To pinpoint the 3D distribution of gas around the Milky Way and to estimate its total mass, the different gas phases (in the density-temperature phase-space) need to be considered. In the following, we separate the Milky Way halo gas in three such phases: (1) neutral/molecular gas, (2) warm ionized gas with temperatures T ≤ 105 K, and (3) hot ionized gas with T > 105 K. First, we discuss the angular distribution of these phases, their galactocentric distances, and their total mass considering recent observational results. Then, we provide estimates on the gas-accretion rate of the Milky Way based on these observations. Note that many of these aspects have also been discussed in previous reviews (Wakker & van Woerden 1998; Richter 2006; Putman, Peek & Joung 2012) and in the book on HVCs (van Woerden et al. 2004).
As discussed earlier, radio observations in the H i 21 cm line have become the most powerful method to study the distribution and internal structure of neutral halo clouds (Bajaja et al. 1985; Hulsbosch & Wakker 1988; Wakker & van Woerden 1991; Hartmann & Burton 1997; Morras et al. 2000; Kalberla et al. 2005; McClure-Griffiths et al. 2009; Winkel et al. 2010). Absorption-line measurements against bright extragalactic background sources (e.g., quasars) provide additional information on the chemical composition of the neutral gas and its connection to the other circumgalactic (ionized) gas phases.
In Fig. 2 we show the sky distribution of the 21 cm IVCs (upper panel) and HVCs (lower panel) plotted in galactic coordinates in an Aitoff-projection centered on l = 180∘. Because of their different kinematics and intrinsic properties, it is useful to discuss the properties of 21 cm IVCs and HVCs individually.
 |
Figure 2. (upper panel) and high-velocity gas (lower panel) in an Aitoff projection centered on l = 180∘. The maps show the sky distribution of neutral gas in the Galactic halo. The maps have been generated from different data sets described in Wakker (2004), Kalberla et al. (2005), and by Tobias Westmeier (priv. comm.). For the IVCs we show (color-coded) the deviation velocity of the gas from a simple model of Galactic rotation (see Wakker 2004) in the range |vdev| = 30−90 km s−1. For the HVCs, we display the color-coded LSR velocity (|vLSR| = 100−500 km s−1). Individual neutral IVC and HVC complexes (see Wakker 2001, 2004) are labeled with numbers. For IVCs: (1) IV Arch, (2) LLIV Arch, (3) IV Spur, (4) Outer Arm, (5) Complex L, (6) AC Shell, (7) Complex K, (8) PP Arch, (9) IV-WA, (10) Complex gp. For HVCs: (1) Complex C, (2) Complex A, (3) Complex M, (4) Complex WA, (5) AC Cloud, (6) Complex H, (7) Complex L, (8) Leading Arm of MS, (9) Magellanic Stream, (10) Complex GCN. |
The sky distribution of IVCs indicates that intermediate-velocity gas has predominantly negative radial velocities. The most prominent IVC features (see labels in Fig. 2) are in the northern sky the IV Arch and its low-latitude extension (LLIV Arch), IV Spur, the low-velocity part of Complex L, Complex K, and the low-velocity part of the Outer Arm. In the southern sky, there is the Anticenter (AC) shell and the Pegasus-Pisces (PP) Arch (see Kuntz & Danly 1996; Wakker 2001). The total sky covering fraction of neutral IVC gas is fc ≈ 0.30 for column densities N(H i) ≥ 1019 cm−2 and deviation velocities |vdev| = 30−90 km s−1 (Wakker 2004; see caption of Fig. 2). Many of the IVC features are spatially and kinematically connected with 21 cm disk gas. The typical H i column densities in IVCs in the 21 cm surveys lie in the range N(H i) = 1019−1020 cm−2, but some cores in the IV Arch and the Outer Arm exhibit column densities above 1020 cm−2 (Fig. 2). At positive radial velocities, there is the low-velocity extension of HVC Complex WA in the north and Complex gp in the south. The positive velocity IVCs are much smaller in angular size and column density and they exhibit a very complex internal structure.
Direct distance measurements (mostly using the bracketing technique) place the IVCs relatively close to the Milky Way disk with z heights < 2.5 kpc, typically. The most massive IVC complexes, for example, have distances of 0.8−1.8 kpc (IV Arch), ∼ 0.9 kpc (LLIV Arch), and 0.3−2.1 kpc (IV Spur; see IVC compilations by Wakker 2001, 2004 and references therein). Most of the IVCs thus belong to the disk halo interface, which possibly represents a crucial component for gas accretion processes in the Galaxy, as will be discussed below. For the most prominent, large-scale IVC complexes, the 21 cm data indicate individual H i masses on the order of 1−8 × 105 M⊙ (e.g., IV Arch, LLIV Arch, IV Spur; Wakker 2001). The total neutral gas mass of the large-scale IVCs thus can estimated to be ∼ 106 M⊙.
