Annu. Rev. Astron. Astrophys. 2012. 50: 491-529 Copyright © 2012 by Annual Reviews. All rights reserved

2. THE GALACTIC HALO

This section focuses on the significant observational progress that has been made in studies of Milky Way halo gas in the past decade (see Wakker & van Woerden 1997 for earlier work) and is divided into sub-sections based primarily on the temperature of the gas. The neutral hydrogen halo gas discussed in Section 2.1 is at temperatures < 10,000 K and detected through the 21-cm line in emission. This is traditionally the primary tracer of Galactic halo gas, and we therefore separately discuss its physical properties (Section 2.1.1). The warm and warm-hot gas is defined here as largely ionized with temperatures > 10⁴ K, but < 10⁶ K (Section 2.2). This gas is detected in optical line emission (primarily H) and absorption lines in the optical and ultraviolet . Finally, hot gas is at temperatures greater than 10⁶ K and detectable through x-ray observations close to the disk and indirect methods further out (Section 2.3). The dust, molecules and metals in halo gas are discussed in Section 2.4.

Distances are often an issue when defining gas as a halo medium. There has been significant progress in this area (see Section 2.1.1), but in any case, it should be noted that halo gas has largely been defined based on its velocity not falling within the expected Galactic disk velocity in a given direction. This definition may in some (relatively rare) cases pick up Galactic gas that has been recently imparted with a large amount of kinetic energy, and it definitely misses halo gas that overlaps with the velocity of the disk. In general, the transition between halo and disk gas is blurry, and a continuum of gas between the disk and halo seems likely. Halo gas is defined here to be at heights above the Galactic Plane (z-heights) of a few kpc and not part of the Galactic warp, and gas below this that shows some connection to the disk is discussed in the disk-halo interface section (Section 2.5).

2.1. Neutral Hydrogen Halo Gas

The first halo gas detected was in neutral hydrogen emission at high velocity relative to Galactic disk velocities (Muller et al. 1963). These discrete gas clouds were subsequently referred to as high-velocity clouds (HVCs). HVCs are the coldest and densest component of halo gas and are found throughout the sky (Figures 1 & 2). They are material stripped from satellites and/or cooling components of the accreting IGM and feedback material (see Section 4). Significant progress in mapping the distribution of HVCs has come through multibeam instruments that have been used to observe large areas of sky at relatively high resolution (4' to 16'; Putman et al. 2002, Saul et al. 2012, McClure-Griffiths et al. 2009). In addition, the entire sky has now been mapped at a uniform resolution of 35' and 1 km s^-1 with the Leiden/Argentine/Bonn (LAB) HI survey (Kalberla et al. 2005; see Figures 1 & 2). The full-sky successor to the LAB survey will be a combination of multibeam surveys with a resolution of 9-16' and 1 km s^-1 (McClure-Griffiths et al. 2009, Winkel et al. 2010).

Figure 1. The distribution of HI (shaded clouds and plus symbols) and ionized high velocity gas (circles and diamonds) on the sky with color denoting the LSR velocity of the detection. The HI map was created with LAB data by Westmeier (2007) by removing the HI model of the Milky Way from Kalberla & Kerp (2009). The plus symbols represent the the small HI compact HVCs detected with other datasets (Putman et al. 2002, de Heij et al. 2002), open circles are the O VI absorption line detections (Sembach et al. 2003), solid circles are the Si absorption line detections (Shull et al. 2009), and diamonds are the ionized HVCs (IHVCs) at 15 kpc (Lehner & Howk 2011). Multiple colors for a symbol indicate multiple absorbers along the line of sight. The positions of the major HI HVC complexes (with S to indicate the Smith cloud) and the LMC, SMC, M31 and M33 galaxies are noted. The AC Shell appears in this map, but is considered an intermediate velocity cloud.

Figure 2. The distribution of HI (shaded clouds and plus symbols) and ionized high velocity gas (circles and diamonds) on the sky with color denoting the GSR velocity of the detection. Otherwise this figure is the same as Figure 1, with the exception of showing the view from the North (left) and South (right) Galactic Poles at the bottom.

