Absorption systems produced by the circumgalactic medium hold key kinematic signatures into unlocking the the behavior of gas around galaxies. High resolution spectroscopy of the background quasars is critical to resolving the velocity substructures within these complex absorption systems. These data can be used to differentiate between scenarios of gas accretion, disk-rotation and outflows.
3.1. Internal or Intrinsic Gas Kinematics
For the most part, metal-line absorption systems are not composed of a uniform velocity distribution of “clouds”, but tend to exist in groupings closer together in velocity with occasional higher velocity clouds offset from the groupings. For a handful high-resolution Mg ii absorption profiles, Lanzetta & Bowen (1992) inferred that their velocity structure was dominated by coherent motions as oppose to random. They further showed that the absorption profiles are consistent with a rotating ensembles of clouds similar to a co-rotating disk. With a larger sample of high-resolution absorption profiles, Charlton & Churchill (1998) applied statistical tests for a variety of kinematic models and concluded that pure disk rotation and pure accretion models are likely ruled out. However, models with contributions from both a rotating disk and infall/halo can reproduce velocities that are nearly consistent with the observed kinematics.
Similar work focusing on the low-ion transitions (such as Si ii) associated with damped Ly α systems, Prochaska & Wolfe (1997) examined if a range of models such as rotating cold disks, slowly rotating hot disks, massive isothermal halos, and a hydrodynamic spherical accretion models could explain the observed absorption kinematics. They determined that thick rapidly rotating disks are the only model consistent with the data at high confidence levels. Their tests suggest that disk rotation speeds of around 225 km s−1 are preferred, which is typical for Milky Way-like galaxy rotation speeds. Furthermore, the gas is likely to be cold since ratio of the gas velocity dispersion over the disk rotation speed must be less than 0.1. These data are suggestive of thick, cold and possibly accreting disks surrounding galaxies.
All these studies are based on absorption-line data alone and therefore it is important to quantify how the absorption kinematics changes with galaxy properties.
Using primarily Lyman limits systems, Borthakur et al. (2015) found Ly α absorption from the circumgalactic medium has similar velocity spreads to that of their host galaxy's interstellar medium as observed via H i emission. The combination of the correlation between the galaxy gas fraction and the impact-parameter-corrected Ly α equivalent width is consistent with idea that the H i disk is fed by circumgalactic gas accretion (Borthakur et al., 2015). Furthermore, they find a correlation between impact-parameter-corrected Ly α equivalent width and the galaxy specific star formation rate suggesting a link between gas accretion driving star formation (Borthakur et al., 2016).
Nielsen et al. (2015) has quantified the Mg ii absorber velocity profiles using pixel-velocity two-point correlation functions and determined how absorption kinematics vary as a function of the galaxy orientation and other physical properties as shown in Figure 3. While they find that absorption profiles with the largest velocity dispersion are associated with blue, face-on galaxies probed along the projected minor axis, which is suggestive out outflows, they find something different for edge-on galaxies. For edge-on galaxies probed along the major axis, they find large Mg ii absorber velocity dispersions and large column density clouds at low velocity regardless of galaxy color, which is used as a tracer of star formation rate. It is suggested that the large absorber velocity dispersions seen for edge-on galaxies (Figure 3 – right) may be caused by gas rotation/accretion where the line-of-sight velocity is maximized for edge-on galaxy inclination. Furthermore, the large cloud column densities may indicate that co-rotating or accreting gas is fairly coherent along the line-of-sight. The only way to test this scenario is to compare the absorption velocities relative to the rotation velocity of the host galaxies.
 |
Figure 3. Shown are pixel-velocity two-point correlation functions (TPCFs) examining how the spread in the pixel velocities of absorption profiles differ for different galaxy properties. The TPCFs presented give the probability of having pixel velocities at a given velocity offset from the absorption redshift defined by the optical depth weighed mean. The shaded regions are the 1σ bootstrap uncertainties. For all azimuthal angles, the TPCFs are shown for blue and red face-on galaxies (left) and edge-on galaxies (right). Note the dramatic differences between the absorber velocity dispersions for blue and red galaxies for face-on orientations (left). The larger velocity spread is suggestive of of high velocity outflows being ejected from the galaxy, while red galaxies are less active. On the other-hand, there is no difference between blue and red for edge-on orientations (right). The velocity spread is comparable to the rotation speed of galaxies, which is consistent with gas accretion and/or gaseous disk co-rotation around the galaxies. |
In addition, some evolution in the circumgalactic medium has also been observed. Blue galaxies do not show an evolution in the velocity dispersions and cloud column densities with redshift (between 0.3 ≤ z ≤ 1), while red galaxies have a circumgalactic medium that becomes more kinematically quiescent with time (Nielsen et al., 2016). This is suggestive that the gas cycle in blue star-forming galaxies is active, be it via accretion or outflows, while red galaxies exhibit little-to-no gas activity. This is consistent with the little-to-no O vi found around z ∼ 0.25 quiescent galaxies from the COS−Halos survey (Tumlinson et al., 2011).
