As discussed above, the initial focus in the literature on Na I absorption kinematics biased these samples toward the highest surface-brightness systems at λrest ∼ 5900 Å – i.e., infrared-selected starbursts, massive ellipticals, and bright AGN hosts. It was not until large surveys obtaining deep, high-S/N spectroscopy in the rest-frame ultraviolet were performed that the first constraints on cool gas kinematics toward a significant (> 50) sample of individual, “normal” star-forming galaxies were discussed. Moreover, the Mg II and Fe II absorption transitions covered in these spectra can be significantly stronger than Na I due to a number of factors; e.g., Mg and Fe are more abundant than Na, and their singly-ionized transitions trace a much broader range of temperature and density than neutral Na. UV galaxy spectra therefore have the potential to trace more diffuse material, including the halo gas which is known to give rise to Mg II absorption in background QSO sightlines to projected distances of R⊥ ∼ 100 kpc (Bergeron 1986, Steidel et al. 1994, Kacprzak et al. 2007, Chen et al. 2010a). Rest-frame UV transitions are thus significantly more sensitive to both inflow and outflow, and are accessible at observed-frame optical wavelengths for galaxies at z > 0.3.
The first unambiguous detection of inflow observed down the barrel toward a sample of star-forming galaxies was reported in Rubin et al. (2012). The galaxies were targeted over the course of a high-S/N Keck/LRIS survey of ∼ 100 galaxies at redshifts 0.3 < z < 1.4 (Rubin et al. 2014). This sample, selected to a magnitude limit BAB < 23, spans the star-forming sequence at z ∼ 0.5, and is thus representative of the “normal” star-forming galaxy population (Fig. 2, left). In addition, this survey targeted fields with deep HST / ACS imaging, facilitating a detailed morphological analysis (Fig. 2, right).
Figure 2. The first survey to detect cool gas inflow traced by Mg II and Fe II absorption onto a sample of distant star-forming galaxies (Rubin et al. 2012, 2014). Left: Colored and black points show the sample of ∼ 100 galaxies targeted in Rubin et al. 2012, 2014), and gray contours show the SFR-M* distribution of the underlying galaxy population (Barro et al. 2011). Blue open diamonds indicate galaxies with detected winds, red filled diamonds mark objects with inflows, and black circles indicate objects with neither winds nor inflow. The size of the blue diamonds is scaled with the Mg II equivalent width of the outflow. Winds are detected in ∼ 2/3 of the galaxies, while detected inflows occur at a rate ∼ 6%. Right: Distribution of inclinations (i) for the disk-like galaxies selected from the same study. The distribution for galaxies with winds is shown in blue, and the distributions for galaxies with inflows and without winds/inflows are shown in red and black, respectively. While face-on galaxies (at low i) are significantly more likely to drive a detected wind than galaxies viewed edge-on, the galaxies with detected inflow are nearly all highly inclined. Panels are adapted from Rubin et al. (2012, 2014).
Blueshifted absorption tracing outflows was detected in the majority (∼ 66%) of this sample. Moreover, the detection rate of these winds does not vary significantly with the star formation rate (SFR) or stellar mass of the host, but rather depends primarily on galaxy orientation (Fig. 2, right). This finding is suggestive not only of ubiquitous outflows, but also of an approximately biconical morphology for these flows across the star-forming sequence.
In the same survey, redshifted absorption tracing cool inflow was detected in six of the remaining galaxy spectra (red diamonds in Fig. 2; Fig. 3) with velocities ∼ 80 − 200 km s−1. The galaxies themselves have SFRs ranging from ∼ 1 − 40 M⊙ yr−1, and have stellar masses in the range 9.6 < log M* / M⊙ < 10.5. Perhaps most significantly, five of these six galaxies have disk-like morphologies and are viewed in a nearly edge-on orientation (with inclinations > 55∘; Fig. 2, right). The authors suggest that the preferential detection of inflows toward edge-on galaxies indicates that cool infall is more likely to occur along the plane of galactic disks, rather than along the minor axis. We also note here that higher sensitivity to inflows in more edge-on systems is a natural consequence of biconical winds.
