3. UV MEASUREMENTS OF THE WHIM

3.1. Past and present UV instruments

The first and second generations of space based UV spectrographs such as Copernicus and the International Ultraviolet Explorer (IUE) did not have sufficient sensitivity to systematically study intervening absorption in the intergalactic medium along a large number of sightlines. The early low- and intermediate resolution spectrographs installed on the Hubble Space Telescope (HST), namely the Faint Object Spectrograph (FOS) and the Goddard High Resolution Spectrograph (GHRS), were used to study the properties of the local Ly forest and intervening metal-line systems (e.g., Stocke et al. 1995; Shull et al. 1998). While intervening O VI absorption has been detected with these instruments (e.g., Tripp et al. 1998), the concept of a warm-hot intergalactic gas phase was not really established at that time. With the implementation of the high-resolution capabilities of the Space Telescope Imaging Spectrograph (STIS) installed on HST the first systematic analyses of WHIM O VI absorbers as significant low-redshift baryon reservoirs came out in 2000 (see Tripp et al. 2000), thus relatively soon after the importance of a shock-heated intergalactic gas phase was realised in cosmological simulations for the first time (e.g., Cen & Ostriker 1999; Davé et al. 2001). The STIS echelle spectrograph together with the E140M grating provides a high spectral-resolution of R 45000, corresponding to a velocity resolution of ~ 7 km s^-1 in the STIS E140M wavelength band between 1150 and 1730 Å (e.g., Kimble et al. 1998; Woodgate 1998). An example for a STIS quasar spectrum with intervening hydrogen and metal-line absorption is shown in Fig. 4. Note that at the spectral resolution of the STIS E140M grating all intergalactic absorption lines (i.e., hydrogen and metal lines) are fully resolved. In 1999, the Far Ultraviolet Spectroscopic Explorer (FUSE) became available, covering the wavelength range between 912 and 1187 Å. Equipped with a Rowland-type spectrograph providing a medium spectral resolution of R 20000 (FWHM ~ 20 km s^-1) FUSE is able to observe extragalactic UV background sources brighter than V = 16.5 mag with acceptable integration time and signal-to-noise (S/N) ratios (for a description of FUSE see Moos et al. 2000; Sahnow et al. 2000). With this resolution, FUSE is able to resolve the broader intergalactic absorption from the H I Lyman series, while most of the narrow metal-line absorbers remain just unresolved. This is not a problem for O VI WHIM studies with FUSE, however, since the spectral resolution is very close to the actual line widths and the O VI absorption usually is not saturated. FUSE complements the STIS instruments at lower wavelengths down to the Lyman limit and consequently combined FUSE and STIS spectra of ~ 15 low redshift QSOs and AGN have been used to study the low-redshift WHIM via intervening O VI and BLA absorption (see Tripp et al. 2007 and references therein). Unfortunately, since 2006/2007 both STIS and FUSE are out of commission due to technical problems.

Figure 4. STIS spectrum of the quasar PG 1259+593 in the wavelength range between 1300 and 1400 Å. Next to absorption from the local Ly forest and gas in the Milky Way there are several absorption features that most likely are related to highly ionised gas in the WHIM. Absorption from five-times ionised oxygen (O VI) is observed at z = 0.25971 and z = 0.31978. Broad H I Ly and Ly absorption is detected at z = 0.08041, 0.09196, 0.10281, 0.13351, 0.14034, 0.14381, 0.14852, 0.15136, and z = 0.31978. From richter2004.

Fresh spectroscopic UV data from WHIM absorption line studies will become available once the Cosmic Origins Spectrograph (COS) will be installed on HST during the next HST service mission (SM-4), which currently is scheduled for late 2008. COS will observe in the UV wavelength band between 1150 and 3000 Å at medium resolution (R 20000). COS has been designed with maximum effective area as the primary constraint: it provides more than an order of magnitude gain in sensitivity over previous HST instruments. Due to its very high sensitivity, COS thus will be able to observe hundreds of low- and intermediate redshift QSOs and AGN and thus will deliver an enormous data archive to study the properties of WHIM UV absorption lines systems in great detail (see also Paerels et al. 2008 - Chapter 19, this volume).