Absorption-line measurements indicate that the metallicity of most of the large northern IVCs is relatively high with typical values between 0.5 and 1.0 solar (e.g., Wakker 2001; Richter et al. 2001a, 2001b). For the Outer Arm, Tripp & Song (2012) derive a lower metallicity of 0.2−0.5 solar, suggesting that this gas (albeit being close to the disk) might have an extragalactic origin, such as many HVCs (see below). Another interesting feature is the core IV 21, which has a metallicity of just 0.4 solar (Hernandez et al. 2013) at a z height of ∼ 300 pc above the disk, thus also pointing toward an extragalactic origin. Dust is also present in IVCs, as is evident from the observed depletion patterns of heavy elements in intermediate-velocity gas (Richter et al. 2001a, 2001b; Wakker 2001; see also Savage & Sembach 1996) and from the observed excess of extra-planar infrared emission in the direction of IVCs (Desert et al. 1988, 1990; Weiss et al. 1999). Also molecules (H2, CO) have been detected in several IVCs (Richter et al. 2001b, 2003; Wakker 2006; Gillmon et al. 2006; Hernandez et al. 2013; Röhser et al. 2016), but the molecular gas fraction is very small, so that the molecular phase does not contribute significantly to the overall IVC masses. Yet, the presence of molecular gas implies the presence of substantial small-scale structure in the gas down to AU scales (Richter, Sembach & Howk 2003). Small-scale structure is also evident from high-resolution 21 cm data, which show, next to the coherent large-scale 21 cm IVC complexes, a population of several thousand compact H i clumps at z = 1−2 kpc (e.g., Lockman 2002; Kalberla & Kerp 2009; Saul et al. 2012). These clumps, that have very small masses of only 101−104 M⊙, may represent cloudlets that have condensed out of the ambient multi-phase medium. They are raining down to the disk, thus fueling it. Because of their small cross section for absorption spectroscopy, the metal and dust content of these clumps remains unknown so far.
In view of the measured overall chemical composition of the large IVCs and their location in the disk-halo interface, the favored scenario for the origin of near solar-metallicity IVCs is, that these structures represent the back-flow of cooled gas from the galactic fountain process (Shapiro & Field 1976; Houck & Bregman 1990); i.e., they originate from metal-enriched gas that has been ejected from the disk by supernova explosions that is now cycling back to the disk due to gravitational forces.
The sky distribution of the Milky Way HVCs is far more complex than that of the IVCs (Fig. 2, lower panel); HVCs span a huge LSR velocity range of ∼ 700 km s−1.
Among the most prominent Milky HVCs is Complex C, which covers ∼ 1500 square-degree on the northern sky in 21 cm emission, which is about 4 percent of the entire sky (Wakker 2004). Complex C is a cloud that presumably is being accreted from the IGM or from a satellite galaxy (e.g., Sembach et al. 2004). Another prominent 21 cm HVC is the Magellanic Stream (MS) in the south. The MS also covers an area of ∼ 1500 square-degree in 21 cm, but it spreads over the entire southern sky, forming a coherent stream of neutral gas (D'Onghia & Fox 2016). The MS represents a tidal feature expelled from the Magellanic Clouds as they approach the Milky Way halo (e.g., Gardiner & Noguchi 1996; Connors et al. 2006; Besla et al. 2010, 2012; Nidever et al. 2010; Diaz & Bekki 2011, 2012). Over its entire body the MS spans a distance range of d = 50−100 kpc (or even further) from the Galactic disk, suggesting that it extends over several hundred kpc in the outer halo of the Milky Way (Putman et al. 1998, 2003; Stanimirovic et al. 2002, 2008; Brüns et al. 2005). Other prominent Galactic HVCs are Complex A, Complex H, the Anti-Center Cloud, and Complexes WA−WE. Their positions are indicated in Fig. 2. The individual properties of all HVC complexes are discussed in detail in Wakker (2001).
The entire HVC population of the Milky Way shown in Fig. 2 has a total sky covering fraction in 21 cm of fc ≈ 0.35 for neutral gas column densities N(H i) ≥ 7 × 1017 cm−2 (Murphy, Lockman & Savage 1995; Wakker 2004 and references therein). The covering fraction reduces to fc ≈ 0.15 for larger column densities N(H i) ≥ 2 × 1018 cm−2. The H i column densities in HVCs follow a well-defined column-density distribution function of the form f(NHI) ∝ NHI−β with β = 1.42 for log N(H i) ≥ 18 (Lockman et al. 2002). With the exception of the MS, all HVCs for which direct distance information from the bracketing method is available, are located within 20 kpc. For instance, the Complexes A & C have distances d ≈ 10 kpc (van Woerden et al. 1999; Wakker et al. 2007; Smoker et al. 2011; Thom et al. 2008), the Cohen Stream and Complex GCP are at d = 5−15 kpc (Wakker et al. 2008), and the HVC towards the LMC has d ≈ 9 kpc (Richter et al. 2015). For these structures in the inner Milky Way halo the distances estimated indirectly from the measured Hα fluxes agree well with the values derived from the bracketing method (e.g., Bland-Hawthorn & Putman 2001; Putman et al. 2003; Tufte et al. 2002). Combining the 21 cm emission in Milky Way HVCs and in M31 halo clouds, Richter (2012) predicts that the H i covering fraction in HVCs around Milky Way-type galaxies declines exponentially with galactocentric distance with fc < 0.01 for d > 70 kpc. From deep 21 cm observations of the M31 halo (Westmeier et al. 2007) further follows that also the so-called compact high-velocity clouds (CHVCs), isolated high-velocity 21 cm gas clumps with very small angular sizes of < 2deg (Braun & Burton 1999; de Heij, Braun & Burton 2002), are located at d < 50 kpc, disproving a previous scenario in which CHVCs are regarded as gas-filled dark matter halos residing in the Local Group (Blitz et al. 1999; Braun & Burton 1999).