3.2. Relative Gas-Galaxy Kinematics
It is interesting that the internal velocity structure of absorption systems are reflective of their host galaxy type, orientation and redshift, however the question then arises of how/if the circumgalactic medium is kinematically connected to their host galaxies. The most direct measure of gas accretion is observing it down-the-barrel by using the host galaxy as the background source. This method is ideal since there are no degeneracies in the line-of-sight direction/velocity. Although this method does have its difficulties too since metal-enriched outflows and the interstellar medium will tend to dominate the observed absorption over the metal-poor accreting gas. These down-the-barrel gas accretion events have has been observed in a few cases with absorption velocity shifts relative to the galaxy of 80−200 km s−1 (Martin et al., 2012; Rubin et al., 2012). These down-the-barrel gas accretion observations are discussed in detail in the chapter by Kate Rubin. It is still unknown if these observations are signatures of cold accretion or recycled material falling back onto galaxies, yet they are the most direct measure of accreting gas to date.
Using quasar sight-lines, we find that the distribution of velocity separations between Mg ii absorption and their host galaxies tends to be Gaussian with a mean offset of 16 km s−1 and dispersion of about 140 km s−1 (Chen et al., 2010). Although there are much higher velocity extremes, typically expected for outflows, it is interesting that the velocity range is more typical of galaxy rotation speeds having masses close-to or less-than that of the Milky Way.
When galaxy masses are known, one can compare the relative galaxy and absorption velocities to those of the escape velocities of their halos. Figure 4 shows escape velocities, computed for spherically symmetric Navarro–Frenk–White dark matter halo profile (Navarro et al., 1996), as a function of halo mass. It can be seen that very few absorption velocity centroids exceed the estimated galaxy halo escape velocities. This is surprisingly true for a full range of galaxy masses from dwarf galaxies probed by C iv absorption (Bordoloi et al., 2014a) to more massive galaxies probed by O vi (Tumlinson et al., 2011). Consistently, it has been found that the vast majority of absorption systems that reside within one galaxy virial radius are bound to the dark matter halos of their hosts (Stocke et al., 2013; Mathes et al., 2014; Tumlinson et al., 2011; Bordoloi et al., 2014a; Ho et al., 2016). It is worth noting that some of the velocity ranges covered by the absorption profiles are comparable to the escape velocities but are typically for the far wings of the profiles where the gas column densities are the lowest. Therefore two possible scenarios, or a combination thereof, can be drawn: 1) gas that is traced by absorption can be driven into the halo by star formation driven outflows and eventually fall back onto the galaxy (known as recycled winds) and/or 2) the gas is new material accreting from the intergalactic medium. With these data alone, we likely cannot distinguish between these scenarios.
 |
Figure 4. Combined C iv (Bordoloi et al., 2014a) and O vi (Tumlinson et al., 2011) absorption velocity centroids with respect to the systemic redshift of their host galaxies as a function of the inferred dark matter halo mass for star-forming (blue squares) and passive (red diamond) galaxies. The range bars indicate the maximum projected kinematic extent of each absorption system. The histogram represents the distribution of individual component velocities. The dashed lines show the mass-dependent escape velocities at R = 50, 100, and 150 kpc, respectively. Note that all absorption-line systems appear to be bound to their halos and have velocities (and velocity ranges) comparable to galaxy circular velocities. This means that both outflowing and accreting gas could give rise to the observed kinematics. Image courtesy of Jason Tumlinson, Rongmon Bordoloi and the COS−Halos team. |
To test how the circumgalactic gas is kinematically coupled to their galaxy hosts, Steidel et al. (2002) presented the first rotation curves of five intermediate-redshift Mg ii selected absorbing galaxies. Interestingly, they found that for four of the five cases, the absorption velocities lie entirely to one side of the galaxy systemic redshift and consistent with the side expected for rotation. Using simple thick disk-halo models, they concluded that the bulk of Mg ii gas velocities could be explained by an extension of disk rotation with some velocity lag (Steidel et al., 2002). This was further confirmed by Kacprzak et al. (2010a) who also showed that infalling gas or lagging rotation is required to explain the gas kinematics. Using cosmological hydrodynamical galaxy simulations to replicate their data allowed them to concluding that coherently rotating accreting gas is likely responsible for the observed kinematic offset.
