CGM emission is a probe of galactic accretion complementary to absorption measurements. The principal advantage of emission measurements is that they provide spatial resolution in individual halos, which can be used to identify galactic accretion using morphological signatures, such as filaments. At present, the main challenge with emission observations is that circum-galactic gas is typically very faint, so emission measurements are currently only possible for dense gas relatively close to galaxies, or in halos with a luminous quasar that can power CGM emission out to larger radii. High-quality CGM emission observations will, however, become increasingly common over the next few years as a number of optical integral field spectrographs (IFS) with the capacity to detect low surface brightness, redshifted rest-UV CGM emission have recently been commissioned or are planned for the near future. The Palomar Cosmic Web Imager (PCWI; Matuszewski et al., 2010) started taking data in 2009 and the first science results on luminous spatially extended Lyα sources at z ∼ 2−3 have been reported (Martin et al., 2014a; Martin et al., 2014b). Its successor, the Keck Cosmic Web Imager (KCWI, Martin et al., 2010), to be mounted on the Keck II telescope, is currently being developed. The Multi-Unit Spectroscopic Explorer (MUSE; Bacon et al., 2010) on the Very Large Telescope (VLT) completed its commissioning in August 2014 and early science results are being reported (e.g., Fumagalli et al., 2016a; Wisotzki et al., 2016; Borisova et al., 2016). These IFSs provide kinematic information not available with narrowband imaging, and their spectroscopic resolution also enables more accurate background subtraction for line emission.
In this section, we provide a brief summary of simulation results regarding Lyα emission (§ 3.1), UV metal line emission (§ 3.2), and X-ray emission from the CGM (§ 3.3), as well as a summary of the observational status for each.
3.1. Lyα Emission from the CGM
Lyα emission is typically the brightest emission line from the CGM. Our first glimpses of CGM emission have indeed come from spatially extended Lyα sources known as “Lyα blobs” (LABs). The classical LABs have line luminosities up to ∼ 1044 erg s−1 and spatial extents sometimes exceeding 100 proper kpc (Steidel et al., 2000; Matsuda et al., 2004; Yang et al., 2009). The physical nature of LABs is not yet well understood, but several studies suggested that they could be powered by the conversion of gravitational potential into Lyα photons as gas accretes onto halos or galaxies (“cooling radiation”). In this model, weak shocks continuously heat cold accreting gas to temperatures T ∼ 104 K and this energy is efficiently converted into Lyα emission via collisional excitation of HI (Haiman et al., 2000; Fardal et al., 2001; Dijkstra & Loeb, 2009; Goerdt et al., 2010; Rosdahl & Blaizot, 2012). However, a major hurdle in identifying diffuse Lyα radiation with this process is that the expected luminosity remains uncertain at the order-of-magnitude level. Fundamentally, this is because the Lyα collisional excitation emissivity is an exponentially steep function of temperature near T = 104 K, so that small errors in the thermodynamic history of the accreting cold gas can result in large differences in the predicted Lyα luminosity (e.g., Faucher-Giguère et al., 2010). There are two sources of error that can affect the thermodynamic history of accreting gas in cosmological simulations.
The first is the accuracy of the hydrodynamics, which must be able to correctly capture the properties of both weak and strong shocks experienced by accreting gas. The latter point regarding strong shocks is also important for the identification of cold accretion flows in simulations, and is worth expanding on. In both particle-based and grid-based hydrodynamic codes, shocks are often broadened across several resolution elements, which can lead to “in-shock cooling” (e.g., Hutchings & Thomas, 2000; Creasey et al., 2011). This problem arises, for example, when a strong shock should produce hot T ≳ 106 K gas with a long cooling time but in the code the gas cools artificially as it passes through the numerically broadened shock and encounters the peak of the cooling function at T ∼ 105 K. In such circumstances, the hydrodynamic solver can overestimate radiative energy losses via low-energy processes, including Lyα. A specific situation where this effect likely occurs in cosmological simulations is when cool accreting gas impacts a galaxy, where cooling times can be very short owing to the relatively high local gas densities. In this case, not only will there be an error in the predicted Lyα emission, but also artifacts can be introduced in simple algorithms for identifying cold mode accretion based on the maximum temperature to which gas is heated (e.g., Kereš et al., 2005; Kereš et al., 2009b; van de Voort et al., 2011; Nelson et al., 2013).
The second reason for the large uncertainties in predicted Lyα cooling luminosities is the treatment of ionizing radiative transfer. As discussed in § 2.1, cold accretion streams are traced by LLSs, which by definition are optically thick to ionizing photons at the Lyman edge. Since most cosmological simulations to date do not include self-consistent ionizing radiative transfer, they do not accurately capture photoheating in dense self-shielded gas (but see Rosdahl & Blaizot, 2012). In particular, simulations run with a uniform cosmic UVB and assuming optically thin ionization balance overestimate the amount of photoheating within cold streams. Faucher-Giguère et al. (2010) tested the sensitivity of their predictions to the treatment of dense gas and found that different assumptions produced Lyα luminosities differing by up to ∼ 2 orders of magnitude.