From the 21 cm data and the available distance information it follows that the total neutral HVC gas mass is MHI,HVC ≈ 2.5 × 108 M⊙ (Wakker 2004; Brüns et al. 2005). The MS contributes with more than 60 per cent to this mass (MHI,MS ≈ 1.6 × 108 M⊙ for d = 55 kpc), while the other HVCs have substantially smaller masses (e.g., MHI,Complex C ≈ 5 × 106 M⊙; Wakker 2004). Since the two Magellanic Clouds are located at only ∼ 80 kpc distance, their interstellar gas content adds to the gas mass being accreted by the Milky Way. Therefore, if we add the neutral gas mass of the Magellanic Clouds and their gaseous interconnection, the Magellanic Bridge, the total H i budget in the Milky Way halo sums up to a value of MHI,MWhalo ≈ 1.3 × 109 M⊙, which is ∼ 20 percent of the neutral gas mass in the Milky Way disk (MHI,MWdisk ≈ 7 × 109 M⊙; Ferriere 2001).
Other tracers of predominantly neutral gas in the Milky Way halo are absorption lines of neutral and low ions, such as O i, N i, Ar i, S ii with transitions in the UV, and Ca ii and Na i in the optical regime. From a survey of high-velocity Ca ii / Na i absorption in the Milky Way halo against several hundred extragalactic background sources, Ben Bekhti et al. (2008, 2012) derive a covering fraction of fc ≈ 0.50 for log N(Ca ii) > 11.4, which is ∼ 2 times higher than the 21 cm covering fraction. These results further indicate the widespread presence of cold, neutral gas structures away from the large 21 cm complexes. Such structures possibly are too small to be seen in all-sky 21 cm survey because of the limited angular resolution of these surveys in combination with beam-smearing effects.
While most of the IVCs have near-solar metallicities, the metal abundance in many HVCs is substantially lower (by a factor 5−10, typically). For Complex C, the (mean) metallicity has been constrained to 0.15 solar (Wakker et al. 1999; Richter et al. 2001a; Tripp et al. 2003; Collins, Shull & Giroux 2003; Sembach et al. 2004). The main body of the Magellanic Stream also has a metallicity of only 0.1 solar (Fox et al. 2010, 2013), but the MS contains a filament that is more metal rich (0.3 solar; Richter et al. 2013; Gibson et al. 2001). Similarly, Complex A most likely has metallicity of only ∼ 0.1 solar (Wakker 2004). Such low metallicities are in line with the idea that HVCs represent gas infalling from the pre-enriched intergalactic medium (or intragroup gas), but the clouds may also trace material stripped from satellite dwarf galaxies as they are being accreted by the Galaxy. Moreover, it cannot be ruled out that some of the gas has been part of the Milky Way a long time ago, then was ejected (at relatively low metallicity) by a former Galactic outflow or wind, and now is raining back towards the disk ("intergalactic fountain"). An example for this latter scenario is the Smith Cloud (also called Complex GCP), which has a metallicity of 0.5 solar and is believed to originate in the outer Galactic disk (Lockman et al. 2008; Hill, Haffner & Reynolds 2009; Fox et al. 2016).
There appears to be only little dust in neutral and ionized HVCs (e.g., Wakker & Boulanger 1986; Bates et al. 1988; Tripp et al. 2003; Richter et al. 2001, 2009; Williams et al. 2012). The few tentative detections of far-IR emission in some HVCs (e.g, Miville-Deschênes et al. 2005; Peek et al. 2009; Planck Collaboration 2011) remain inconclusive with respect to their dust abundance. Diffuse molecular gas is present only in high-column density regions of the Magellanic Stream (Sembach et al. 2001; Richter et al. 2001c, 2013), the Magellanic Bridge (Lehner 2002; Murray et al. 2015), and in a dense clump of an HVC projected onto the LMC (Richter et al. 1999). However, the molecular component is not of importance for the total mass of HVCs.
