An Adaptive-Optics Near-IR Integral field Spectrograph

for the Gran Telescopio de Canarias 

The main capability of this instrument that singles it out from the standard long-slit and Fabry-Perot spectrographs is the possibility of  simultaneous acquisition of medium resolution integral-field spectroscopy and high resolution imaging of a given field of view. 

Atlantis is foreseen to be a general-user instrument for Gran Telescopio de Canarias (GTC). 
 

Large collecting areas are essential for spectroscopic studies aimed at disentangling the emission components of a source at the sub-arcsec level. To cope efficiently with those studies requires instrumentation that benefit form two key observational parameters: low interstellar extinction and diffraction-limited resolution. For a ground-based observatory, both possibilities are currently only accessible in the near-IR domain. If furthermore a spatially detailed study is sought, the new instrumentation should combine two recent main developments: integral field spectroscopy and adaptive optics. 

The near-IR domain is spectrally very rich, including a large number of emission lines: molecular, nebular, coronal lines as absorption lines, altogether allowing us a fairly complete analysis of the physics of the emitting objects. In some cases, the IR may be the only way to achieve useful spectroscopic information as extinction may lead to a featureless optical spectrum if the object light is screened by large dust concentrations. In others, the near-IR is the only available door to the optical light of objects up to redshift of 3. 
 

Atlantis is currently designed for covering the 1-2.4 micron range 

The core of the instrument is the image-slicer. A monolithic bundle of fibers will  "slice" the field of view in the focal plane and then rearrange it into a long slit.   The instrument is designed to work with the new generation of IR arrays (2048 x  2048 pixels) and will utilize a bundle of 1024 fibers with the condition of one fiber per two pixels to minimize cross-talking. 

The fiber bundle design follows the novel solution of "flared fibers" introduced by the Max-Planck-Institute in the ESO VLT Sinfoni instrument. This technique has the advantage of making the instrument throughput almost virtually seeing independent. 

The instrument is designed to realise the promise of adaptive optics and so to provide near-IR spectroscopy at scales corresponding to the diffraction limit of a 10 m telescope. Pixels scales in the range 0.25 - 0.03 arcsec are considered together with the ability to rapidly change the scale to ensure optimal performance under different seeing conditions and/or scientific requirements. 

Initially, two set of gratings with resolving power, R, of 2000 and 5000 are considered.  R=2000 guarantees the coverage of two atmospheric bands in one single exposure while keeping still suitable velocities resolution (~ 150 km/s) for extragalactic studies. The higher resolution will be  about 5000  at which point it becomes feasible to suppress OH line emission from the atmosphere by software procedures.  R=5000 guarantees the coverage of only one atmospheric band  in one single exposure. 

The instrument is fully cryogenic; the working temperature will be down to about 77 K. This will guarantee maximum performance of the instrument at the K band. 

The instrument performance is mostly dictated by the dark-current levels of the available IR detectors, in particular with the adaptive optics (AO) pixel scales and at high spectral resolution where the sky background per pixel is very low. This is of particular relevance in the best parts of the J and H windows. 

The expected performance of the instrument is given in the table below. The estimated sensitivities are for the case of a point source observation, assuming a dark current of 1 e/s (e.g. Rockwell HAWAII array), minimum fibers emissivity, 0.1 arcsec scale, R=2000 and adaptive optics. 
 

Band	Mag	ExpT (min)	S/N	Limited by
H	19	30	40	Detector
K	~ 18	30	40	Background

Currently, the "basic" instrument concept is formed by the three following main subsystems: 
 
1. The focal extender: which projects the image given by the telescope onto the image slicer with the  appropriate  scaling. To obtain different scales, there would have to be a series of focal extenders, each one corresponding  to a specified image scale. These focal extenders shall be mounted on a rotating turret. 
2. The image slicer: a fiber bundle of 1024 fibers 
3. The spectrograph: that includes a collimator (3-mirrors de-centered system), exchangeable gratings and a camera (six lenses). 

In designing the camera  of the spectrograph it was found that the optical solution has excellent chromatic properties, thereby  permitting the design to serve for direct imaging as well.  Hence, an optional upgrade of Atlantis to an integrated field spectrograph and direct imaging is proposed. The possible field of view would be up to 3.4 x 3.4 arcmin, or about 1x1 arcmin if designed for adaptive mode. This upgrade  has evident impact on design volume and cryogenic, and inevitably, costs. 