Figure 3. Imaging and spectroscopy of two galaxies with ongoing gas inflow. The left column shows HST / ACS images, and the middle and right columns show Fe II and Mg II transitions in the galaxy spectra. The line profiles are redshifted with respect to systemic velocity (marked with vertical dotted lines). Analysis of the imaging indicates that these galaxies are disk-like with edge-on orientations. The figure is a reproduction of a portion of Figure 1 from the article “The Direct Detection of Cool, Metal-Enriched Gas Accretion onto Galaxies at z ∼ 0.5”, by K. H. R. Rubin et al. (2012, ApJL, 747, 26). ©AAS. Reproduced with permission.
Assuming a metallicity for these flows of Z = 0.1 Z⊙ and adopting constraints on the inflow Mg II and Fe II column densities from their absorption line analysis, Rubin et al. (2012) estimated lower limits for the mass accretion rates onto this sample of six objects in the range ∼ 0.2−3 M⊙ yr−1. These observed flow rates are approximately consistent with the rate of mass flow onto our own Galaxy (Lehner & Howk 2011).
Nearly concurrently, Martin et al. (2012) carried out a similar but completely independent study of cool gas kinematics in Keck/LRIS spectroscopy of ∼ 200 galaxies. The galaxy sample has a somewhat higher median redshift than that of Rubin et al. (2012) (< z > ∼ 0.5 vs. ∼ 1) and is selected to a fainter magnitude limit BAB < 24.0. However, the vast majority of the targets are star-forming, with SFRs ranging from ∼ 1−98 M⊙ yr−1 and stellar masses 8.85 < log M* / M⊙ < 11.3. This survey made use of a somewhat lower-resolution spectroscopic setup for ∼ 70% of the sample (having a FWHM resolution element ∼ 435 km s−1 vs. ∼ 282 km s−1), which in principle limits sensitivity to lower-velocity flows. In spite of this, redshifted absorption profiles were detected in nine spectra (see Fig. 4), yielding a detection rate ( ∼ 4%) consistent with that of Rubin et al. (2012). Martin et al. (2012) also performed a careful analysis of the two-dimensional spectra of these objects, identifying weak nebular line emission offset from the continuum trace at the same velocity as the redshifted absorption in a few (4) cases. They speculated that the inflows in these systems are being fed by relatively dense, star-forming structures (e.g., satellite dwarf galaxies) rather than diffuse accretion streams from the circumgalactic medium.
Figure 4. Resonant absorption lines and [O II] emission profiles for five galaxies with securely-detected inflows from Martin et al. (2012). Galaxy ID numbers are indicated at the top of each column. The solid red vertical lines mark the rest-frame velocity of each transition, and dashed vertical lines mark the wavelengths of nearby resonant transitions. Single-component absorption line fits are shown in the row of Fe II λ2374 profiles, and fits including both an absorption component with a velocity fixed at 0 km s−1 and a “flow” component are overplotted on the Fe II λ 2586 profiles. These objects exhibit strong redshifts in most of the absorption transitions shown. The [O II] profiles are used to determine the systemic velocity. This figure is a reproduction of Figure 16 from the article “Demographics and Physical Properties of Gas Outflows/Inflows at 0.4 < z < 1.4”, by C. L. Martin et al. (2012, ApJ, 760, 127). ©AAS. Reproduced with permission.
Martin et al. (2012) found that the galaxies exhibiting inflows span the range in stellar mass and SFR occupied by the parent sample. They also noted that among the four inflow galaxies for which quantitative morphologies are available, only one has a high inclination (i ∼ 61∘); the remaining three galaxies have i < 55∘ (Kornei et al. 2012). The detection of inflow toward these objects requires that the idea put forth by Rubin et al. (2012) that infall is more likely to be observed along the plane of galactic disks be considered more carefully and investigated with a significantly larger spectroscopic sample.
These two studies have provided us with the first, unequivocal evidence for gas accretion onto distant, star-forming galaxies. However, the physical nature of these flows remains an open question. There are numerous potential sources for the enriched material producing the observed absorption, including gas which has been tidally stripped from nearby dwarf galaxies, or wind ejecta from the central galaxy which is being recycled back to the disk. Indeed, such wind recycling is predicted in numerous cosmological galaxy formation simulations (Oppenheimer et al. 2010, Vogelsberger et al. 2013). Current absorption line data cannot distinguish between these alternatives; however, detailed constraints on the covering fraction or cross section of the inflow from upcoming integral field spectroscopic surveys may aid in differentiating between these scenarios. This topic will be discussed further in Sections 4.2 and 5.