LOW-TO-INTERMEDIATE RESOLUTION OPTICAL SPECTROSCOPY

I   SCIENCE DRIVERS

One kind of optical spectroscopy that will be done on SOAR unless we make it impossible is low-to-intermediate resolution observations of one object at a time. This remains the largest single use of the Blanco Telescope despite having had a multi-object fiber spectrograph available for years, because it encompasses a huge variety of projects. Multi-object programs on SOAR will be limited in FOV to a few arcmin which tend to exclude some interesting multi-target projects like kinematics of galaxy clusters and velocity dispersion in globular clusters. There are however multi-object projects that can be pursued with a small FOV.

I.1 Science drivers for single object spectroscopy

On SOAR, single object projects would include large doses of the following:

Source Identification.

 Many surveys (ISO, ROSAT, FIRST....) will provide samples of objects with space densities too low to follow up efficiently with multi-object spectroscopy. The next step in these programs is to obtain low-resolution spectra over a broad wavelength range, to try to figure out what the objects are. This has been a major science driver at CTIO over the past decade or more, including being a major use of the optical spectrograph on the 1.5m telescope (for the objects bright enough that it can reach them). Among this kind of programs are the optical identification of galactic black-holes, massive X-ray binaries and pulsars.

Monitoring Programs.

 Because the optical spectrograph on SOAR will be fixed and rapidly accessible, it will be well suited for programs featuring synoptic monitoring of specific objects. This is something which we cannot do well from CTIO at the present, but for which we get more and more demand. Types of objects would include stars, supernovae, novae, AGN, cataclysmic variables, X-ray binaries, etc. Sample programs would include determining H_0 and q_0 from SN, reverberation mapping of BLRs, and perhaps astroseismology from monitoring of Balmer equivalent widths. A range of resolutions would be of more interest here, but R~3000 or so would cover many programs.

 "Snapshot" studies of individual targets.

 These are projects that require detailed non-synoptic study of few objects. Some examples include the diagnostic of AGN, PN, Novae and galactic/LMC HII regions at low dispersion. There are also programs aiming to refine the luminosity function at low main sequence. Others like Stellar abundances determination and atmosphere modeling (including WDs, accretion disks in CVs and pre-main sequence stars) require low to intermediate dispersion. Radial velocity studies and mass functions in close binaries at intermediate dispersion (R = 5000 - 10000) and the study of spots in late type stars are being pursued at LNA/1.6m. Some of the projects above require flux calibration at medium resolution. Many of them require or are enhanced by two-dimensional spectral information. Additional extra-galactic topics include chemical abundances and population synthesis of galaxies, the study of the diffuse interstellar medium in spirals and the velocity dispersion in dwarf ellipticals.

 Surveys.

 Spectroscopic surveys presently being conducted with the Blanco telescope include abundances and kinematics of Milky Way bulge giants, and LMC field giants, kinematics of Carbon stars in the Magellanic Clouds, velocity dispersion of globular clusters and dwarf galaxies, kinematics of clusters and groups of galaxies, studies of cluster ellipticals and distance indicators and large scale motion mappings. Programs tantalizingly just beyond the edge of Blanco's capabilities would include velocities and metal abundances of giants, Horizontal Branch and main sequence turnoff stars in the MW halo, and a wide range of studies of stars in the LMC halo, including the kinematics and abundance distributions of RR Lyrae stars.

  I.2 Science drivers for Multi-object spectroscopy

 There are sample projects that require multi-object observations on a small field that would benefit from SOAR image quality. These are mostly related to photoionization structure of HII regions in the galaxy and Magellanic clouds. In addition there are studies on the dynamics and interactions in galaxy clusters that may be carried out over a small FOV with multi-object capabilities. Probing the central region of ellipticals with long-slit/multiple spectra is a topic of interest. Spatially resolved spectroscopy of bars can be used to investigate their connection with the AGN phenomenon. Star formation in triaxial bulges is another field that could highly benefits from surface spectroscopy. Even without a large FOV it is also possible to observe clusters of galaxies at z > 0.3.

  I.3 Science drivers for high spatial resolution spectroscopy

 Here we give some random examples of projects that make use of high-resolution 2D spectroscopy.

 Merging events in nearby Universe may trigger the birth of dwarf galaxies formed out of recycled material from the parent galaxies. 2D spectroscopy on these targets can be used to determine the fraction of dwarfs in groups, and binary systems, formed through this process. The physical properties and chemical abundances of nova shells can be studied with proper evaluation of geometry and and mass distribution structure. Observations of PN can be used to improve 3 dimensional photoionization codes that provides a detailed analysis of asymmetrical and/or inhomogeneous nebulae. In our Galaxy the study of bright condensations immersed in the ionized gas of HII regions can be used to clarify the star forming processes. Such 2D capability can also be used in the study of the star formation history of the circumnuclear rings of nearby spiral galaxies. In AGNs Spectroscopic imaging of nearby SY 2 would provideinformation with a spatial resolution of about 10 pc, allowing to test the shock+photoionization model.