HVCs range in size from 1000 deg² to the resolution limit of the observations, and large clouds are often found to be composed of a number of smaller clouds when observed with better resolution and equivalent sensitivity (Putman et al. 2002, Stanimirovic et al. 2008, Hsu et al. 2011). There are many catalogs of clouds that have been created with the various HI survey data that use different catalog methods. The number of individual clouds in these catalogs ranges from ~ 600 over the entire sky in the lowest resolution surveys (35' beam; Wakker & van Woerden 1991), to 2000 in just the southern sky (15.5' beam; Putman et al. 2002), to ~ 700 clouds smaller than 20' in 7600 deg² of the sky (4' beam; Saul et al. 2012). It remains difficult to separate the Galactic disk gas from halo gas. One method of cataloguing distinct halo features is to run algorithms on the entire HI data cube to determine structures that are distinct from each other in position-velocity space (Maller et al. in prep.; Sharma & Johnston 2009).

Clouds of all sizes can be grouped into complexes of most likely related clouds based on their spatial and kinematic proximity as labeled in Figures 1 and 2. In the Southern hemisphere the HVCs that dominate in sky coverage and mass are associated with the Magellanic System. In particular, the complexes labeled MS and LA are the trailing Magellanic Stream and Leading Arm (recent references include: (Putman 2004, Putman et al. 2003b, Putman et al. 1998, Brüns et al. 2005, Westmeier & Koribalski 2008, Stanimirovic et al. 2002, Stanimirovic et al. 2008, Nidever et al. 2010). These complexes were created from the interaction of the Large and Small Magellanic Clouds (LMC and SMC) with each other and with the Milky Way. The Magellanic Stream is a long, continuous structure with a well-defined velocity and column density gradient, while the Leading Arm is a collection of clouds throughout the region leading the Magellanic Clouds. In the Northern hemisphere there are a number of complexes of similar spatial size, but Complex C is the largest. The best way to view the HI component of the complexes at their approximate distances (see 2.1.1) is through the three-dimensional Figure 3. In both hemispheres there are a number of small clouds called compact HVCs (CHVCs) that have been claimed to be at large distance due to their size (< 2° in diameter) and relative isolation (Braun & Burton 1999, de Heij et al. 2002). CHVCs can largely be associated with known HVC complexes in position-velocity space as shown by the plus symbols in Figures 1 and 2, and are therefore unlikely to be at greater distance. Even the ultra-compact HVCs ( 20') have been shown to be related to HVC complexes (Saul et al. 2012). None of the CHVCs or HVC complexes have a clear association with stellar features in the halo (Hopp et al. 2007, Siegel et al. 2005), excluding the association of MS and LA to the Magellanic Clouds.

Figure 3. 3D, interactive model of the Galactic HVC complexes. The Galactic disk is shown with a 25 kpc radius and a 1 kpc thickness. The central regions of our Galaxy are indicated with a gray 3 kpc radius sphere, and the Solar position is shown with a yellow sphere. HVC complexes are placed at the average of their most stringent distance constraints (Section 2.1.1). The Magellanic Stream is shown starting at a distance of 50 kpc, and increasing linearly to 150 kpc at a Magellanic longitude of 150 degrees, consistent with the models of Besla et al. (2010). The positions of the LMC and SMC are noted with gray spheres. Cloud thickness is an arbitrary function of column density along the line of sight. If the Figure does not appear interactive, please view with a current version of Adobe Reader and click and drag to rotate.

The majority of the kinematic range of the HVCs can be seen in Figures 1 and 2. The range has been extended further with new surveys and is currently approximately -500 to +450 km s^-1 in the Local Standard of Rest velocity frame (V_LSR) or between -300 and +300 km s^-1 in the Galactic Standard of Rest frame (V_GSR). HVCs can be defined by the cloud's deviation velocity (V_dev), or the amount the velocity of the cloud deviates from a simple model of Galactic rotation. Figures 1 and 2 were created with |V_dev| > 75 km s^-1 and thus there are some small differences from maps created with the classical |V_LSR| > 70-90 km s^-1 definition (e.g. Westmeier 2007, Wakker & van Woerden 1991). The clouds at both the positive and negative velocity extremes are largely associated with the Leading Arm and the tip of the Magellanic Stream, respectively. Since HVCs have not been catalogued in the velocity range obscured by Galactic emission, it is difficult to assess the dispersion of the population. Using the GSR frame, an estimate for the velocity dispersion of the HVCs is ~ 100-120 km s^-1.