There have now been over 50 galaxies/absorbers pairs that have been compared this way and the vast majority exhibit disk-like and/or accretion kinematics (Steidel et al., 2002; Chen et al., 2005; Kacprzak et al., 2010a; Kacprzak et al., 2011a; Bouché et al., 2013; Burchett et al., 2013; Keeney et al., 2013; Jorgenson & Wolfe, 2014; Diamond-Stanic et al., 2016; Bouché et al., 2016; Ho et al., 2016) and some show outflowing wind signatures (Ellison et al., 2003; Kacprzak et al., 2010a; Bouché et al., 2012; Schroetter et al., 2015; Muzahid et al., 2016; Schroetter et al., 2016) while some exhibiting group dynamics (Lehner et al., 2009; Kacprzak et al., 2010b; Bielby et al., 2016; Péroux et al., 2016). Even for systems with multiple quasar sight-lines (Bowen et al., 2016) or for multiply-lensed quasars near known foreground galaxies (Chen et al., 2014) provide the same kinematic evidence that a co-rotating disk with either some lagging rotation or accretion is required to reproduce the observed absorption kinematics. One caveat is that above works have a range of galaxy inclinations and quasar sight-line azimuthal angles, which could complicate the conclusions drawn. It is likely best to select galaxies where the quasar is located along the projected major axis where accreted gas is expected to be located.
Ho et al. (2016) designed an experiment where they selected Mg ii absorbers associated with highly inclined (i > 43 degrees) star-forming galaxies with quasars sight-lines passing within 30 degrees of the projected major axis. Presented in Figure 5 are the rotation curves and the velocity spread of the absorption shown as a function of distance within the galaxy viral radius. It is clear that there strong correlation between the Mg ii absorption velocities and the galaxy rotation velocities. The majority of the Mg ii equivalent widths are detected at velocities less than the actual rotation speed of the dark matter halo (blue squares), while the Keplerian fall-off from the measured rotation curve provides lower limits on the rotation speed of the circumgalactic medium (dashed, black line). The cyan curves illustrate constant Rvrot(R) and show that the infalling gas would have specific angular momentum at least as large as that in the galactic disk, for which some of the gas has comparable specific angular momentum. The Mg ii absorption-line velocity widths cannot be generated with circular disk-like orbit and a simple disk model with a radial inflowing accretion reproduced the data quite well (Ho et al., 2016).
 |
Figure 5. Galaxy rotation curves of 10 galaxies, normalized by the rotation speed as a fraction of the halo virial radius, are compared to the kinematics of their circumgalactic gas. The Mg ii absorption velocities are deprojected such that the velocity shown represents the tangential motion in the disk plane that would give to the observed sight-line velocities. The measured and intrinsic velocity range of each Mg ii absorption system after deprojection is indicated by the green and orange bars with the Mg ii absorption velocity along the quasar sight-line (green circles). Note that the absorption systems align with the expected side for extended disk rotation. Also shown are dark matter halo rotation speed models (blue squares) and the Keplerian fall off from the galaxy rotation curves, which sets a lower limits on the rotation speed in the circumgalactic medium (dashed, black line). The cyan curves illustrate constant Rvrot(R) and indicates that the infalling gas would have specific angular momentum at least as large as that in the galactic disk. |
We have shown that lagging or infalling gas appears to be a common kinematic signature of the circumgalactic gas near star-forming galaxies from 0.1≤ z ≤ 2.5 and is consistently seen for a range of galaxy inclination and position angles. These observations are consistent with current simulations that show that large co-rotating gaseous structures in the halo of the galaxy that are fueled, aligned, and kinematically connected to filamentary gas infall along the cosmic web. (Stewart et al., 2011; Danovich et al., 2012; Stewart et al., 2016; Danovich et al., 2015; Stewart et al., 2013). The predictions and results from simulations are discussed further in the chapters by Kyle Stewart and Claude-André Fauchèr-Giguere. Stewart et al. (2013) demonstrated that there is a qualitative agreement among the majority of cosmological simulations and that the buildup of high angular momentum halo gas and the formation of cold flow disks are likely a robust prediction of ΛCDM. These simulations naturally predict that accreted gas can be observationally distinguishable from outflowing gas from its kinematic signature of large one-sided velocity offsets. Thus, it is plausible we have already observed gas accretion through these kinematic velocity offsets mentioned above.
It is also important to note that the vast majority of systems discussed above have column densities typical of Lyman Limit Systems (N(H i) > 1017.2 cm−2), which are exclusively associated with galaxies. At the lowest column densities of N(H i) < 1014 cm−2, Ly α absorption was found to not mimic the rotation of, and/or accretion onto, galaxies derived from H i observations (Côté et al., 2005). Maybe this is not unexpected since this low column density gas is not likely to be associated with galaxies and likely associated with the Ly α forest (gas associated with galaxies have N(H i) > 1014.5 cm−2− Rudie et al., 2012).