Finally, it is worth noting that some of the above mentioned neutral halo clouds (e.g., Complex L) exhibit a radial velocity range that extends from values below 100 km s−1 to values above this threshold, i.e., these complexes can be regarded as both IVCs and HVCs, although they each represent a single, kinematically coherent structure. The question arises, whether the separation of IVCs and HVCs as different halo-cloud populations is justified, or whether they just represent the same population of objects with just different radial velocities. As discussed above, distance measurements place the IVCs within 2.5 kpc of the Milky Way disk, while most of the HVCs are located much further away. This, together with the on average higher metallicity of IVCs compared to HVCs, indeed indicates that low-velocity halo clouds (with LSR velocities typically < 100 km s−1) predominantly reside in the disk-halo interface, while high-velocity halo clouds (with LSR velocities typically ≥ 100 km s−1) predominantly trace gas at d = 2−100 kpc.
Warm ionized gas in the Milky Way halo is even more widespread than the neutral gas traced by 21 cm emission. We here define warm-ionized gas as gas that is predominantly ionized (i.e., with small neutral gas fractions) and has a temperature < 105 K. Circumgalactic gas at such temperatures is expected to be photoionized by the combined ionizing radiation from stars in the Milky Way disk and the ambient extragalactic UV background at z = 0 (see model by Fox et al. 2005). Warm ionized gas in the Milky Way halo can be detected either in emission in recombination lines such as Hα or in high-velocity UV absorption of so-called intermediate ions that have lower ionization boundaries in the range 15−40 eV (e.g., C iii, N iii; Si iii, Si iv, Fe iii; Morton 2003; R17).
In Fig. 3, upper panel, we show the sky distribution of Hα emission in the range 0.3−100 Rayleigh, based on data obtained from the Wisconsin Hα Mapper (WHAM; Haffner et al. 2003, 2016). As can be seen, Hα emission is widespread at latitudes b < 30∘, situated in distinct coherent spatial structures such as lobes and arches. These features reflect the complex motions of diffuse ionized gas (DIG) in the disk-halo interface (DHI) that is believed to be shaped by the on-going star-formation in the Milky Way disk. Based on the derived gas densities and volume filling factors (Reynolds et al. 2012; Haffner et al. 2003), the total mass of ionized gas in the DHI of the Milky Way can be estimated to be ∼ 108 M⊙, in line with estimates for extra-planar ionized gas in other low-redshift disk galaxies (e.g., NGC 891; Dettmar 1990).
 |
Figure 3. All-sky maps of Hα emission in the Milky Way (upper panel) and high-velocity UV absorption of Si iii absorption (lower panel) in an Aitoff projection centered on l = 180∘ The maps show the sky distribution of diffuse ionized gas (DIG) in the lower and upper Galactic halo, respectively. The Hα map (kindly provided by Matt Haffner) has been compiled from data of the Wisconsin Hα mapper (WHAM; Haffner et al. 2003, 2016). It shows emission from relatively dense gas that predominantly resides in the inner halo and disk-halo interface of the Milky Way (DIG layer; Reynolds 1991). The Si iii data stem from the high-velocity (|vLSR| = 100−500 km s−1) UV absorption survey from Richter et al. (2017; hereafter referred to as R17) using 265 HST/COS spectra of extragalactic background sources. Diffuse ionized gas has a substantially larger sky covering fraction (∼ 2−3 times higher) than the neutral gas (Fig. 2). |
 |
Figure 4. Velocity profiles of high-velocity UV absorbers in the Galactic halo along three lines of sight (from R17), based on HST/COS data. Absorption profiles of various transitions from Si iii, Si ii, C iv, and C ii are shown, where the high-velocity absorption at |vLSR| ≥ 100 km s−1 is indicated in blue. We also show the H i 21 cm emission profiles (from GASS/EBHIS data) for the same sightlines in the lowest panels. The individual high-velocity absorption features towards RX J1230.8+0115 and SDSS J123426.80+072411.3 trace gas streams in the halo that are predominantly ionized (without H i 21 cm counterpart), while the high-velocity absorption towards ESO−031−G−008 traces neutral and ionized halo gas related to the MS (UV absorption plus H i 21 cm emission). |
Many of the 21 cm HVCs at much larger distances from the disk are also detected in Hα emission (e.g., Weiner & Williams 1996; Tufte, Reynolds & Haffner 1998; Bland-Hawthorn et al. 1998), indicating that the neutral gas clouds are surrounded by envelopes of ionized gas, whose masses are comparable with or even larger than the neutral gas body (e.g., Fox et al. 2004). Also the Magellanic Stream at d = 50−100 kpc is detected in Hα (e.g., Putman et al. 2003; Fox et al. 2014), proving that the MS is surrounded by substantial amounts of warm H ii. From the models of Bland-Hawthorn et al. (2007) it follows that the mass-weighted H ii column density in the MS is >1020 cm−2, thus larger than the mass-weighted H i column density. A similar conclusion was drawn by Fox et al. (2014), who determine an ionized-to-neutral hydrogen mass ratio of ∼ 3 based on the absorption strength of intermediate and high ions associated with the 21 cm body of the Stream.