The HVCs have typical linewidths of 20-30 km s^-1, corresponding to velocity dispersions of = 8.5-13 km s^-1 (Putman et al. 2002, de Heij et al. 2002, Kalberla & Haud 2006). This linewidth is indicative of a warm neutral medium for HVCs at ~ 9000 K (Hsu et al. 2011). Many HVCs also show a narrow component in their velocity profiles corresponding to cold gas at _~ ^< 500 K (Kalberla & Haud 2006). The two-phase structure of HVCs is expected for clouds in pressure equilibrium with a diffuse hot Milky Way halo medium (Section 2.3; Wolfire et al. 1995). It is possible that HVCs that do not show this two-phase structure are currently in a more turbulent environment, or are embedded in a lower pressure part of the halo.

HVCs have typical peak column densities of ~ 10¹⁹ cm^-2, and a small number of clouds have peak column densities above 10²⁰ cm^-2 with the 4-15.5' resolution HI surveys (Putman et al. 2002, Hsu et al. 2011). Additional dense cores with HI column densities > 10²⁰ cm^-2 have been detected within some HVCs with synthesis observations (e.g., Wakker et al. 2002, Ben Bekhti et al. 2006). The clouds are found to extend down in column density to the limits of the HI observations (> 10¹⁷ cm^-2; (Lockman et al. 2002, Braun & Thilker 2004). This suggests HI surveys may be detecting only a fraction of the area of many complexes, although they are likely to be detecting the majority of the HI mass (e.g., Peek 2009). There is also evidence that the clouds detected in HI emission are linked to lower column density absorption line systems (Ben Bekhti et al. 2012, Murphy et al. 1995, Nidever et al. 2010). At ~ 7 × 10¹⁷ cm^-2 in HI emission the covering fraction is ~ 37% of the sky (Murphy et al. 1995, Lockman et al. 2002).

When the detailed HI structure of HVCs is examined, the clouds show evidence for being ionized by an external radiation field (Maloney & Putman 2003, Sternberg et al. 2002). The HI structure also often indicates the clouds are moving through a diffuse halo medium and being disrupted (Peek et al. 2007, Brüns et al. 2000, Stanimirovic et al. 2006, see Figure 4 and Section 5). Head-tail clouds, that form as the leading side of the HVC is compressed and gas is stripped behind it, are particularly abundant in the vicinity of the Leading Arm (Putman et al. 2011b).

Figure 4. Left: Head-tail HI halo clouds with the outer contour at NHI = 2 × 1018 cm-2 (15.5' resolution; Putman et al. 2011b). Right: A HI HVC that shows signatures of interaction with the halo medium from the edge features (4' resolution; Peek et al. 2007).

2.1.1    DISTANCES AND PHYSICAL PROPERTIES    HVC distance measurements allow for estimates of their physical properties such as mass, size, volume density, and pressure (e.g., Wakker et al. 2001, Hsu et al. 2011). The best method to measure the distance to a HVC is to use a halo star of known distance toward the cloud and search for absorption lines at the HVC's velocity in the star's spectrum. A detection of absorption from the HVC shows that the cloud is in front of the star, while not detecting absorption may indicate the HVC is behind the star. Usually the optical Ca II H and K and/or Na II doublet absorption lines from the HVC are used for this distance measurement. Concluding the HVC is behind the star with a non-detection requires very deep observations. This is because the line of sight to the star may have a lower HI column density than measured by the radio observations and there is uncertainty in the gas-phase abundance of the metals (Wakker et al. 2001). The identification of halo stars with the Sloan Digital Sky Survey (SDSS) and other surveys has enabled distances to be obtained for almost all of the large, labeled HVC complexes in Figure 1. Most of the complexes are at distances between 2-15 kpc and z-heights 10 kpc (Wakker et al. 2008, Thom et al. 2006, Thom et al. 2008, Wakker et al. 2007), Wakker et al., in prep.), and, as viewed from the Galactic Center, they are within 30° of the disk plane. The clear exceptions to these distances are the Magellanic Stream and Leading Arm, which are at ~ 55 kpc near the Magellanic Clouds and may extend to 100-150 kpc at the tip of the Stream (e.g., Besla et al. 2010). The complexes can be viewed at their distances in the three-dimensional map of Figure 3.

