OSIRIS Hieroglyph The OSIRIS User's Manual

Spectroscopic Mode Overview

OSIRIS has three spectroscopic modes: low-resolution (R~1200) and high-resolution (R~3000) long-slit modes plus a low-resolution cross-dispersed mode (R~1200) with a short slit. Separate slits are provided for each mode as described below.

OSIRIS is put into spectroscopic mode by retracting the prefilter wheel and pupil mask, and then flipping the grating into the beam in place of the imaging flat. This operation requires approximately 2 minutes to complete. The imaging flat may be flipped back into the beam rapidly (3 seconds) to provide a quick-look Acquisition Mode to center the object on the position of the slit (either by taking an image of the field or viewing through the slit).

This section describes:

Spectroscopic observing strategies and techniques (including target acquisition and calibration issues) are discussed in the section on Spectroscopic Observing.

OSIRIS Spectroscopic Modes

OSIRIS has three spectroscopic modes:
R~1200 single-order long-slit spectroscopy covering all of the J, H, or K bands in a single spectrum. The Low-Res long slit covers the entire width of the field of view of the f/2.8 camera at a fixed width. The grating tilt should be set at 950 in low-res mode.

R~3000 single-order long-slit spectroscopy covering most of the J, H, or K bands. The wavelength region adjusted by tilting the grating. The High-Res long slit covers approximately 80 arcseconds by 0.48 arcseconds.

Low-Resolution (R~1200) multi-order cross-dispersed spectroscopy with a short slit and the f/2.8 camera that covers all of J, H, and K bands simultaneously in adjacent orders. The cross-dispersed slit is only ~15% the height of the low-res long slit at a fixed width.

All modes use a single diffraction grating blazed at 6.6-microns in first order, which places the K-band in 3rd order, H-band in 4th order, and J-band in 5th order. A grism and JHK filter (integrated into the short slit) are used to cross-disperse orders 3,4, and 5 (and part of order 6) across the detector.

The High- and Low-resolution modes are selected by changing the camera and slit, and using one of the filters as an order sorter. Cross-dispersed ("X-Disp") mode is selected by introducing the cross-dispering grism (in the filter wheel) and short slit into the beam. This slit has an integrated JHK filter to provide blocking outside of the bands of interest. The table below summarizes the camera, slit, and filter configuration used by each of the OSIRIS spectroscopic modes.

Table 1: OSIRIS Spectroscopic Configurations
Mode Camera Slit Filter
Low-Res f/2.8 0.34mm x 63mm
(LoRes Long)
J, H, K or LPK
High-Res f/7 0.14mm x 24mm
(HiRes Long)
J, H, K or LPK
X-Disp f/2.8 0.34mm x 8.6mm
(XDisp Slit)

The order separation filters are as follows

The following table summarizes the basic properties of the OSIRIS spectroscopic modes. The properties of the slits depend on the telescope, and are summarized in below.

Table 2: Spectroscopic Parameters
Mode Filter
Wavelength (Å)
Linear Dispersion
Wavelength Range
(2 pix)
Low-Resa J (5) 12500 5.525 2500 1200
  H (4) 16500 6.906 3000 1200
  K (3) 22000 9.208 3800 1200
High-Res J (5) selectb 2.208 1435 3000
  H (4) selectb 2.762 1795 3000
  K (3) selectb 3.683 2394 3000
X-Disp JHKc (5-3) n/a (low-res)d 10000-24000
(3.5 orders)

Table Notes:

a) Low-Res mode central wavelengths are given for the optimal grating tilts (see Table 4) that center the entire band within the limits of the filter transmissions.

b) Center wavelength is selected by tilting the grating (see below), within limits prescribed by the filter transmissions and the physical limits of the grating travel.

c) The "JHK" filter used in X-Disp mode is integrated into the cross-dispersed slit, and not in the filter wheel proper (which is occupied by the cross-dispersing grism).

d) Dispersion and resolution depend on the order, and are the same as for the individual low-resolution orders given in the table.

e) Because of vignetting by the f/7 camera, the true coverage for extracted spectra should be restricted to pixels between about 250 and 900 even though there will be dispersed light outside this range.
[ Top of Page | Contents ]


Slit widths and heights are fixed (pre-cut masks) as described in Table 3 below, and depend on which telescope you are using.

Table 3: Slit Properties
  Physical CTIO 4-m CTIO 1.5-m
Slit Width Height Width Height Width Height
Low-Res Long Slit 0.34mm 63.0mm 1.2" 216" 3.6" 666"
X-Disp Slit 0.34mm 8.64mm 1.2" 30" 3.6" 90"
High-Res Long Slit 0.14mm 24.0mm 0.5" 81" 1.5" 254"

The Low-Res and High-Res long slit heights span the entire unvignetted imaging fields of the f/2.8 and f/7 cameras, respectively. The X-Disp slit is only 14% the length of the long slit so that the different orders can be stacked on the array without overlap.