The most sensitive ions to trace the warm ionized gas in the halo are C iii and Si iii with strong transitions in the UV at 977.02 Å (C iii) and 1206.50 Å (Si iii; see Richter et al. 2016). In their recent legacy survey of high-velocity UV absorption in the Milky Way halo, R17 found that warm ionized halo gas, as traced by Si iii at velocities |vLSR| ≥ 100 km s−1 and column densities log N(Si iii) ≥ 12.1, has a covering fraction as high as fc = 0.74, confirming earlier results based on much smaller samples (Collins, Shull & Giroux 2009; Shull et al. 2009; Lehner et al. 2012; Herenz et al. 2013). This covering fraction is more than twice the value obtained for the neutral HVCs from the 21 cm observations (see above). In the lower panel of Fig. 3 we show the sky distribution of high-velocity Si iii absorption, which can be directly compared with the 21 cm HVC map (Fig. 2, lower panel). Like the Hα emission, high-velocity Si iii absorption is often associated in radial velocity with the 21 cm HVC features, even if located several degrees away from the 21cm contours. This further implies that HVCs represent coherent multi-phase gas streams (with a neutral gas body surrounded by an ionized gas layer) that move through the Milky Way halo (Lehner et al. 2012). Figs. 2 and 3 also indicate that there are several regions in the high-velocity sky that exhibit pronounced Si iii absorption but do not show significant large-scale H i. In contrast to the 21 cm HVCs and their ionized envelopes, gas in these regions can be regarded as coherent ionized gas streams in which patchy condensations of cooler, neutral gas clumps are embedded. Particularly interesting are the regions l > 200∘, b > 0∘ and l < 120∘, b < 0∘, which form a velocity dipole on the sky in UV absorption (Fig. 2; R17; Collins, Shull & Giroux 2003) in a direction that forms the major axis of the Local Group cosmological filament (N14). The observed kinematically distinct absorption features at high positive and high negative radial velocities possibly indicate that the Milky Way is ramming into ionized intragroup gas because it follows the general flow of galaxies in the direction of the Local Group barycenter (Peebles et al. 2001, 2011; Whiting 2014), while it is moving away from Local Group gas in the opposite direction along the filament (R17). Therefore, the Milky Way's accretion of warm ionized gas might be strongly influenced by the local galaxy environment and cosmological structure formation in the Local Group (N14). This important aspect will be further discussed in Sect. 3.2.
The ionized gas components that are associated with the 21 cm features obviously have the same distances as the neutral halo clouds. They also have comparable metallicities, if the gas has not yet been mixed with the ambient hot coronal gas. This implies that the majority of the diffuse ionized halo clouds that are not associated with the MS are located at d < 20 kpc, while the ionized envelope of the Stream is at d = 50−100 kpc. The exact angular extent of the Stream's ionized gas component is unknown, but it may well cover 30−50 percent of the entire sky (R17; Fox et al. 2014). From their survey of high-velocity UV absorption towards Galactic halo stars with known distances (d < 15 kpc) and extragalactic background sources Lehner & Howk (2011) and Lehner et al. (2012) find that the sky-covering fraction of high-velocity UV absorption increases only marginally from the halo-star sample to the QSO sample, if velocities |vLSR| ≤ 170 km s−1 are considered. This indicates that HVCs in this velocity range are predominantly located at d < 15 kpc. In contrast, HVCs with absolute LSR velocities larger than 170 km s−1 (e.g., the MS) are only seen against extragalactic background sources, demonstrating that the gas is located at d > 15 kpc, and being in line with the distance constraints for the neutral gas (see above). The halo-star sample of Lehner & Howk (2011) and Lehner et al. (2012) covers only a limited fraction of the sky, however, so that the possible presence of more distant ionized gas structures even at low velocities (in particular in the directions of a possible Local Group filament) cannot be ruled out with these data.
The total mass of diffuse ionized high-velocity gas in the Galactic halo is dominated by the extended envelope of the MS (Fox et al. 2014; R17). Assuming d = 55 kpc and calculating the amount of H ii from the observed ion abundances in combination with an ionization model, both studies obtain a gas mass of the ionized component of the MS of MMS ≈ 1−3 × 109 M⊙. This mass would be substantially higher, if some of the gas from the MS was located at larger distances. For instance, if the distance of MS would lie in the range d = 100−150 kpc (see Besla, et al. 2012; Jin & Lynden-Bell 2008; Bland-Hawthorn et al. 2013), the mass of the ionized component of the MS would be as large as ∼ 3−7 × 109 M⊙, thus very close to the total ISM gas mass in the Galactic disk (Ferriere 2001). The contribution of high-velocity absorbers at d < 20 kpc to the ionized gas mass in the halo is small instead; their gas mass sums up to a total value of no more than MHVCs,d < 20 kpc = 2 × 107 M⊙ (R17). This value still is comparable to or even higher than the mass of the neutral gas in the same distance range.