Distance measurements for HVCs obtained with halo stars are referred to as direct distance constraints. Indirect distance constraints are model dependent and available for almost all of the HVCs. H observations of HVCs can be used to estimate their distance if one assumes that the H emission is the result of the ionizing radiation from the Galaxy reaching the surface of the cloud (the fraction escaping normal to the disk is f_esc ~ 6% in this model; (Bland-Hawthorn & Maloney 1999, Bland-Hawthorn & Maloney 2002, Putman et al. 2003a, Bland-Hawthorn & Putman 2001, Bland-Hawthorn et al. 1998). Most of the distances derived from H observations of HVCs have been consistent with their direct distance constraints using halo stars. Clouds without direct distance constraints (in particular the CHVCs) are also usually detected in H emission, which is consistent with them being within 40 kpc of the Galactic disk (Tufte et al. 2002, Putman et al. 2003a). The striking exception to the escaping radiation model is the Magellanic Stream, which is too bright in H emission given its distance (see Section 2.2).

The most important additional indirect constraint on HVC distances is deep HI observations of systems similar to the Milky Way and/or Local Group. If one assumes the distribution of HVCs is similar, searches of Local Group analogs for HI clouds place even the compact HVCs within 80 kpc of the Galactic disk (Pisano et al. 2007) and deep observations of Andromeda suggest the clouds are within 50 kpc (Westmeier et al. 2007). The distance constraints have ruled out the possibility that the majority of the small HVCs are dark matter halos found throughout the Local Group (Blitz et al. 1999, Braun & Burton 1999). Recent Arecibo surveys are finding smaller clouds that do not yet have distance constraints (Giovanelli et al. 2010, Begum et al. 2010), but most of these clouds are associated with HVC complexes in position-velocity space (Saul et al. 2012).

The HI masses and sizes of HVC complexes can be determined once distances are known. Calculating the HI mass requires a measurement of the total flux of the cloud and is proportional to the square of the distance. Determining the physical size of a HVC requires scaling the observed size directly with the distance, and this measurement may vary with the beam size and sensitivity of the observations. The HI masses of the HVC complexes (each consisting of many individual clouds) are in the range of 10⁵ - 5 × 10⁶ M, and their sizes are a few to 15 kpc across (excluding MS and LA). There are now enough distance constraints that we can determine the total mass in halo clouds. For the 13 complexes with direct distance constraints, we calculate the mass by placing each complex at the average of its upper and lower distance limits. We place the remaining clouds ( ~ 23% of the total HVC HI flux) at a distance of d=10 kpc, since the majority of the complexes are at approximately this distance and indirect distance constraints indicate there are not clouds at significantly larger distances. This gives a total HI mass in non-Magellanic System HVCs of 2.6 × 10⁷ M. This mass should be roughly doubled and multiplied by 1.4 to account for the ionized component (Section 2.2) and Helium respectively, thus giving a total mass of 7.4 × 10⁷ M. The expected ongoing accretion rate from the HVC complexes is addressed in Section 6.1. The Magellanic System alone contributes a minimum of 3 × 10⁸ M (d / 55 kpc)² just in HI to the halo from the Magellanic Stream, Leading Arm and Magellanic Bridge which joins the two Clouds (Putman et al. 2003b, Brüns et al. 2005). Including the LMC and SMC galaxies would add another 7 × 10⁸ M in HI to the Milky Way halo. All other satellites with HI gas are beyond the virial radius of our Galaxy (Grcevich & Putman 2009).

The derivations of volume densities and pressures require additional assumptions about the HVCs and therefore should be adopted with some caution. For volume densities, it is often assumed the cloud's path length is similar to the size of the dimensions observed. For spherical clouds this may be a valid assumption if the linewidth does not seem unusually broadened. Volume density estimates for clouds with direct distance constraints are ~ 0.05-0.15 cm^-3, and when simulated halo clouds are 'observed' the measured volume densities are within a factor of 3-4 of their actual volume densities (Wakker et al. 2001, Hsu et al. 2011). An estimate of a HVC's pressure is made with the ideal gas law using the derived volume density and the kinetic temperature calculated from the linewidth (see Section 2.1). Using the above range of densities and a temperature of 9000 K typical for the HVCs' warm neutral component, the pressures for the clouds with distance constraints are log(P/k) = 2.7 - 3.1 K cm^-3. The HVCs are at z-heights from 3-9 kpc and the pressures derived are consistent with the calculations by Wolfire et al. (1995) for clouds in pressure equilibrium with a hot halo medium. The estimates for the pressures associated with the Magellanic Stream clouds are substantially lower, and this would be expected for clouds in a more distant, lower density halo medium (Stanimirovic et al. 2008, Hsu et al. 2011).

