The AAOmega IFU will provide a powerful integral field capability with the existing SPIRAL IFU, with modifications to the slit end to feed the new AAOmega spectrograph. The AAOmega IFU will keep the SPIRAL lenslet array, feeding 512 optical fibres (32x16); see Figures 1 and 2 below. Each lenslet/fibre is 0.7"x  0.7" on the sky, critically sampling the median AAT seeing of 1.4", and the total field of view is 22"x11". The AAOmega IFU will normally be at the Auxiliary Cassegrain focus (spectropolarimetry, if available, would use the Normal Cassegrain), with 18-20m of fibre feeding the AAOmega spectrograph in the Coude room.

It is expected that Nod&Shuffle will be the default mode of operation for the AAOmega IFU, giving the best sky subtraction performance (Poisson-limited from SPIRAL experience). However, for objects which don't fill the IFU f.o.v., it will also be possible to designate regions of the IFU to use for sky-subtraction, thus allowing observers to avoid the Nod&Shuffle overheads. All 512 IFU fibres will feed the AAOmega spectrograph, thus the spectra will be separated by ~8 pixels in the spatial direction (4000 pixels). The spatial FWHM of the IFU spectra is 2.1 pixels, so there will be some overlap between spectra. The peak throughput in IFU mode is predicted to be 13-15% Spectral resolutions from R=2000-13,000 will be available (R=lambda/delta(lambda)).

We are hoping to have spectropolarimetry capability with the AAOmega IFU, but this would be an upgrade and not part of the AAOmega project. In spectropolarimetry mode, the f.o.v. would be 11"x11", half the normal f.o.v.

Some concerns have been expressed as to whether an IFU can be competitive with a slit system: in particular, can an IFU provide accurate sky subtraction, spectrophotometry, spectropolarimetry, and UV coverage?   A summary of the performance of the current SPIRAL spectrograph can be found in Appendix A of the AAO Users Committee report (July 2000).  Analysis of SPIRAL data shows that sky subtraction can be performed to an accuracy of 2-4% without Nod&Shuffle, and to better than 1% with Nod&Shuffle (Poisson-limited). Results with the SPIRAl and PMAS IFUs show that spectrophotometry can be done to an accuracy of 1%. Spectropolarimetry with the SPIRAL IFU has not yet been tested. Regarding UV coverage, the AAOmega design wavelength coverage only extends to 3700 Ang in the blue, and the throughput is dropping rapidly below this point. Remember, though, that there are no slit losses associated with an IFU; with a long-slit spectrograph, slit losses are worse in the UV because of the dependence of seeing on wavelength.

A summary of the characteristics of the AAOmega IFU is given in the table below.
 

AAOmega - integral field mode
Field of view	22" x 11"
Number of fibres	512
Spatial sampling	0.7"
Wavelength range	0.37 - 0.95 microns
Dispersion	0.2 - 0.5 Å/pixel)
Spectral resolution (/)	2000 - 13000
Spectral resolution (FWHM)	2.1 Å
Peak system throughput1	13-15%
Limiting magnitude, 4 hours, S/N=10	B = 23.5

¹End-to-end throughput, from top of atmosphere to ccd.
 

Fig. 1: SPIRAL lenslet array, consisting of 32 x 16 elements.		Fig. 2: SPIRAL input fibre array, prior to attaching the microlens array.