Multi-object Spectroscopy


IRIS2 MOS Announcement of Availability

The AAO is pleased to announce the availability of infrared multi-object spectroscopy (MOS) for its IRIS2 imager/spectrograph, with effect from Semester 2003A. This facility will permit the observation of up to three (3) IRIS2 MOS fields for each IRIS2 run (i.e. each time IRIS2 is installed on the telescope). IRIS2 MOS fields are ~4'x8' in size and can be used for spectroscopy of ~40 objects in the J, H, or K bands at R~2400. Because H-band spectroscopy requires a different grism and slit axis from that used for J/K, the same MOS mask cannot be used for all 3 bands without sacrificing significant wavelength coverage in H.

Astronomers wishing to propose for IRIS2 MOS must contact Chris Lidman (Chris.Lidman -@- in advance of submitting their proposal, to discuss their needs and the functionality of this developing facility. No proposal for IRIS2 MOS will be accepted, which has not been discussed with the AAO in advance.



IRIS2 MOS : Results from rho Oph in July 2002

Multi-object spectroscopy was attempted for the first time with IRIS2 on the Director's nights of July 24 & 25, 2002. Masks were manufactured for a 4'x8' field in the north-west extension of the rho Ophiuchi young star-forming region, using 2MASS photometry and astrometry. Successful MOS requires a precise estimate of the AAT's f/8 focal plane scale, at the lowered f/8 focus at which IRIS2 is used, and a precise estimate of the mask shrinkage from the room temperature it is manufactured at, to the 120K temperature it is stabilised at in the fore-dewar. For this run, we estimated these numbers by dead reckoning. For this run, objects were allocated down to J~17 or H~16.

Because this would be our first attempt with MOS, we were somewhat conservative in designing our masks - we used a 1.5" (3-pixel) slit width, and 10" (or 22.4 pixel) slit length, with nodding performed along the slit. In this way we could allocate ~25 slits (and 3 acquisition holes) on an IRIS2 MOS field. In future, the optimal use of IRIS2 MOS will probably involve pairs of 1"x ~3" slits offset on the mask, with nodding between pairs of slits for sky subtraction. This should permit the allocation of ~50 objects per mask.

Dispersion is achieved using the standard IRIS2 sapphire grisms. These deliver R~2400 (for a 1" slit) in the J, H and K passbands (depending on which of the J, H or K filters is used for order sorting).

The images below show the mask, and the field on the sky it goes with. Orientation is N to bottom, W to the left. Comparison of the two shows that we got our `dead reckoning' of the relevant plate scales wrong by just 0.4%. This is enough that, when objects are acquired to go down the slits in the field centre, they just miss the slits (in the E-W direction). However, now we know the offset to apply, we are confident we can get all the objects down all the holes for the next IRIS2 MOS run!


rho Oph mask with fiducial holes
rho Oph field in J
mask image


Which brings us to the actual data! Here is 50 minutes of on-sky data (1hr with overheads, with telescope nods every 5 minutes) in the Ks passband pair-subtracted. Only the objects in the centre of the field have decent S/N because they are the only objects which are centred on their slits.




IRIS2 MOS Details

IRIS2 MOS masks are installed in one of three "full field" positions in the IRIS2 slit wheel. This wheel is installed in the fore-dewar of IRIS2, which can be thermally cycled separately from the main-dewar (in which the grisms, filters, optics and detector are installed). This fore-dewar takes several days to thermally cycle, which imposes an operational constraint on the changing of masks, i.e. that masks can only be changed at the start of an IRIS2 run.

Masks are laser-cut from brass sheet, which is mounted on a frame which allows relief for the differential expansion that brass experiences relative to the aluminium slit wheel in which they are mounted. The cosmetic quality of these laser-cut slits is excellent. The smallest dimension which can be cut in this way is ~150 microns (or 1").

Because the AAT's Ritchey-Chretien design has no astrometric distortion over the field of view that IRIS2 sees, and a rigid mirror, the manufacture of masks for IRIS2 merely requires the provision of precise external co-ordinates. IRIS2 itself has significant astrometric distortion - around 1% at the field corners. However, the relevant focal plane for getting light through the slits in a MOS mask is that for the telescope, not the instrument.

This means applicants must be absolutely confident they have absolute astrometry for their targets with the image scale known to better than 0.1%, no radial distortion, and no boundaries from the mosaicing of separate detectors on other instruments, and positions good to +/-0.2" (at least). Checking this for your data against a trusted astrometric catalogue (e.g.. HIPPARCOS Tycho, USNO2, or 2MASS) is an essential requirement for IRIS2 MOS observing.

You must also select from exactly the same data set a sample of (somewhat) brighter stars for use in field acquisition. IRIS2 can easily detect objects at J/K~14 in 10s acquisition images. It is absolutely essential that these fiducial stars be on the same co-ordinate system as the spectroscopic targets, or else you will be wasting your time.

Currently, the only field orientation offered for IRIS2 MOS is with the slits running N-S (i.e. dispersing light E-W, instrument PA=90), or with the slits running E-W (i.e. dispersing light N-S, instrument PA=180).

Mask design and fabrication must currently be overseen by the IRIS2 MOS astronomer. Mask designs are created using in-house AAO-developed code. So observers will have to provide target selection lists, and the AAO will create mask designs, which must be checked by the observers for the suitability of allocations. This will usually require several iterations, so this process must start well in advance (at least 8 weeks) of scheduled runs. Observers who fail to fulfil these requirements will have their time returned to ATAC for reallocation to other programs.