2.2. Warm Ionized and Warm-Hot Halo Gas

The warm gas (~ 10^4-5 K) in the Galactic halo is detected with deep H emission line observations and low and intermediate ion absorption lines (e.g., Si II, Si III, C II, C III, O I) using background QSOs or stars. The H observations are generally targeted toward known HI HVCs, while the location of the absorption line observations are dictated by the positions of background probes. Both methods have led to numerous detections of warm Galactic halo gas. This gas probes the interaction of HI clouds with photons from the disk and with other halo gas (Section 4.2), and also comes directly from the IGM and/or satellites (Section 4.1 & 4.3).

There are numerous detections of HI halo clouds in H emission with Fabry-Perot instruments and deep long-slit spectroscopy (Putman et al. 2003a, Tufte et al. 1998, Tufte et al. 2002, Bland-Hawthorn et al. 1998, Weiner & Williams 1996, Weiner et al. 2002). All of the high-velocity H detections are located on or near HI HVCs. Extensions of the H emission from the HI have been noted for one complex (Hill et al. 2009), but this does not appear to be a generic condition (Haffner 2005, Haffner et al. 2001, Putman et al. 2003a). The detections of H emission are typically < 0.5 Rayleigh (R), where 1 R is 10⁶ / 4 photons cm^-2 s^-1 sr^-1. There are some non-detections of H emission for HVCs, but it is possible they will be detected in the future if the cloud is fully mapped to deep levels. One distinct non-detection may be the Leading Arm of the Magellanic System, which is not detected in H for numerous pointings (< 30-50 mR). This is in contrast to the Magellanic Stream, which is particularly bright in many locations and requires models that invoke collisional ionization mechanisms beyond photoionization by radiation from the disk (Bland-Hawthorn et al. 2007, Weiner & Williams 1996, Putman et al. 2003a); Section 2.1.1 and Section 4.2). The velocities of the HI and H lines are approximately the same for all HVC complexes, and there is no evidence for a correlation between HI column density and H brightness. Other species such as [NII] and [SII] have also been detected in emission for some HVCs (Bland-Hawthorn et al. 1998, Tufte et al. 1998, Putman et al. 2003a).

The mass of the warm ionized component of halo clouds should be added to the neutral hydrogen mass to obtain the total mass in halo clouds. The H detections have been used to calculate that the ionized component of the HVC complexes is on the order of the mass of the neutral component in many cases (Hill et al. 2009, Wakker et al. 2008, Shull et al. 2011). This suggests the warm ionized component of the HVCs contributes 3-4 × 10⁷ M to the Galactic halo (excluding the MS and LA; see Section 2.1.1). It should be noted that the ionized gas masses vary by significant factors depending on if one assumes the ionized and neutral gas are mixed together, or that the ionized gas forms a skin around the neutral component. The numerous H detections along the Magellanic Stream indicate this feature also has a significant amount of ionized gas associated with it (see also Fox et al. 2010).

Absorption line observations are able to detect much lower column density gas at a range of temperatures compared to emission line observations. An important conclusion from the absorption line observations is that multi-phase high-velocity gas covers a significant fraction of the sky. A covering fraction of ~ 81% was found for T ~ 10^4-4.5 K gas using high velocity Si III lines (< N_Si III = 2.6 × 10¹³ cm^-2 >; Shull et al. 2009, Collins et al. 2009), and a similar covering fraction (> 60%) was found for T ~ 10^5-6 K gas using O VI absorption (N_O VI > 2.5 × 10¹³ cm^-2; Sembach et al. 2003). These results can be used to infer that low column density ionized hydrogen covers the majority of the sky (see also Lehner et al. 2012). Models show that the Si III absorbers represent photoionized material, while the O VI is largely from collisionally ionized gas produced as cool clouds interact with a hot halo medium (Sembach et al. 2003). A combination of photo- and collisional ionization is commonly needed in models to reproduce the line ratios for all sightlines (Shull et al. 2011, Fox et al. 2006, Fox et al. 2010, Haffner & Madsen 2011). Lehner & Howk (2011) recently constrained the distance to a collection of ionized HVCs (IHVCs) using various ultraviolet absorption lines detected in the spectra of halo stars. They found the ionized clouds with |V_LSR| < 170 km s^-1 are within ~ 15 kpc and contribute on the order of 10⁸ M in gas to the halo (assuming a metallicity of 0.2 solar; see also Shull et al. 2009). Thus, the largely ionized component of halo gas most likely dominates in mass over the HI component, although one must be careful that the same velocity cuts are made with these comparisons.