3.2. Intervening WHIM absorbers at low redshift

We start with the O VI absorbers which are believed to trace the low-temperature tail of the WHIM at T < 5 × 10⁵ K. Up to now, more than 50 detections of intervening O VI absorbers at z < 0.5 have been reported in the literature (e.g., Tripp et al. 2000; Oegerle et al. 2000; Chen & Prochaska 2000; Savage et al. 2002; Richter et al. 2004; Sembach et al. 2004; Savage et al. 2005; Danforth & Shull 2005; Tripp et al. 2007). All of these detections are based on FUSE and STIS data. Fig. 5 shows two examples for intervening O VI absorption at z = 0.23351 and z = 0.26656 in the direction of PG 0953+415 and H 1821+643, as observed with STIS. The most recent compilation of low-redshift intervening O VI absorbers is that of Tripp et al. (2007), who have analysed 16 sightlines toward low-redshift QSOs observed with STIS and FUSE along a total redshift path of z 3. They find a total of 53 intervening O VI absorbers (i.e., they are not within 5000 km s^-1 of z_QSO) comprised of 78 individual absorption components ¹. The measurements imply a number density of O VI absorbing systems per unit redshift of dN_OVI / dz 18 ± 3 for equivalent widths W 30 mÅ. The corresponding number density of O VI absorption components is dN_{O VI} / dz 25 ± 3. These values are slightly higher than what is found by earlier analyses of smaller O VI samples (Danforth & Shull 2005), but lie within the cited 2 error ranges. The discrepancy between the measured O VI number densities probably is due to the different approaches of estimating the redshift path z along which the O VI absorption takes place. If one assumes that the gas is in a collisional ionisation equilibrium, i.e., that ~ 20 percent of the oxygen is present in the form of O VI (f_{O VI} 0.2), and further assumes that the mean oxygen abundance is 0.1 Solar, the measured number density of O VI absorbers corresponds to a cosmological mass density of _b(O VI) 0.0020-0.0030 h₇₀^-1. These values imply that intervening O VI absorbers trace ~ 5-7 percent of the total baryon mass in the local Universe. For the interpretation of _b(O VI) it has to be noted that O VI absorption traces collisionally ionised gas at temperatures around 3 × 10⁵ K (and also low-density, photoionised gas at lower temperatures), but not the million-degree gas phase which probably contains the majority of the baryons in the WHIM.

Figure 5. Examples for H I and O VI absorption in two absorption systems at z = 0.23351 and z = 0.6656 towards PG 0953+415 and H 1821+643, respectively, plotted on a rest frame velocity scale (observed with STIS). Adapted from Tripp et al. (2007).

The recent analysis of Tripp et al. (2007) indicates, however, that this rather simple conversion from measured O VI column densities to _b(O VI) may not be justified in general, as the CIE assumption possibly breaks down for a considerable fraction of the O VI systems. From the measured line widths of the H I Ly absorption that is associated with the O VI Tripp et al. conclude that ~ 40 percent of their O VI systems belong to cooler, photoionised gas with T < 10⁵ K, possibly not at all associated with shock-heated warm-hot gas. In addition, about half of the intervening O VI absorbers arise in rather complex, multi-phase systems that can accommodate hot gas at relatively low metallicity. It thus appears that - without having additional information about the physical conditions in each O VI absorber - the estimate of the baryon budget in intervening O VI systems is afflicted with rather large systematic uncertainties.

