A grism , or Carpenter prism, is a
transmission grating mounted on a prism that has its angle chosen in
such a way that the desired order of the grating passes through the
grism undeviated. Figure 5 shows a schematic
grism and the optical light path through it. We define 
to be
the prism angle, 
to be the grating groove angle which is also
the grating blaze angle in its reflection mode, 
to be the
deflection angle, 
to be the refractive index of the prism
material, and d to be the grating groove spacing.
In the simplest case for a grism 
. Then consideration
of the phase lag between adjacent facets of the grating in the prism
material and external to the prism shows that the undeviated
wavelength ( 
0) is given by:
where N is the order of the grating.
For deviated rays the grating equation becomes:
Geometrical ray trace shows that a ray with normal incidence passes
undeviated ( 0) when . That puts the
grating blaze at 0. For different from or
different from the refractive index of the replicating resin
used to produce the grating, the blaze is at a different wavelength,
but the above grating equations remain valid.
Figure 5: Optical diagram of a simple grism.
Thus the grating groove spacing and the prism angle for a given prism material define the location of the spectrum on the detector, and the grating groove angle and resin index define the grating blaze function.
The long slit J, H, and K grisms in CASPIR provide two pixel
resolving powers of 
500 over each of the respective photometric
passbands through a 1 
128'' slit. Wider long slits in
the Aperture Wheel can be used at the expense of spectral resolution.
Grism spectra should be recorded at two positions along the slit in an
ABBA sequence. This permits better sky subtraction than procedures
which require interpolation of the sky flux. The object can be nodded
between these two positions either by manually moving the telescope or
by acquiring data in Nod mode. Nod mode has the advantages that
nodding is performed automatically, the nod positions are defined as
telescope apertures so are unaffected by changes in the instrument
rotator angle, and the cumulative sum of the AB differences can be
viewed in the Run Display. The longest possible integration time is
required to overcome read noise. In practice, this means using an
integration time of 180 sec in method 2.
To use the grisms, first configure CASPIR for direct imaging with the
fast camera (0.5''/pixel scale) by typing FAST, and obtain an
image of a star in the Idle Display. Select your long slit by
typing, e.g.:
CASPIR/APERTURE=LSLIT1
Note the pixel positions of the center of the slit and the two desired nod positions, and then return to imaging the whole field by typing:
CASPIR/APERTURE=FASTCLR
Center the star on the slit-center pixel and redefine aperture A by typing:
APERTURE/HERE A
into the telescope console terminal. Manually move the star to the first nod position or offset the telescope EW by typing, e.g.:
OFFSET/SCALE 10.0 0.0
into the telescope console terminal and then define aperture N1 by typing:
APERTURE/HERE N1
into the telescope console terminal. Repeat this at the second nod position for aperture N2. Return the star to slit-center by typing:
APERTURE A
into the telescope console terminal and select Nod mode by typing:
CASPIR/MODE=NOD
into the CASPIR command terminal.
Spectroscopic observations are best performed using the Tip-Tilt system in either guide or correct mode. Operation of the Tip-Tilt system with CASPIR is more fully described in §3.13 and in the Tip-Tilt manual. To use the Tip-Tilt system for spectroscopic observations, roughly center the object on the slit-center pixel, then offset the IMB X-Y stage to the location of a suitable reference star by typing into the telescope console terminal, e.g.:
TIPTILT/OFFSET/SCALE -26.5 +10.5
if you know the reference star RA and Dec. offsets in arcsec on the sky, or
TIPTILT/COORD 12 26 25.9 -17 18 42 J2000
if you know the reference star coordinates, or
TIPTILT BS2015
if the reference star coordinates are in a telescope coordinate file, or
TIPTILT/TRACK
if you can use the optical image of the infrared object as the reference star, or
TIPTILT/HERE
if there is a suitable reference star within the Tip-Tilt acquire frame and you wish to drive the reference star to the correct subframe, or
TIPTILT/FIND
if there is a suitable reference star within the Tip-Tilt acquire frame and you wish to move the correct subframe to the pixel location of the reference star. Start correcting by typing:
TIPTILT/CORRECT
The Tip-Tilt system will pull the reference star to the center of the correct subframe, and hence may move the object out of the slit slightly. Recenter the object on the slit-center pixel by moving the Tip-Tilt sensor X-Y stage incrementally (using the above TIPTILT/OFFSET... commands) until the correct box center coincides with the slit-center pixel on the CASPIR array.