In Figures 1 and 2, high-velocity absorption lines are shown over the HI map of HVCs. The absorption line sightlines were not chosen to lie near HI complexes (Sembach et al. 2003, Shull et al. 2009), but in these maps it can be seen that the majority of the absorbers are in the spatial and kinematic vicinity of the HI complexes. In particular, an extension of the Magellanic Stream is evident with absorbers at high negative velocities (see also (Nidever et al. 2010, Lockman et al. 2002, Gibson et al. 2000). There are also extensions of Complexes C, A, and M and the W complexes, and many of the distance constraints for the ionized HVCs are consistent with those of the closest HI HVCs in position-velocity space. This map clearly shows that large scale multi-phase flows are active in the halo. There are some cases where the absorbers do not show a position-velocity link to complexes detected in HI. Most of these are at positive velocities and b > 45°. These largely ionized structures could be longer extensions of the W and LA complexes, gas along local galaxy structures (i.e., toward the Local Supercluster in the north or toward the Sculptor Group for the positive velocity absorber near the South Galactic Pole), or outflows from the disk (e.g., Keeney et al. 2006, Zech et al. 2008).

2.3. Hot Halo Gas

The hot gas ( ~ 10⁶ K) in a spiral galaxy halo traces feedback mechanisms (Section 4.2), and potentially shock-heated IGM (Section 4.1). The extent and mass of this medium is of great interest to cosmological models of galaxy formation, as they predict it extends out to the virial radius and potentially hosts a large percentage of the galaxy's baryons (White & Rees 1978, Maller & Bullock 2004, Fukugita & Peebles 2006, Kaufmann et al. 2006, Sommer-Larsen 2006, Crain et al. 2010).

A hot Galactic halo and/or disk-halo medium is inferred to exist from x-ray observations in emission (Kerp et al. 1999, Wang & McCray 1993) and absorption (O VII and O VIII; (Williams et al. 2005, Wang et al. 2005), and from dispersion measures of pulsars at a range of distances (Taylor et al. 1993, Gaensler et al. 2008). Most of the hot gas detected directly in emission is thought to be within a few kpc of the disk (Fang et al. 2006, Yao & Wang 2007), and this is consistent with other galaxies (see Section 3.3). There is evidence for a lower density medium that extends out to at least the Magellanic System ( ~ 50-100 kpc) from several indirect methods. 1) The abundant detections of high velocity O VI absorption in the Galactic halo is attributed to the interaction of cool halo clouds with the hot halo medium (Sembach et al. 2003). 2) Satellite galaxies are stripped of their gas out to large radii, most likely as they move through this diffuse hot halo medium (Grcevich & Putman 2009, Blitz & Robishaw 2000). 3) The spatial and kinematic structure of the cold halo clouds indicate they are both supported and destroyed by a hot halo medium (see Section 5). 4) There is a low-density bipolar structure of unknown distance detected in x-ray (Bland-Hawthorn & Cohen 2003) and gamma-ray maps (Su et al. 2010) that is consistent with hot gas being sent into the extended halo.

Mass estimates for the hot halo medium are < 10¹⁰ M within 100 kpc of the disk (e.g., Collins et al. 2005, Anderson & Bregman 2010, Yao et al. 2008), and the covering fraction is estimated to be > 60% (Fang et al. 2006, Sembach et al. 2003). As discussed further in Section 4, the amount of hot gas in the halo is below that predicted by cosmological simulations (see also Bregman 2007), but its extent makes it relevant to studies of large scale Galactic winds and/or shock-heated IGM. The precise physical properties of the extended hot halo medium remain uncertain, though densities are generally determined to be between 10^-5 and 10^-4 cm^-2 at 50-100 kpc with the indirect methods mentioned above (Putman et al. 2011b, Hsu et al. 2011, Stanimirovic et al. 2006, Grcevich & Putman 2009, Sembach et al. 2003). The temperature is usually assumed to be 1-2 × 10⁶ K for these halo density estimates, consistent with the x-ray observations and with expectations for gas in equilibrium with a Milky Way mass halo. This temperature also happens to be consistent with the velocity dispersion of the HI HVCs ( ~ 100-120 km s^-1). There is a great deal of additional work required to understand this difficult-to-detect, low density, hot halo medium. In particular, the distribution of this medium and its density at the virial radius need to be measured in order to obtain a more accurate estimate of its total mass.