Ever since the prediction of Lyman Spitzer in 1956 on the existence of a Galactic Corona (see Sect. 1), the search for a low-density, high-temperature (T > 105 K) gaseous medium that surrounds the Milky Way has been of high priority for astrophysicists, as the Corona links the observed properties of the Galaxy to cosmological structure formation (see, e.g., Oort 1966). From more recent theoretical work (e.g., Maller & Bullock 2004), it is indeed expected that all MW-type galaxies are surrounded by massive, hot gaseous halos of typical mass of 1011 M⊙ and temperature T∼ 106 K (the virial temperature of the galaxy's DM halo). If some fraction of the gas was able to cool, it would sink towards the disk, feeding the galaxy with fuel for future star formation. Therefore, hot coronal gas may serve as a huge baryon reservoir from which MW-type galaxies gain their gas. In addition, the hot Milky Way halo might be further fed with gas from a possible large-scale outflow from the Galactic center region (Fox et al. 2015; Lehner et al. 2012; Zech et al. 2008; Bland-Hawthorn & Cohen 2003; Su, Slayter & Finkbeiner 2010).
Despite the obvious importance of the hot, ionized circumgalactic gas phase for galaxy evolution, our knowledge about the properties and spatial extent of hot coronal gas in the Milky Way still is very limited. This is because it is very difficult to detect such coronal gas that is expected to have very low densities (nH < 10−3 cm−3), in particular in the outer regions of the halo. Observational evidence for the existence of a hot Milky Way Corona comes from observations in the X-ray regime, where the gas can be observed in emission or in absorption against extragalactic X-ray point sources. Using ROSAT data, Kerp et al. (1999) systematically searched for X-ray emission spatially associated with neutral HVCs and reported several positive detections, e.g., in the direction of Complex GCN, Complex C, and Complex D. In Fig. 5, we show as an example the ROSAT emission map of hot halo gas in the direction of Complex GCN from the study of Kerp et al. Other observations of the soft X-ray background support the existence of hot coronal gas in the Milky Way (e.g., Kuntz & Snowden 2000). X-ray absorption of O vii and O viii in the Galactic halo has been reported by several groups (Wang et al. 2005; Fang et al. 2002, 2003, 2006; Mathur et al. 2003; Bregman 2007; McKernan, Yaqoob & Reynolds 2004; Williams et al. 2005; Gupta et al. 2012; see also Miller, Kluck & Bregman 2016), but the interpretation of these low-resolution spectra is afflicted with systematic uncertainties (see Richter, Paerels & Kaastra 2008). Also pulsar dispersion measures have been used to constrain the properties of hot coronal gas in the Milky Way halo (e.g., Gaensler et al. 2008). All these observations are biased towards the regions with the highest gas densities in the Corona, however, so that the bulk of the hot gas detected in this manner presumably resides in the inner halo at d < 20 kpc (e.g., Rasmussen et al. 2003). It therefore remains unknown whether the coronal gas really extends to the virial radius of the Milky Way (Rvir ∼ 260 kpc; e.g., Tepper-García et al. 2015) and how it connects to gas gravitationally bound to the Local Group. From the UV observations of O vi in the thick disk of the Milky Way (Widmann et al. 1998; Savage et al. 2003; Wakker et al. 2003; Savage & Wakker 2009) indeed follows that there are large amounts of extra-planar warm-hot gas at vertical heights z < 5 kpc.
 |
Figure 5. Left panel: X-ray emission map of hot gas in the direction of HVC complex GCN, based on ROSAT 0.25 keV soft-X ray background (SXRB) data (see Kerp et al. 1999 for details). The position of the background quasar Mrk 509 is indicated with the black dot. In this direction, hot halo gas has been detected in high-velocity O vi absorption using FUV data (Sembach et al. 2003; see also Winkel et al. 2011). Right panel: model of the SXRB emission in the same direction, based on 21 cm H i data from foreground gas (causing photoelectric absorption of the background X-ray photons) and assuming a homogenous distribution of the X-ray background (Kerp et al. 1999). Both maps indicate the presence of hot coronal gas in the Galactic halo in this area of the sky. Maps kindly provided by Jürgen Kerp. |
Indirect evidence for the widespread presence of hot gas in the Milky Way halo comes from the many UV absorption-line detections of high-velocity O vi (Sembach et al. 2003; Wakker et al. 2003), which has a sky covering fraction as large as fc = 0.60−0.85. O vi arises in warm-hot gas at T ∼ 3 × 105 K and is believed to trace the interface regions between the hot coronal gas and cooler halo clouds embedded therein (i.e., the neutral and ionized HVCs; see above). Isolated, strong O vi absorption at high velocities is also seen in the direction of the Local Group barycenter and, in particular, towards the quasar Mrk 509 in HVC Complex GCN (Fig. 5), where enhanced X-ray emission and high-velocity Si iii absorption is observed (Collins, Shull & Giroux 2005; see above). The coincidence of UV absorption and X-ray emission further indicates that there is an excess of hot gas in this direction that possibly is related to highly-ionized intragroup gas near the Local Group barycenter, towards which the Milky Way is moving (see Sect. 2.2).