For small offsets ( 
) between the slit center and the N1
and N2 slit positions, it may be preferable to leave the IMB X-Y stage
fixed and move the correction subframe on the Tip-Tilt sensor during
each telescope offset. This is achieved by setting the TIPTILT
parameter and unsetting the STAGE_OFFSET parameter by typing:
CASPIR/TIPTILT/NOSTAGE_OFFSET
Larger offsets with the long-slit grisms will require IMB X-Y stage motions:
CASPIR/TIPTILT/STAGE_OFFSET
Now select the appropriate slit, grism, and the fast camera lens, and place the filter wheels in clear positions by typing, e.g.:
CASPIR/APERTURE=LSLIT1/UTILITY=K_GRISM/UFILTER=CLEAR/LFILTER=CLEAR/LENS=FAST
or
KGRISM
The Idle Display image will change to a spectrum and you can begin taking data.
Grism data can be obtained in either the Direct_Imaging observing
mode or the Nod observing mode. In Direct_Imaging mode, data frames
are obtained singly with one frame recorded for each REPEAT requested.
In Nod mode, successive exposures are obtained in an ABBA pattern at
two positions on the sky defined by telescope focal plane
apertures named N1 and N2. This permits accurate sky-subtraction
by differencing the images. The number of AB pairs obtained is set by
the REPEATS parameter which can be changed during data acquisition.
If a CASPIR/SHOW=CURRENT command has been given, the current
A-B difference is displayed in the Run Display. If the SHOW parameter
has been set to MEAN with a CASPIR/SHOW=MEAN command, the
average of the accummulated difference images is displayed in the Run
Display. This allows the observer to assess the quality of the full
dataset.
Data acquisition is started in Direct_Imaging or Nod mode by typing:
CASPIR/RUN
In Nod mode, each ``repeat'' consists of two runs (i.e., two recorded files) taken with the object at the N1,N2 or N2,N1 aperture positions. Set the ``repeats'' value to a large number initially, and reduce it to end the run.
The expected spectral images for common astronomical lines as well as Xenon and Argon arcs with each grism are shown in Figures 6 to 11. Tables of Xenon and Argon arc line wavelengths and OH airglow wavelengths (Oliva & Origlia 1992, A&A, 254, 466) can be found in Appendix J, as well as plots of extracted Xenon and Argon arcs for each grism. Xenon and Argon calibration lamps as well as an incandescent lamp are available in the calibration lamp module of the IMB. These are activated by typing any of:
IMB/CALIBRATION=XENON IMB/CALIBRATION=ARGON IMB/CALIBRATION=INCANDESCENT
It is recommended that lamp-on and lamp-off frames be obtained for the calibration lamps. The lamps can be switched off, without moving other components of the calibration lamp module but typing, e.g.:
IMB/XENON=OFF
The calibration lamps are switched off and the lamp select mirrors removed from the telescope beam by typing:
IMB/CALIBRATION=OFF
or just:
IMB/CALIBRATION
Wavelength calibration can also be achieved by measuring the compact planetary nebulae listed in Table 29 of Appendix J, or by recording mercury spectra of the fluorescent room lights.
Grism spectra are flat fielded using the incandescent lamp in the calibration lamp module. Record pairs of lamp on and lamp off frames, without moving the flip mirror between these frames. This can be done by typing:
IMB/CALIBRATION=INCANDESCENT
to turn the lamp on, and:
IMB/INCANDESCENT=OFF
to switch it off without moving the mirror, or by typing the DCL
symbols LON and LOFF into MOPRA. Note that these
abbreviations have a different effect to the same symbols typed into
the telescope console terminal.
A flat spectrum object must be measured to remove terrestrial
atmospheric absorption features from an object spectrum. A flux
standard must also be measured to flux calibrate the object spectrum.
Stars earlier in spectral type than F are preferred as calibrators for
the grism spectra because they have few intrinsic spectral features,
and the H 
ion bump around 1.6 
m is less pronounced than in G
dwarfs. However, it is necessary to also record a spectrum of a K or
M star in order to measure and remove the hydrogen absorption lines in
the early-type star spectrum. Lists of suitable stars can be found in
Appendix G. Plots of terrestrial atmospheric
absorption can be found in Appendix M.
Figure 6: Predicted astronomical (top) and airglow (bottom) spectra
for the J grism.
Figure 7: Predicted Xenon (top) and Argon (bottom) lamp spectra for
the J grism.
Figure 8: Predicted astronomical (top) and airglow (bottom) spectra
for the H grism.
Figure 9: Predicted Xenon (top) and Argon (bottom) lamp spectra for
the H grism.
Figure 10: Predicted astronomical (top) and airglow (bottom) spectra
for the K grism.
Figure 11: Predicted Xenon (top) and Argon (bottom) lamp spectra for
the K grism.