This page is intended to give the reader who is unfamiliar with the spectroscopy mode of GRIM enough basic information to be able to effectively plan observations, somewhere between a FAQ and the actual User Manual. Basic information about infrared spectroscopy is also provided for those users who are used to optical spectroscopy but may not have prior experience with spectroscopy in the near-IR regime.

A caveat about using GRIM for near-IR spectroscopy should be stated up front: the instrument was not designed to be a workhorse IR spectrometer. It was designed principally as an imager with spectroscopy as an interesting addition, albeit something of an afterthought.

The spectroscopic mode of GRIM uses a grating prism (grism) to preserve the in-line optical path from the entrance window to the detector. A slit is placed in the optical path near the grism; otherwise, slitless spectroscopy can be done in the "objective prism" mode. There are two significant consequences of using a grism rather than a more conventional grating. First, the resulting dispersion is not linear, since the dispersive effects of a prism and a grating are superimposed. Second, multiple orders are imaged simultaneously, leading to the need for a filter to be used in sequence with the grism and slit as an order-sorter. The following table gives lower and upper limits on the spectral window in spectroscopic mode for different f/ratio and filter combinations:

===============================================================================
	       J Band(Order 6) J Band(Order 5) H Band(Order 4) K Band(Order 3)	
Scale and Slit	(low)	(high)	(low)	(high)	(low)	(high)	(low)	(high)
_______________________________________________________________________________
f/5		0.844	1.301	1.013	1.561	1.266	1.951	1.688	2.602	
f/10		0.958	1.187	1.150	1.424	1.437	1.780	1.916	2.373	
f/20 Short	0.970	1.082	1.163	1.298	1.454	1.622	1.939	2.163	
f/20 Mid	1.015	1.127	1.217	1.352	1.522	1.690	2.029	2.253	
f/20 Long	1.068	1.180	1.282	1.415	1.602	1.769	2.136	2.359	
===============================================================================

All wavelengths are in microns. Note that two orders are present in J-filtered spectra.

Unlike DIS, the setup in GRIM for spectroscopy does not include a mirrored slit and slit-viewing camera; users familiar with DIS spectroscopy will find GRIM spectroscopy unfamiliar. A "View Slit" mode is available in the drop-down menu under the "Config" button in the GRIM control panel in Remark. This mode leaves the slit in place but removes the grism from the light path such that the undispersed beam from the slit is imaged onto the detector. To place an object on the slit, first take an image of the field in imaging mode, and compare this with an image in the "View Slit" mode (Note: You need only take one "View Slit" image on a given night as the slit positioning is highly repeatable.) Using the center of the slit determined by, for example, fitting a Gaussian profile, one can calculate the required offset between the slew position of an object and the position on the chip which places the object on the slit.

The slit is not aligned perfectly parallel to either the rows or the columns; the figure below shows the offset in f/5 mode:

Keep this in mind if you chose to dither the position of your object along the slit in order to make a master sky frame for calibration purposes. The derotation angle decreases quickly with f/ratio until at f/20, the slit is very nearly parallel to the CCD rows, so at higher resolutions this problem is less important. Another approach is to set a rotation of the instrument to compensate for the slit derotation, which then puts the slit along E-W. In the case of the figure, the derotation between the slit and the rows of the detector is approximately 1.4 degrees. Slewing to your field via Remark with a 0 degree object rotation places NSEW along the rows and columns with north up and east left; setting an object rotation of +1.4 degrees then makes the slit parallel to E-W.

As with GRIM imaging, three optical setups are available (f/5, f/10 and f/20). Each gives a different spectral resolution and effective slit length and width on the sky:

By f/20, the slit width on the sky is only 0.3 arcsecond, which is well below the typical site seeing in the near-IR (about 0.8 arcsecond) and even below the best seeing we ever achieve (about 0.45 arcsecond). Coupled with the lower throughput at f/20 limits the choice of targets to only the brightest. There are also corresponding problems with getting wavelength comparison spectra at f/20 due to throughput that should be considered when planning observations. Weigh these calibration issues against the higher spectral resolution.

In planning spectroscopic observations with GRIM, one should first begin by determining the spectral region(s) of interest. Next, a spectral resolution should be selected from among the f/ratios in the table above; this will be, of course, driven by science demands. Using this combination of spectral window plus spectral resolution, one can derive the required f/ratio and filter from the first table, above.

Here are some notes on GRIM spectroscopy from the instrument's early days. Most of the information is still relevant to the current configuration of GRIM.

Spectroscopy is a "mode" of GRIM much like imaging. By moving from one mode to another, you're changing the physical setup within the instrument. A grism/slit combination is mounted in the filter wheels allowing the two to be placed in the beam independently. This serves as a proxy slit viewing mode of sorts; by placing the slit in the beam without the grism, the projected position of the slit on the chip can be determined.