2.4. Dust, Metals and Molecules

In this section we discuss the observations of dust and metals in halo gas and the limited detections of molecules. Halo dust and metals are direct tracers of the ejection of material from galaxies as they result from stellar evolution and supernovae. Small dust grains are the dominant heating mechanism of the cold neutral medium and metals the dominant coolant, so their abundances dictate the temperature structure of neutral halo clouds (Wolfire et al. 1995). Dust is destroyed by grain-grain collisions and by particle-grain collisions (sputtering) via shocks (Jones et al. 1996, Draine & Salpeter 1979a, Draine & Salpeter 1979b), and therefore halo dust properties can provide information on the dynamical history of the gas such as the launching mechanism. Understanding halo dust is also important for correcting the observed colors of extragalactic objects. The standard Galactic extinction map was made using a single grain size and composition for all Galactic gas, and a single temperature along a given line of sight (Schlegel et al. 1998). Metallicity measurements can further help to determine the origin of the gas (see Section 4), while molecular gas may point to future star formation in the halo.

Dust can be detected in the Galactic halo through far-infrared (IR) emission, depletion of refractory elements, and the reddening of light from distant objects. While far-IR emission provides a clean indication of dust in the disk, it is difficult to detect in the halo given the lower HI column densities and metallicities of halo clouds compared to the disk. HVCs may have less of their metals in dust grains (Wakker et al. 2001) and have cooler (and dimmer) dust. No definitive detections of HVC dust have been made with far-IR emission, although there are tentative detections of Complexes M and C (Peek et al. 2009, Planck Collaboration et al. 2011, Miville-Deschênes et al. 2005). Measuring the depletion of refractory elements onto grains can in principle indirectly show the presence of dust, but it is difficult given the unknown ionization states of the gas across species, the likely variation in inherent abundance patterns for clouds, and even the significant disagreement in the baseline non-depleted solar abundance ratios (Asplund et al. 2009). Nevertheless, there is no sign of the depletion of S and Si onto grains for Complex C (Collins et al. 2007, Tripp et al. 2003) or ionized HVCs (Richter et al. 2009), while evidence for dust depletion has been found for clouds associated with the Magellanic System (Lu et al. 1998, Fox et al. 2010). Reddening of background sources is independent of temperature and ionization state and may be used in the future to detect or rule out dust in Galactic halo gas at interesting levels.

Metallicity measurements trace the star-formation history of halo gas more directly than the dust. Metallicities have been measured primarily only for the large Complex C and Magellanic System complexes (MS and LA; see Figures 1 - 3), although additional estimates will be obtained with future Cosmic Origins Spectrograph (COS) observations (Wakker 2009). Metallicities for Complex C are 10-30% solar, and there is some evidence for mixing with Galactic material at low latitudes (Shull et al. 2011, Collins et al. 2007). The nearby Complex A has a similar measurement (Wakker et al. 2001). The deuterium abundance was also measured for one sightline in Complex C and is consistent with cosmological expectations (N(D I) = 2 × 10¹⁵ cm^-2; D / H = 2.2 × 10^-5; (Sembach et al. 2004). The Magellanic Stream and Leading Arm metallicity determinations are also ~ 10-30% solar (Fox et al. 2010, Gibson et al. 2000, Lu et al. 1998, Sembach et al. 2001). This metallicity, and the relative abundances, have been used to support an SMC origin for MS and LA; however, the metal content of these features depends on when, and from what region, the gas was stripped from the Magellanic Clouds.

Molecular gas has only been detected along a few sightlines through HVCs. H₂ was detected in the Leading Arm and Magellanic Stream ( ~ 10^16-17 cm^-2; (Sembach et al. 2001, Richter et al. 2001). It was not detected toward a sightline through Complex C, but this may be due to a low HI column along the sightline (Richter et al. 2003). CO and H₂ was detected in the bridge of gas connecting the LMC and SMC, which is consistent with the star formation in the Magellanic Bridge (Muller et al. 2003, Lehner 2002).

2.5. The Disk-Halo Interface

Observations of gas at the disk-halo interface help to determine the relationship between halo gas that comes from the disk and halo gas that fuels the disk. For the Milky Way, the gas at the interface between the disk and halo has traditionally been studied separately from the halo gas. The notable exception is the hot gas because the velocity resolution of these data do not allow the two to be separated. Defining individual objects at the disk-halo interface is difficult at any wavelength. Objects are defined with HI data with the large intermediate velocity clouds (IVCs) and a population of small discrete clouds, but at other wavelengths the properties of the entire gas layer are usually presented.

