Annu. Rev. Astron. Astrophys. 1999. 37: 603-648 Copyright © 1999 by Annual Reviews. All rights reserved

2. INSTRUMENTAL CONSIDERATIONS

Progress in understanding the far-ultraviolet radiation from galaxies has been more circumscribed by instrumental limitations than was the case, for instance, in extragalactic X-ray astronomy. Fewer long-lived ultraviolet facilities have been available, and most of these have not been well suited for the study of galaxies. The problems are both intrinsic and technical. Intrinsically, galaxies are faint, extended sources. For typical elliptical galaxies, incident far-UV photon rates per unit solid angle per unit wavelength are typically over 50 times smaller than in the V-band. The centers of nearby bright ellipticals produce only a few × 10^-15 erg s^-1 cm^-2 Å^-1 arcsec^-2 at 1500 Å averaged over a 10" radius (Burstein et al 1988, Maoz et al 1996, Ohl et al 1998). The paucity of high contrast spectral features in UV hot star spectra at the spectral resolution and S/N possible for E galaxies has also hampered interpretation.

There has never been a large area UV sky survey sensitive enough to detect galaxies. The only all-sky survey yet made in the UV was by TD-1 in 1973 (Boksenberg et al 1973). This has a limit of about 9th magnitude and did not include a single galaxy or QSO. The GALEX mission (Martin et al 1997), now under development, will remedy this situation and produce a survey up to 10 magnitudes fainter. For now, however, the fact remains that the deepest survey of the UV sky is comparable to the Henry Draper catalog of stars, made around 1900. So UV astronomy, at least in this sense, is still 100 years behind optical astronomy.

The technical development of UV instrumentation has been reviewed by Boggess & Wilson (1987, spectroscopy), O'Connell (1991, imaging), Joseph (1995, detectors), and Brosch (1998, surveys). UV telescopes have been small, mostly less than 40 cm diameter. Other than the 2.4-m Hubble Space Telescope (HST), the largest UV instrument available has been the 1-m diameter Astro Hopkins Ultraviolet Telescope (HUT), which as a Shuttle-attached payload had an equivalent dedicated observing lifetime in 2 missions of only about 6 days (Kruk et al 1995). Observations of galaxies are difficult with the small entrance apertures available on most UV spectrometers, for example the International Ultraviolet Explorer (IUE) (10" × 20") or the HST/Faint Object Spectrograph ( 1"), which were designed for point sources. With IUE, long exposures of typically 4–8 hours were needed to register far-UV spectra of galaxies. Newer instruments are better matched to requirements for galaxy work. HUT was the first UV spectrometer designed specifically for galaxies, with apertures as large as 19" × 197" (providing, however, only one spatial resolution element). The Astro Ultraviolet Imaging Telescope (UIT) experiment, designed for filter imaging in the 1230–3200 Å region, had a field of view (40') and spatial resolution (3") well matched to ground-based studies of nearby galaxies. The new Space Telescope Imaging Spectrograph (STIS) offers UV apertures up to 2" × 52", encompassing many spatial resolution elements, and can image 25" × 25" fields with UV photon-counting detectors and 0.05", resolution. The HST Advanced Camera for Surveys, scheduled for installation in 2000, has high throughput UV cameras with fields up to 30" × 30".

The quantum efficiencies of UV detectors such as cesium iodide and cesium telluride photocathodes are only modest (10–30%), and net throughputs are further compromised by the lower reflectivities and transmissions of UV optical components. The most widely used mirror coating, magnesium fluoride, has a short-wavelength cutoff near 1150 Å. To obtain response to the Lyman discontinuity at 912 Å special coatings such as silicon carbide are now available (e.g. Kruk et al 1995), though these do not achieve reflectances typical of standard coatings at longer wavelengths.

Two special requirements for far-UV observations have serious practical consequences. First is the necessity to suppress the effects of the strong geocoronal Ly- emission line at 1216 Å. This is usually straightforward in spectrographs, but in photometers or imagers the only remedy is to use blocking filters that permit response only for 1250 Å. Second is the necessity to suppress residual filter and detector response to long-wave ( > 3000 Å) photons. Even though this may be only a tiny fraction of peak UV response, it covers a wide wavelength range. Because cool sources, such as stars with T_e < 7000 K, can have optical f thousands of times higher than their UV f, there can be serious "red leak" contamination of UV observations. Despite considerable effort (e.g. on Wood's filters), it has not been possible to develop fully satisfactory long-wave blocking devices with good peak UV response. Therefore, red leak suppression depends on the use of "solar-blind" detectors with large photoelectron work functions, such as cesium iodide, which has very small response for > 1800 Å. Such detectors have been used in most UV spectrometers but were not available in the HST Wide Field Camera (WFPC2) or HST Faint Object Camera (FOC), both of which consequently required careful red leak calibrations for use shortward of 2500 Å. The effects of red leaks on HST photometry of stars and galaxies can be dramatic and have been discussed by Yi et al (1995), Chiosi et al (1997). The requirements for simultaneous Ly- and red leak suppression imply smaller bandwidths and lower throughputs for far-UV imaging or photometry than is typical at longer wavelengths.

Because of these technical constraints, the working "far-ultraviolet" (FUV) band covers ~ 1250-2000 Å for imaging or photometry, extended to about 1150 Å for spectroscopy. The "mid-ultraviolet" (MUV) band covers ~ 2000-3200 Å (3200 Å being both the useful sensitivity limit of cesium telluride photocathodes and the short-wavelength cutoff of the Earth's atmosphere). We will call the 3200-4000 Å region accessible from the Earth's surface the "near-ultraviolet" (NUV). The 912-1150 Å region in galaxies has been explored to date only by HUT, though FUSE (launched in 1999) will also cover this range in brighter objects.

Unless noted, magnitudes quoted in this paper will be on the monochromatic system, where m = -2.5 log F - 21.1 and F is the mean incident flux in the relevant band in units of erg s^-1 cm^-2 Å^-1; the zero point is such that m(5500 Å) = V. Notation for colors will be, for instance, 1500-V m (1500 Å) - V.