To estimate the baryon content of the Milky Way's coronal gas, its radial density profile needs to be constrained. Due to on-going accretion of gas and stars from satellite galaxies and the IGM, substantial deviations from a simple hydrostatic density distribution are likely. Even for a hot halo that is not perfectly hydrostatic, however, the average gas density in the coronal gas is expected to decrease for increasing distances to the disk. From X-ray absorption, spectra Bregman & Lloyd-Davies (2007) estimate a gas density of nH ≈ 8 × 10−4 cm−3 for the inner halo (d < 20 kpc). From pulsar dispersion measures instead follows that the average coronal gas density must be smaller, nH < 8 × 10−4 cm−3, in line with studies that estimate nH indirectly from considering the interaction between the cool HVCs and the ambient hot medium (nH ≈ 2 × 10−4 cm−3; Grcevich & Putman 2009; Peek et al. 2007; Tepper-Garcia et al. 2015). Because of the unknown extent and the unknown gas properties at the virial radius, the total mass of the Milky Way's hot coronal gas is very uncertain. For d < 250 kpc the total mass is estimated to be MCorona ≈ 1010−1011 M⊙ (Anderson & Bregman 2010; Yao et al. 2008; Gupta et al. 2012; Miller & Bregman 2013, 2015; Fang, Bullock & Boylan-Kolchin 2013; Salem et al. 2015), in line with the idea, the Milky Way's hot Corona represents a huge baryon reservoir. However, the hot coronal gas must cool and condense into streams of denser gas to be able to sink to the Milky Way disk, i.e., it must transit through the diffuse ionized and/or neutral phase to contribute to the gas-accretion rate. To understand the details of this important phase transition, hydrodynamical simulations are required, which will be discussed in Sect. 3.
2.4. Gas-accretion rates from observations
The observations presented in the previous sections suggest that HVCs (and IVCs) represent coherent entities of multi-phase gas that move within the Milky Way halo (Lehner et al. 2012; R17). Independent of their individual origin (tidal interactions with satellite galaxies, infall from the IGM or intragroup medium, condensations from the hot coronal gas, the back-flow of material expelled previously from the disk), these streams of gas trigger the Galaxy's present-day gas-accretion rate as defined in equation (1) and we will discuss their contribution to dMgas,halo / dt in the following.
While the 3D distribution of neutral and diffuse ionized gas around the Galaxy (i.e., Mgas,halo, d) is constrained by observations, the space motion of the gas is not, as only the radial component of the velocity can be observed. In lack of further information, the mean infall velocity often is assumed to be constant for all halo clouds, although it is likely that vinfall spans a large range for the IVCs and HVCs and systematically depends on d (see Sect. 1). For the MS, recent studies assume ⟨ vinfall ⟩ = 100 km s−1 (e.g., Fox et al. 2014), a value that also has been used for other HVCs (e.g., Complex C; Wakker et al. 1999, 2008). A more complex model for vinfall is presented by Putman, Peek & Joung (2012), where they try to separate for the HVC population the azimuthal velocity component from the accretion velocity in the Galactic center (GC) direction and derive ⟨ vinfall,GC ⟩ ≈ −50 km s−1. This value is consistent with the mean radial velocity of HVCs of ⟨ vrad ⟩ ≈ −50 km s−1, but the most distant halo structures, such as the MS, might have somewhat larger accretion velocities (Mathewson et al. 1974).
Because of its large mass, the MS dominates by far the total present-day gas accretion rate of the Milky Way. From their UV absorption-line survey Fox et al. (2014) derive a mass inflow rate of neutral and diffuse ionized gas from the Stream of dMMS / dt ≈ 2 M⊙ yr−1 for a fixed distance of the MS of d = 55 kpc and ⟨ vinfall ⟩ = 100 km s−1. At this distance and infall velocity, the gas (or better said, a fraction η of it; see equation (2)) would reach the disk in ∼ 540 Myr. If, instead, the distance of the Stream was d = 100 kpc, then it would take ∼ 1 Gyr for the gas to reach the disk and the accretion rate would be higher by a factor of ∼ 2, because the estimate of MMS depends on its assumed distance. Taking d = 55 kpc as a conservative lower limit and adding the gas mass associated with the Magellanic Clouds and the Magellanic Bridge (with all these components together forming the Magellanic System), the total accretion rate from all these components sums up to a value of dMMSys / dt ≥ 3.7 M⊙ yr−1; Fox et al. 2014)
Based on the observed properties discussed above, the contribution of the other individual, 21 cm-selected HVC Complexes at d < 20 kpc to the mass-inflow rate (including both neutral and ionized gas) is expected to be rather small (e.g., Complex C: 0.1−0.2 M⊙ yr−1; Complex A: 0.05 M⊙ yr−1; Cohen Stream: 0.01 M⊙ yr−1; see Wakker et al. 1999, 2007, 2008; Thom et al. 2006, 2008; Putman, Peek & Joung 2012; R17). Putman, Peek & Joung (2012) derive a maximum accretion rate of 0.4 M⊙ yr−1 for all HVCs except the MS. If we consider only the neutral gas mass in the Galactic HVC population, the total H i gas accretion rate from all 21 cm HVCs (including the MS) comes out to dMHI / dt = 0.7 M⊙ yr−1 (Richter 2012), again assuming that the MS is at d = 55 kpc.