In high-column density O VI systems at redshifts z > 0.18, such desired additional information may be provided by the presence or absence of Ne VIII (see Sect. 2.2.1), which in CIE traces gas at T ~ 7 × 10⁵ K. Toward the quasar PG 1259+593 Richter et al. (2004) have reported a tentative detection of Ne VIII absorption at ~ 2 significance in an O VI absorber at z 0.25. The first secure detection of intervening Ne VIII absorption (at ~ 4 significance) was presented by Savage et al. (2005) in a multi-phase O VI absorption system at z 0.21 in the direction of the quasar HE 0226-4110. The latter authors show that in this particular absorber the high-ion ratio Ne VIII / O VI = 0.33 is in agreement with gas in CIE at temperature of T ~ 5 × 10⁵ K. With future high S/N absorption line data of low-redshift QSOs (as will be provided by COS) it is expected that the number of detections of WHIM Ne VIII absorbers will increase substantially, so that an important new diagnostic will become available for the analysis of high-ion absorbers.

One other key aspect in understanding the distribution and nature of intervening O VI systems concerns their relation to the large-scale distribution of galaxies. Combining FUSE data of 37 O VI absorbers with a database of more than a million galaxy positions and redshifts, Stocke et al. (2006) find that all of these O VI systems lie within 800 h₇₀^-1 kpc of the nearest galaxy. These results suggest that O VI systems preferentially arise in the immediate circumgalactic environment and extended halos of galaxies, where the metallicity of the gas is expected to be relatively high compared to regions far away from galactic structures. Some very local analogs of intervening O VI systems thus may be the O VI high-velocity clouds in the Local Group that are discussed in the next subsection. Due to apparent strong connection between intervening O VI systems and galactic structures and a resulting galaxy/metallicity bias problem it is of great interest to consider other tracers of warm-hot gas, which are independent of the metallicity of the gas. The broad hydrogen Ly absorbers - as will be discussed in the following - therefore represent an important alternative for studying the WHIM at low redshift.

As described in Sect. 2.2.1, BLAs represent H I Ly absorbers with large Doppler parameters b > 40 km s^-1. If thermal line broadening dominates the width of the absorption, these systems trace the WHIM at temperatures between 10⁵ and 10⁶ K, typically (note that for most systems with T > 10⁶ K BLAs are both too broad and too shallow to be unambiguously identified with the limitations of current UV spectrographs). The existence of H I Ly absorbers with relatively large line widths has been occasionally reported in earlier absorption-line studies of the local intergalactic medium (e.g., Tripp et al. 2001; Bowen et al. 2002). Motivated by the rather frequent occurrence of broad absorbers along QSO sightlines with relatively large redshift paths, the first systematic analyses of BLAs in STIS low-z data were carried out by Richter et al. (2004) and Sembach et al. (2004). Richter et al. (2006a) have inspected four sightlines observed with STIS towards the quasars PG 1259+593 (z_em = 0.478), PG 1116+215 (z_em = 0.176), H 1821+643 (z_em = 0.297), and PG 0953+415 (z_em = 0.239) for the presence of BLAs and they identified a number of good candidates. Their study implies a BLA number density per unit redshift of dN_BLA / dz 22-53 for Doppler parameters b 40 km s^-1 and above a sensitivity limit of log (N(cm^-2) / b(km s^-1)) 11.3. The large range for dN_BLA / dz partly is due to the uncertainty about defining reliable selection criteria for separating spurious cases from good broad Ly candidates (see discussions in Richter et al. 2004, 2006a) and Sembach et al. 2004). Transforming the number density dN_BLA / dz into a cosmological baryonic mass density, Richter et al. (2006a) obtains _b(BLA) 0.0027 h₇₀^-1. This limit is about 6 percent of the total baryonic mass density in the Universe expected from the current cosmological models (see above), and is comparable with the value derived for the intervening O VI absorbers (see above). Examples for several BLAs in the STIS spectrum of the quasar H 1821+643 are shown in Fig. 6.

Figure 6. Broad Lyman absorbers towards the quasars H 1821+643 and PG 0953+415 (STIS observations), plotted on a rest frame velocity scale. If thermal line broadening dominates the width of the absorption, these systems trace the WHIM at temperatures between 105 and 106 K. From Richter et al. (2006a).