Slit Orientation

The slit is oriented E-W on the 4m in the standard installation. By default,the instrument rotator = 270 degrees when OSIRIS is mounted. We have measured the true E-W position to be 271.5 degrees. The mounting of OSIRIS seems to reproduce this value to within about 0.5 degrees. Therefore, you should ask the Night Assistant to adjust the rotator to this angle at the start of your run. You can then check the accuracy by doing a slitscan in Acquisition mode, measuring the stellar Y pixel position. The instrument may be rotated to larger angles, but this should be arranged prior to an observing run.

[ Top of Page | Contents ]

Selecting Wavelengths

The wavelength region is selected by tilting the diffraction grating. How much you need to tilt for a given configuration is described for each mode below.

Low-Res Long-Slit Mode

Since all of the given band is on the detector at once in low-resolution long-slit mode, the optimal grating tilts to center each band have been measured in advance and are summarized in Table 4. In general, it is not necessary to set the grating tilt to other than these settings.

The "center" here is defined as the wavelength at row 575 (the center of the f/2.8 imaging mask) in a vertical cut made along column 515 (the center of the low-resolution long slit). The slit is curved as seen on the detector such that the central wavelength of an object shifts to the RED as its position moves toward the end of the slit from the center. The shift is about 10 pixels from center to end.

Table 4: Low-Res Mode Grating Tilts
(1999 Jan 25)
Filter Order Tilt Central
J 5 950 12500Å
H 4 950 16500Å
K 3 950 22000Å

If adjustments are required, the general rules of thumb for tilting the grating in low-res mode are as follows:

Cross-Dispersed Mode

Since all of the principal orders (J through K) are on the detector at once in X-Disp mode, the optimal grating tilt for this mode has been measured in advance. We do not recommend using a different grating tilt in X-Disp mode unless there have been serious mechanical changes in the instrument (at which point the support scientist should be immediately informed of this).

X-Disp Grating Tilt: 950
Prospero Command: XDSPEC
In general, the optimal tilt for X-Disp mode is selected so as to best center 4th order (H-band) on the array. The adjacent orders (J and K, respectively), sort themselves out across the detector with minimal losses off the edges of the field of view (most of these regions are in bad atmospheric windows, anyway).

See the Cross-Dispersed Mode Order Map for details of how the orders are arranged across the array and descriptions of some of the other stuff you might see.

High-Res Long-Slit Mode

It may be necessary to tilt the grating in order to cover the alternate ends of the JHK bands in high-resolution long-slit mode. The "center" for high-res mode is defined approximately as the pixel at row 560 (center of the f/7 camera imaging mask) in a cut along column 500 (the center of the high-res long slit). Since the image of the slit is slightly curved, the central wavelength of an object will get REDDER (the wavelength at constant y pixel gets bluer) as you move out from the center to the end by a little over 4 pixels (8-12Å, depending on the order).

The grating tilt can generally place the wavelength of interest within ~1 pixel of the center (as defined above).

In general, the rules of thumb for adjusting the grating in high-res mode are as follows:

[ Top of Page | Contents ]

Spectroscopic Performance


Accurate sensitivity numbers have not yet been measured. Proposers can make rough estimates for signal to noise calculations by using the IR spectroscopy Exposure Calculator. OSIRIS was designed to work at f/17, hence the collimated beam overfills the grating at f/14.5 resulting in a loss of throughput (presently estimated up to factor of ~2). It is hoped that for many programs, the combination of increased wavelength coverage at higher spectral resolution (relative to the IRS) and increased visibility between the OH sky lines for the XD mode will make up for this.

Grating Stability & Repeatability

Tests show that once a grating tilt has been selected it will be stable to within 0.05 pixels RMS if not moved. If you are only going to be using one spectroscopic setting for a night (e.g., only taking cross-dispersed spectra), it is best to set the grating once and then leave it alone, unless you have evidence that the grating has been moved.

If you change the grating tilt, even setting it to the same tilt as it was before, this operation involves some mechanical imprecision in the repeatability of the encoders. Typically, if you ask the instrument to reset the grating to its home position and then tilt it to a desired setting, the grating will be driven to within ~1 pixel RMS in high-res mode, and ~0.4 pixels RMS in the low-res and X-disp modes. The non-repeatability in tilt is due to the discreteness of the tilt encoders (recall that 1 encoder step moves a line by 2.5 pixels in high-res mode, and 1 pixel in low-res mode). You will probably have to take new standards and wavelength calibration frames if you move the grating and restore it to the same position. New flats will probably not be necessary.

There are occasional mechanical glitches when flipping between spectroscopic and acquisition modes that can introduce approximately 0.3 pixels of shift in the high-res mode, and 0.1 pixels in the low-res mode. It is not repeatable, and appears as a slight jump in the spectral lines on close inspection. If you see a "large" jump of 2-3 pixels, you should reset the flip by typing ACQMODE then SPMODE to "cycle" the grating/mirror flip mechanism. These glitches are more likely to occur if the instrument has not yet reached equilibrium after being filled following warm up, or if it has run out of cryogen and is starting to warm up.


To be added.

[ Top of Page | Contents ]

Updated: 1999 February 16