   II  FAINT LIMITS

 With a 4m, these sorts of observations will mostly be background-limited. SOAR will inherently be 4x faster than Blanco because of the better image quality, but 4x slower than Gemini because of the smaller aperture. Presumably if we build a spectrograph like the one on Gemini it will have the same throughput that they get. If SOAR were provided with a spectrograph specialized for the highest-possible throughput, we might gain an extra factor of two as compared to the existing spectrograph on the Blanco and to GMOS on Gemini. Then SOAR would be 8x faster than the present Blanco capability, and only 2x slower than Gemini. This would make it a very nice alternative to using Gemini for all except the very faintest objects.

 Scaling from the Blanco's measured sensitivity, such an instrument would reach a signal:noise ratio of 10 per pixel at a per-pixel resolution of R=4200 (1.1A/pix) at 5000A, for a continuum source with monochromatic magnitude AB(5000A) = 22.5, in 1 hr in 0.25" FWHM seeing and 4 hr in 0.5" seeing. So with 4x binning to lower resolutions, SOAR could work down to about V=24 in good seeing.

 Stellar radial velocity studies using SOAR may reach mag 20 with internal errors smaller than 5 km/s. Compared to GMOS, a Nasmyth or bench spectrograph should be capable of better velocity measurements of bright objects at intermediate resolution (R~10000) due to superior mechanical stability when compared to Cassegrain mounted instruments. Flexure and tracking errors will possibly limit the velocity precision for stars brighter than I=20 in GMOS spectra.

  III SPECTROGRAPH REQUIREMENTS

 The above types of programs have the following general requirements:

 Efficiency:

 Rigorously maximized. Detect at least 30% of photons hitting primary mirror, over at least 50% of the observed passband. GMOS possibly achieve a value of 15% to 20%. Innovative optical designs should be pursued for maximum throughput.

 Field of View:

 1 arc-min minimum.
This will allow good sky measurements for observations of faint point sources, plus will cover the highest s:n parts of extended objects with peaked surface-brightness distributions. Sky subtraction in crowded fields may be problematic.

 Bigger is better.
A longer slit in a long-slit mode means a chance of getting simultaneous data over a larger galaxy, HII region, PN or whatever, and equally important, a better chance of getting outside the object where you can get a reasonably uncontaminated sky measure. The 5' slit length on the Blanco's RC spectrograph seems to be enough for the majority of cases. A field of several arcmin also invites multi-slit use (at the cost of additional complexity).

Wavelength Coverage:

Wide.
This is a circular argument, but the survey and identification projects sketched out above benefit greatly from covering the widest possible spectral range as efficiently as possible. Depending on cost, this could mean an octave in the spectrum (eg. 3500-7000A), the whole optical passband (0.31-1.1 microns), and also out into the near-IR for the brighter or redder objects. However, precisely simultaneous observations over these large wavelength ranges is rarely the goal. Rather, we just want to gather data over a wide wavelength range in a way that makes efficient use of the telescope. Steep sensitivity curves dropping just below 4000 A may affect several science projects - the impact of good UV response on the overall instrument capabilities is significant in the context of the proposed projects.

What is too wide?
The problem of achieving a good efficiency over a wide wavelength range involve many compromises in the optical design and detector choices. Optical dual-beam spectrographs often turn out to be less wonderful than they at first seemed because of the great difference in exposure times for the two channels; you buy lots of extra complexity for marginal gain in telescope efficiency. Trying to combine optical and near IR spectroscopy is even more problematical; there is a big difference in background levels between the optical and near-IR bands, so depending on their colors fainter objects will tend to be observable in one wavelength range but not in the other. These are the things to ask about when it comes time to design the spectrograph.

Resolving Power:
 There should be a mode that simply gives the highest R consistent with wide wavelength coverage as described above. I am guessing that this is R=3000-4000, but it will depend on the spectrograph design. For lower resolution observations of the faintest objects, it would be adequate to just retain the ability to bin pixels. There are programs that require intermediate resolutions (5000-10000), implying interchangeable gratings. These resolutions would be possibly available at GMOS. However, radial velocity measurements with rms < 8 km/s could be obtained with SOAR for 21 magnitude stars. In addition, R~10000 is often necessary for precision teluric correction from 6800 A to the IR limit. On the other hand, from the standpoint of the kind of science being pushed here the intermediate resolution capabilities should not be allowed to decrease the throughput of the spectrograph.

 2D - Spectroscopy:
 The proposed science require this capability. SOAR competitive image quality would be fully exploited by 2D spectroscopy projects of non-punctual objects. This statement suggests an integral field mode designed (or upgradable) to work with low order AO. The possibility of using IFU devices for this purpose has some advantages in the standard case of point source observations.