The available field of view is ~4' x 8', though if full wavelength coverage is desired in each spectrum, this may have to be limited to more like 2.5' x 8'. Experience with the best field dimensions for each passband is being gained "as we go", so consult the IRIS2 MOS astronomer for more details. Keep in mind also that because H-band spectroscopy with IRIS2 requires a different grism and slit axis from that used for J/K, the same MOS mask cannot be used for all 3 bands (without sacrificing significant wavelength coverage in H). Thus, any set of 3 masks would allow you to observe 3 different fields in J+K, or in H; or one field in J+H+K, and one field in J/K or in H.




Assuming your masks have been correctly designed, and are in place in the IRIS2 dewar, the following steps describe how to acquire your field. The basic idea of acquisition is to take an image of the field and an image of the mask, then calculate and apply the correct rotation and offset to match the two together. In practice this can be accomplished with greater ease by following an iterative procedure, taking one step at a time.

You will need to define an aperture set to use based on your mask. The aperture set contains the offsets applied to the telescope to switch between beams A and B. The offset between A and B should be about half the length of the shortest slit on the mask. Nods up and down the slit in a North-South direction correspond to an offset in the X direction in the aperture set. Offsets in the aperture set are in mm at the focal plane, and the scale at the focal plane in Cassegrain focus at f/8 is 6.6"/mm.


The acquisition of your field is considerably simplified by the use of the MOS-acquisition tool, available as part of Chris Tinney's version of Skycat. To run this version log in as aatobs, on the aatvme14 console and type

aatvme14{aatobs}: cgt_skycat_test

aatvme14{aatobs}: skycat & 

As the near-infrared sky is bright the following steps may be easier to perform if you subtract an appropriate bias image taken at the reference axis using File...Bias Image in the Skycat GUI.

  1. After you have the correct aperture set loaded slew to a SNAF star, go to aperture A and take a short exposure. Use View -> Pick Object to select the SNAF star and get a good centroid on it. This position is the axis of rotation for aperture A, which will be used later so make a note of it now.
  2. Now we want an image of the mask. Configure IRIS2 and expose a short image through the mask and identify the fiducial holes. It may be useful at this stage to select on of the holes as Mark 1, by Shift...Left Click...Set as Mark 1, as this will allow a rough offset to be made by moving the relevant fiducial star to this mark.
  3. Now open the MOS-Acquisition tool in Skycat, Telescope -> MOSAcquire. Use View -> Pick Object to centroid each of the fiducial holes. As the profile of a fiducial hole is a top-hat function, rather than a Gaussian, a much more accurate centroid is found by zooming out on the image panner in the Pick Object GUI. After picking each hole enter it into the MosAcquire GUI, by means of theAdd Hole button. This will mark the position of each hole with a green circle, and number them in the order in which they were picked. This is because the fiducial stars must be chosen in the same order as the holes.
  4. Next slew to your object field, go to aperture A, and take an image. Identify the fiducial stars. If there is a large offset between the fiducial stars and the marks identifying the holes, then you may want to offset the telescope by selecting the appropriate star and then Shift...Left Click...Move object to Mark 1 and take another exposure.

    If the stars are reasonably well aligned with the holes then ask the Night Assistant to start guiding. Use the Pick Object GUI to centroid the fiducial stars and use Add Star to add them to the MosAcquire GUI, in the same order as the holes. A red diamond will be displayed marking the positions of the stars.

  5. Now we want to calculate the necessary rotation to align the field and the mask. Enter the axis of rotation into the MosAcquire GUI. Click Calculate Offset and Rotation. This brings a pop-up window with the derived rotation and offsets. Also shown is the RMS value. Check that this is reasonably small (less than 1). Have a look at the individual RMS values for Star/Hole pair, in the bottom table of the MosAcquire GUI. There is a column listing the RMS error without that Star/Hole pair. If the RMS could be significantly improved by ignoring that pair, then check the centroiding of the hole and the star.

     Once you are happy with the derived offset and rotation click Apply dervied offset and rotaion. This brings up a window with the necessary rotation. Ask the Night Assitant to rotate the Cass cage by this amount.

     Next a window will appear with the derived offset. It is better at this point not to apply the offsets. The calculated offsets are more accurate when the rotation angle is very small. Take another image and repeat this step.

  6. When the derived rotation is negligible (less than 0.1 degrees) repeat the above step, do not apply any rotation but do apply the offsets. If the offsets are small and you are guiding it is probably better to ask the Night Assitant to make the necessary offsets.

     At this point the mask should be correctly aligned with the field. How accurately you are able to do this will depend on the mask design and the seeing at the time.

  7. Take an image through the mask. With any luck you should be able to see the fiducial stars shining through the fiducial holes. You can make fine adjustments at this point by simply looking how well the stars and holes are aligned. If they are all off in a certain direction, work out the necessary shift (1 pixel=0.45") and ask the Night Assistant to apply it.
  8. Check the alignment of the slits. Take an image at A and an image at B, and subtract one from the other using File...Bias Image in the Skycat GUI. Hopefully you should see the positive and negative images of the objects nicely centred on the slits. If you can't it may be necessary to tweak the aperture set paramters as in point 1. Remeber to alter the B aperture since all the acquisition was done in aperture A, so this should be correct.

Good luck! Acquiring the field for MOS is a fiddly business, but by performing each step carefully and not trying to do the whole operation in one go, good results have been had. Iterate each step until reasonable accuracy has been achieved before moving on to the next step.

 After the field has been acquired and observing is underway, it is a good idea to check the telescope is nodding properly. After an hour or half an hour repeat step 8 and check the objects are still centred on the slits. Depending on how this goes you may then choose an appropriate interval to check the alignment.


The MOS Acquire Tool