These drawbacks should however be addressed in relation to the fact that this optional imaging capability may free a focus station of GTC. 
 

Atlantis - System diagram

Two "extreme" scientific cases are considered: the observation of the high redshift gravitational lens "the Einstein cross" at z=1.7 (Fig.1) and the observation of the nearby Andromeda galaxy (Fig. 2).

Figures 1 show the observation of the prototype gravitational lens "the Einstein cross" in the UV redshifted frame (Mediavilla et al. 1998, ApJL).  In Fig. 1a,  first panel shows the distribution of the fibers bundle projected on the sky; each fiber covers 0.45 arcsec in the sky. The medium panel shows the corresponding optical spectra of the field associated with each fiber, and so with each point of the sky. The last panel shows the reconstructed  image of the field after summing up all the spectra in fig. 1b. 
With Atlantis,  the same type of fibers configuration and data extraction as those presented in Figs 1 will apply. In  particular, the Atlantis  observation of the Einstein cross in the J/H bands  will provide us with information about the optical light of the system, and at  a much finer image scale  of 0.25 arcsec per fiber.
 

Fig. 1b shows a close-up view of the Einstein cross  from the inner 3x3 arcsec region as seen by INTEGRAL. The spectra associated with each point in the field are shown on top of the re-constructed image; the numbers identify the corresponding fiber position. The observed  emission line coinciding with the brighter emitting regions is CIII] 1909A, the remaining spectra being featureless as they correspond to the blank sky. At the redshift of the system (1.7), the expected most prominent line in the IR to be seen  with Atlantis is Halpha. 

Note that in the particular case of this observation,   sky information is simultaneously collected with object information. This is of particular relevance  for IR observations as  it can allow  an accurate background subtraction, possibly   minimizing  observing time as may not be need for chopping. 

The spectra in Fig1b have a resolution of about 90 km/s, which already allows for some of the complex structure of the line profile to be seen (e.g. spectrum 111). With  Atlantis , a higher resolution of about 60 km/s will be available, which will permit  any of the  atmospheric bands to be covered at once in one single observation. Because imaging is made after integrating the field spectra, it is up to the observer to decide whether to collapse the spectra around a given emission line or region of the continuum (narrow band imaging), or to collapse a broad region of the continuum spectrum (broad band imaging). 

Fig 1b.-  Close-up view of the Einstein cross as seen by INTEGRAL. The numbers on top of the spectra identify the corresponding fiber position in the field-of-view. 
Atlantis will deliver similar type of information for the near IR domain: with a resolution of about 60 km/s, a complete atmospheric band will be covered by each individual spectra in the field. This resolution will allow the  study of complex line profiles in  emission line systems with FWHM of few thousand km/s.

Closer in distance, the physics of sources in the local universe still remains poorly unexplored due to the non adequate matching of the spatial resolution to the physical scales of interest. Diagnosing the presence of central black holes by studying the stellar dynamics in the core of ellipticals, spectroscopic mapping of HII regions and starburst galaxies,  detailed kinematic mapping  of the interaction zone in merging systems, disentangling the nuclear light from the circumnuclear star-forming region in active galaxies, these are some of the challenging problems, essentially unknown on scales less than a few hundred parsecs, that can be tackled with high-resolution integral-spectroscopy in the IR.

Figure 2  shows the  observation of the Andromeda galaxy with INTEGRAL (del Burgo, 1998 PhD): the case study is the stellar kinematics in the inner parsecs region. As in the former case, the distribution of fibers projected on the sky is the same as in figure 1a. The left panel shows the reconstructed V image of the central 6x6 arcsec region of Andromeda. For sake of clarity, the individual spectra associated with each fiber (or point in the galaxy) are not shown but the image is the result of integrating all the spectra out of the fiber bundle. The right panel shows the associated stellar kinematics as derived from MgI 5175A. The spatial scale is  0.45 arcsec, the spectral resolution is about 90km/sec. 
With Atlantis,  stellar kinematic studies in nearby objects will be accessible via the measurement of the  relatively strong  CO  bandheads in the K spectrum. The finer spatial and spectral resolutions  provided by Atlantis, of 0.25 arcsec and 60 km/s respectively, will permit a rather detailed kinematic  analysis of the central region in nearby galaxies; definitively, an improved close view in the case of heavily obscured central regions.

.