There are three steps to dropping your object on the slit to take spectra: locating the projection of the slit on the detector, taking an image to check the relative position of an object and the slit, and iterating the first two steps for optimal positioning. In the "Config" window of the GRIM control in Remark, select "View Slit" from the "Mode" pull-down menu. Take an image and load it into your favorite FITS viewer. Then put the instrument back into "Image" mode and take an image of sufficient duration to locate your object well but not saturating it. Note that if your object is extended or invisible in a relatively short-duration exposure that you will have to take more care in initially locating the object relative to the slit. Load the object image into another frame in your FITS viewer, then measure its pixel position with your cursor. Measure the center of the slit where you plan to place the object for your first spectrum, then calculate the offset required for the move. If you are taking spectra of multiple sources simultaneously, this is the time to set the required instrument rotation. It may take more than one iteration of the procedure for calculating the offset before you land your object(s) on the slit, but taking a short-duration exposure in "Spectral" mode once you think you're there will help confirm the pointing. Remember that there's no use in wasting a long-duration science exposure if you remain unsure about the relative position of the object and the slit.

Once your object is on the slit, taking spectra proceeds much like for spectrographs like DIS. Make sure "Spectral" is selected on the 'Mode' pull down menu in the GRIM control window, and remember to click "move", then "sync" to make sure the mode change was actually affected. Exposure times will be longer than for the same object in imaging mode, depending on the spectral energy distribution. It should be noted, however, that the sky levels will not saturate as quickly as in imaging mode, although individual night sky (e.g., OH) lines may. Refer to the published line lists and atlases of the night sky lines to determine which lines are present in the bandpass of the filter you use to select a spectral order, and how intense those lines are.

Wavecals can be obtained from the overhead HeNeAr lamps on the telescope truss, controlled with the xlamps application on tycho. As the f/ratio increases, the throughput is such that the exposure times for lamps lengthen considerably; by f/20, it is impractical to take Ar lamp spectra with the instrument mounted on the telescope. We find a much better arrangement of placing a portable lamp box with an Ar tube near the entrance window of GRIM itself. If you wish to take wavecals at high f/ratio, consult with your Observing Specialist to make arrangements.

Proper sky subtraction can be done easily without explicit sky pointings for a given object by taking spectra at several positions along the slit. The analogy here is observing in imaging mode with GRIM by using one of the so-called "dither" scripts that move objects around the GRIM frame to simultaneously record both object and sky information. There are no pre-built dither scripts for dithering objects along the GRIM slit in spectroscopic mode, but such observations could easily be scripted. With a sufficient number of pointings along the slit and enough of an offset between pointings, all object spectra for a given filter and exposure time can be median-combined (with a high/low sigma pixel rejection algorithm) to result in an average sky frame in spectral space. Simply subtracting this resulting sky frame off of the appropriate object spectra takes care of all your calibrations at once: bias, dark current, and flat fielding. Alternately, one can obtain explicit calibration frames and apply them to all the raw data before making the median sky image, arriving at the same result. Note, however, that due to software limitations, it's not possible to take true bias frames with GRIM, given that the shortest permissible exposure time is 1.2 s.

Dithering object positions along the slit is complicated slightly at the lower f/ratios by the fact that the slit is not parallel to the rows of the chip. Therefore, in instrument coordinates, a combination of both x and y motion is required to move from one location to another along the slit. (A plot of the GRIM slit center across the chip can be found here.) However, as mentioned previously, by performing a small rotation one can align the slit with E-W on the sky and then do offsets in object coordinates to simplify the dithering process to one dimension. Recent measurements of the slit position separated by several months show that the positioning of the slit relative to the chip is highly repeatable, so again there is no need to take repeated "View Slit" images throughout the night to confirm slit position.

Wavelength calibrations can be obtained in different ways. Argon lamp spectra can be taken off the mirror covers as with DIS; labeled plots taken with GRIM are available here. Alternately, night sky (OH 'airglow') lines can be measured in sky exposures to provide a wavelength reference. A spectral atlas of OH lines in the ~1 - 2.25 micron region is available in Rousselot et al. (2000). Another resource for both arc lamp as well as night-sky spectra in the near-IR is available at the Joint Astronomy Centre site. For on-site observers, we have this atlas and some additional print resources available in the 3.5m control room for on-site use or photocopying.

A set of line plots of argon lamp spectra taken with various modes of GRIM is available; the modes are JHKK'K_s at f/5, f/10 and f/20. 

The resulting images can be reduced as normal spectroscopic data. Be sure to set the FITS card "DISPAXIS" equal to 2, since the dispersion is parallel to the columns of the detector.