500 (slightly greater than the present 12 l/mm grating). The usable slit length would be reduced because of the multiple orders on the array, to about 30 pixels. This grating would replace the 12 l/mm grating (one could get back the longer slit using a J, H, or K blocking filter for coverage of one order at a time). We would expect gains in efficiency of roughly a factor of 2 (because K spectra usually take longer than J and H when done separately). It might also be more attractive than the 75 l/mm, 1.9 um blaze grating for some applications. While it ought to work, we are not in a position to guarantee availability, let alone specify performance. Therefore, anyone whose program requires the cross-dispersed grating should contact one of us in late September to find out the status of the grating and may be advised to wait until second semester 1995. Anyone whose program would benefit from the cross-dispersed grating should specify both it and a back-up on their observing proposal, and should provide calculations of exposure times, etc. appropriate to the back-up configuration. Richard Elston, Jay Elias, Brooke Gregory