The general distribution of UV-absorbing gas in the halo (and its total mass) yet implies, that the ionized component (independent of whether it is associated with H i emission or not) dominates the gas accretion not only for the MS, but also for the nearby HVCs at d ≤ 20 kpc. From their absorption-line survey towards halo stars and extragalactic background sources, Lehner & Howk (2011) determine an accretion rate of predominantly ionized high-velocity gas at d < 15 kpc of 0.45−1.40 M⊙ yr−1. Putting it all together, R17 estimate from their UV absorption-line survey of 265 sightlines a total gas-accretion rate of neutral and ionized high-velocity gas in the halo (including gas from the Magellanic System) of dMHVC / dt ≥ 5 M⊙ yr−1. This limit is higher by a factor of > 2 than the current star-formation rate of the Milky Way (∼ 0.7−2.3 M⊙ yr−1; e.g., Robitaille & Whitney 2010; Chomiuk & Povich 2011).
The contribution of the 21 cm IVCs and the DIG in the disk-halo interface (DHI) at z-heights < 2.5 kpc to the Milky Way's gas-accretion rate is significantly smaller than that of the neutral HVCs. Considering the IVC distances and neutral-gas masses discussed in Sect. 2.1.1, the neutral gas-accretion rate from IVCs is only ∼ 0.01−0.05 M⊙ yr−1. A much higher gas-accretion rate can be determined considering the ionized gas reservoir in the DIG (Sect. 2.2). In principle, the overall gas flow in the DIG of Milky-Way type is expected to be strongly influenced by the various feedback processes from the disk (e.g., radiative and mechanical feedback from supernovae, stellar winds, and AGN; e.g., Bland-Hawthorn & Maloney 2002; MacLow & Klessen 2004; Springel & Hernquist 2005; Marasco, Marinacci & Fraternali 2013), but the exact role of these processes in the gas-circulation cycle of the Milky Way's DHI is uncertain. Assuming that at least half of the diffuse ionized gas in the disk-halo interface is currently being accreted (the rest being related to outflowing gas), the observational constraints imply dMDHI / dt = 1−2 M⊙ yr−1 for vinfall ≤ 20 km s−1, thus in line with the current star-formation rate (see also Fraternali et al. 2013). A net-infall of ionized gas is in line with the observed Hα kinematics (Haffner et al. 2003).
Some interesting conclusions can possibly be drawn from these numbers. Obviously, the amount of large-scale neutral gas in the disk-halo interface is much smaller than the amount of large-scale neutral gas at larger distances; it is also much smaller than the amount required to keep up the current star-formation rate in the disk. Infalling neutral gas structures thus might be disrupted and ionized when entering the disk-halo interface, where it might re-cool and condense again before it enters the disk. This re-processing of infalling gas in the disk-halo interface is sometimes referred to as “quiet accretion”. The galactic-fountain flow is believed to play a crucial role in the the gas cooling and condensation processes (Marinacci et al. 2010; Armilotta, Fraternali & Marinacci 2016). The infalling gas thus might enter the disk in the form of low-velocity, mildly-ionized gas clumps or tiny 21 cm drops (Lockman 2002; Begum et al. 2010; Ford, Lockman & McClure-Griffiths 2010; see Sect. 2.1), thus in a form that is difficult to identify observationally.
Another possible reason for the apparent discrepancy between the neutral gas budget in the DHI and that at larger distances might be the existence of neutral DHI gas that is “hidden” to us: clouds that have low radial velocities (LVCs; see Sect. 1.1), similar to those in the disk, but that reside in the (lower) Galactic halo (e.g., Zheng et al. 2015; Peek et al. 2009). Finally, also gas from the outer disk might contribute to the fueling of star formation in the inner regions of the Milky Way disk through a radial inflow of gas (e.g., Elson et al. 2011; Sellwood & Binney 2002; see also Putman, Peek & Joung 2012 and references therein).
Obviously, observations alone cannot provide a full insight into the complex processes that govern the past, present, and future gas-accretion rate of the Milky Way. Hydrodynamical simulations represent an important toolkit to further study the dynamics and physical properties of gas falling toward Milky-Way type galaxies and to pinpoint the overall mass-inflow rate to the disk. The most relevant of these aspects will be discussed in the following section.