More recently, Lehner et al. (2007) have analysed BLAs in low-redshift STIS spectra along seven sightlines. They find a BLA number density of dN_BLA / dz = 30 ± 4 for b = 40-150 km s^-1 and log N(H I) > 13.2 for the redshift range z = 0-0.4. They conclude that BLAs host at least 20 percent of the baryons in the local Universe, while the photoionised Ly forest, which produces a large number of narrow Ly absorbers (NLAs), contributes with ~ 30 percent to the total baryon budget. In addition, Prause et al. (2007) have investigated the properties of BLAs at intermediate redshifts (z = 0.9 - 1.9) along five other quasars using STIS high- and intermediate-resolution data. They find a number density of reliably detected BLA candidates of dN_BLA / dz 14 and obtain a lower limit of the contribution of BLAs to the total baryon budget of ~ 2 percent in this redshift range. The frequency and baryon content of BLAs at intermediate redshifts obviously is lower than at z = 0, indicating that at intermediate redshifts shock-heating of the intergalactic gas from the infall in large-scale filaments is not yet very efficient. This is in line with the predictions from cosmological simulations.

3.3. The Milky Way halo and Local Group gas

One primary goal of the FUSE mission was to constrain the distribution and kinematics of hot gas in the thick disk and lower halo of the Milky Way by studying the properties of Galactic O VI absorption systems at radial velocities |v_LSR| 100 km s^-1 (Savage et al. 2000; Savage et al. 2003; Wakker et al. 2003). However, as the FUSE data unveil, O VI absorption associated with Milky Way gas is observed not only at low velocities but also at |v_LSR| > 100 km s^-1 (Sembach et al. 2003). The topic of cool and hot gas in the halo of the Milky Way recently has been reviewed by Richter et al. (2006c). These detections imply that next to the Milky Way's hot "atmosphere" (i.e., the Galactic Corona) individual pockets of hot gas exist that move with high velocities through the circumgalactic environment of the Milky Way. Such high-velocity O VI absorbers may contain a substantial fraction of the baryonic matter in the Local Group in the form of warm-hot gas and thus - as discussed in the previous subsection - possibly represent the local counterparts of some of the intervening O VI absorbers observed towards low-redshift QSOs.

From their FUSE survey of high-velocity O VI absorption Sembach et al. (2003) find that probably more than 60 percent of the sky at high velocities is covered by ionised hydrogen (associated with the O VI absorbing gas) above a column density level of log N(H II) = 18, assuming a metallicity of the gas of 0.2 Solar. Some of the high-velocity O VI detected with FUSE appears to be associated with known high-velocity H I 21 cm structures (e.g., the high-velocity clouds complex A, complex C, the Magellanic Stream, and the Outer Arm). Other high-velocity O VI features, however, have no counterparts in H I 21 cm emission. The high radial velocities for most of these O VI absorbers are incompatible with those expected for the hot coronal gas (even if the coronal gas motion is decoupled from the underlying rotating disk). A transformation from the Local Standard of Rest to the Galactic Standard of Rest and the Local Group Standard of Rest velocity reference frames reduces the dispersion around the mean of the high-velocity O VI centroids (Sembach et al. 2003; Nicastro et al. 2003). This can be interpreted as evidence that some of the O VI high-velocity absorbers are intergalactic clouds in the Local Group rather than clouds directly associated with the Milky Way. However, it is extremely difficult to discriminate between a Local Group explanation and a distant Galactic explanation for these absorbers. The presence of intergalactic O VI absorbing gas in the Local Group is in line with theoretical models that predict that there should be a large reservoir of hot gas left over from the formation of the Local Group (see, e.g., Cen & Ostriker 1999).

It is unlikely that the high-velocity O VI is produced by photoionisation. Probably, the gas is collisionally ionised at temperatures of several 10⁵ K. The O VI then may be produced in the turbulent interface regions between very hot (T > 10⁶ K) gas in an extended Galactic Corona and the cooler gas clouds that are moving through this hot medium (see Sembach et al. 2003). Evidence for the existence of such interfaces also comes from the comparison of absorption lines from neutral and weakly ionised species with absorption from high ions like O VI (Fox et al. 2004).