For the cold gas, the thickness of the HI layer in the inner Galaxy is 220 pc, but approaches < 100 pc in the nuclear regions and flares dramatically at R > R (Dickey & Lockman 1990, see Figure 6). In some regions, there are kpc extensions from the disk due to the presence of IVCs or the Galactic warp (Levine et al. 2006, Kalberla & Dedes 2008). There is observational evidence that the HI temperature increases with increasing z-height. This is also seen in the warm and warm-hot gas (see below). Determining the variation in the movement of the gas with z-height is difficult given our location. The model of (Levine et al. 2008) has a vertical falloff of the rotation curve of -22 ± 6 km s^-1 kpc^-1 near the disk, while (Marasco & Fraternali 2011) find lower values of -15 ± 4 km s^-1 kpc^-1. Both the temperature and kinematic gradient of the disk-halo gas is depicted in Figure 5.

Figure 5. A cartoon representing the disk-halo interface given the observations discussed in Section 2.5 and Section 3. The temperature gradient for the diffuse gas is shown with the pink color bar, while the kinematic gradient of the denser neutral gas (HI) is shown by the blue-gray shading. There are distinct populations of large IVCs and small discrete clouds, but in reality the edges of the HI features are not this sharp.

The large HI IVCs are not easy to isolate as individual objects due to their connection to Galactic HI. They are found to reach z-heights of 1-2 kpc and their HI mass estimates are 0.5-8 × 10⁵ M. IVCs have near-solar metallicities (Wakker et al. 2001, Wakker et al. 2008), and dust is readily detected through IR emission. IVCs tend to have somewhat smaller, hotter dust grains than the disk, consistent with dust shattering through shocks (Heiles et al. 1988, Peek et al. 2009, Planck Collaboration et al. 2011). Small amounts of molecular gas have been detected in numerous IVCs with H₂ column densities of 10^14-17 cm^-2, consistent with the presence of dust (Richter et al. 2003). Deuterium has also been detected (Savage et al. 2007).

Recently a large number of small, discrete HI clouds have been identified as a distinct disk-halo population and modeled as clouds rotating with the Galaxy at z-heights of 1-2 kpc (Begum et al. 2010, Saul et al. 2012, Ford et al. 2010, Lockman 2002, Kalberla & Kerp 2009). The GALFA-HI survey has classified the largest population of these clouds ( ~ 1250), and can split the discrete HI clouds at intermediate velocities (20-90 km s^-1) into a cold (T < 5000 K) population that appears to be relatively uniformly distributed and related to the disk, and a warm population that has a bias toward negative velocities and is distributed similarly to the large IVCs (Saul et al. 2012). The HI mass of the clouds depends on their distance and range from ~ 1 M for clouds at high Galactic latitudes at a distance of 1-2 kpc (Saul et al. 2012), to ~ 700 M for clouds toward the inner galaxy with distances obtained with the tangent point method (Ford et al. 2010). The metallicity and dust content of these clouds has not yet been measured.

The warm gas detected in H emission at the disk-halo interface is referred to as the Reynolds or warm ionized medium (WIM) layer (Haffner et al. 2003, Reynolds 1993). This is a layer of warm gas extending ~ 2 kpc above the disk with a volume averaged density of 0.01 - 0.1 cm^-3 and a filling factor of >30% at 1-1.5 kpc (Reynolds 1991, Gaensler et al. 2008, Haffner et al. 2009). As illustrated in Figure 5, at lower z-heights more of the volume is filled in with HI and at higher z-heights the hot gas fills more of the volume (Savage & Wakker 2009, Gaensler et al. 2008). The scale height of gas at multiple temperatures, as well as its relation to the Galactic disk, has been investigated via absorption line studies (Savage & Wakker 2009, Bowen et al. 2008). The cool and warm components have scale heights calculated from the absorption line results that are consistent with that obtained from the emission line observations. The warm-hot gas, that is often referred to as transition temperature gas, is observed with Si IV, C IV and O VI absorption lines and has a larger scale height ( ~ 3-4.5 kpc) than the cold and warm gas. This warm-hot halo gas is found toward the vast majority of the sightlines in the Galactic disk. The disk-halo interface region is discussed further in Section 6.2.