Even if a significant fraction of the Lyα emission in LABs comes from cooling radiation, it is difficult to observationally separate cooling radiation from Lyα photons produced by other processes, such as star formation or AGN activity in galaxies. One reason is that bright cooling radiation requires high galactic accretion rates, which lead to SFRs (or AGN activity) sufficient to power most observed diffuse Lyα halos. Diffuse Lyα halos are now in fact generically observed around ordinary star-forming galaxies (e.g., Steidel et al., 2011) and these observations are consistent with the Lyα photons being produced by star formation inside galaxies. There are several ways in which star formation or AGN power can mimic the spatially extended emission expected from galactic accretion. Lyα photons produced inside galaxies can result in diffuse halos due to scattering of the Lyα photons in the CGM (e.g., Dijkstra & Kramer, 2012). Ionizing photons that escape galaxies but are absorbed in the CGM can also produce fluorescent Lyα emission (e.g., Gould & Weinberg, 1996; Cantalupo et al., 2005; Kollmeier et al., 2010). Alternatively, energy injected in the CGM as galactic winds driven by stellar or AGN feedback encounter halo gas can also power diffuse Lyα emission (Taniguchi & Shioya, 2000; Taniguchi et al., 2001). Since Lyα photons typically scatter a large number of times before escaping the CGM, the apparent Lyα spatial and velocity extents are not necessarily representative of the gas producing the Lyα photons.
A more promising avenue for using Lyα emission to identify galactic accretion is to simply use the Lyα photons as a tracers of CGM gas at last scattering. For example, many Lyα sources have a filamentary morphology reminiscent of cosmic web filaments and their extensions into galactic halos as cold streams (e.g., Rauch et al., 2011; Cantalupo et al., 2014; Martin et al., 2014a; Martin et al., 2014b; Rauch et al., 2011; Rauch et al., 2013; Rauch et al., 2016). Of course, one must be careful not attribute every filamentary features to a cold accretion stream, since other phenomena such as tidally stripped gas can appear elongated. Nevertheless, a statistical study of the morphological properties of spatially extended Lyα, along with a comparison to the incidence rate of accreting filaments predicted by cosmological simulations, can test simulation predictions for galactic accretion. In at least one case with detailed spatially resolved kinematic observations (the Lyα image shown in Fig. 7), there is evidence that the filamentary structure traced by Lyα emission smoothly connects to a large, rotating proto-galactic disk (Martin et al., 2015).
Figure 7. Lyα image of the nebula around the UM 287 quasar (‘a') at z ≈ 2.3 (adapted from Cantalupo et al. 2014). The second bright spot labeled ‘b' marks the location of another, optically faint quasar at the same redshift. The extended filamentary morphology suggests that the Lyα emission traces a cold accretion flow. Follow-up integral field observations suggest a smooth kinematic profile consistent with a giant, rotating proto-galactic disk for the brightest portion of the filament, which appears to connect smoothly to the cosmic web (Martin et al., 2015).
Observations of particularly extended and luminous Lyα nebulae at high redshift provide further evidence for compact dense clumps in the gaseous halos of massive galaxies (see § 2.2 for evidence from absorption measurements). Even if a luminous quasar can in principle power the observed Lyα luminosity through reprocessing of its ionizing radiation in the CGM, the integrated recombination rate over the nebula (∝ ∫dV α(T) ne nHII, where α is the hydrogen recombination coefficient) must be sufficiently high to account for the Lyα luminosity attributed to fluorescence. Recent detailed analyses of giant Lyα nebulae indicate that in at least some cases the Lyα-emitting gas must be highly clumped and reach densities ∼ 1 cm−3 (more typical of ISM gas than CGM gas) over spatial scales ∼ 100 kpc (e.g., Cantalupo et al., 2014; Hennawi et al., 2015). If giant proto-galactic disks are relatively common at high redshift, one possibility is that much of the observed Lyα luminosity in giant nebulae originates in fluorescence due to a luminous quasar shining on such a disk rather than CGM gas (e.g., Martin et al., 2015).
The contribution by S. Cantalupo in this volume provides a more exhaustive review of recent results on spatially extended Lyα sources.
3.2. UV Metal Line Emission from the CGM
Because metals are not as abundant, metal line emission is generally significantly fainter than Lyα. Metal lines can, however, provide very useful complementary information on the physical conditions in the CGM. Since most metal lines are optically thin, they are not subject to photon scattering effects and therefore more directly probe the spatial distribution and kinematics of the emitting gas. Furthermore, different metal ions probe different temperature regimes (e.g., Frank et al., 2012; van de Voort & Schaye, 2013; Corlies & Schiminovich, 2016). On the other hand, since metal emission preferentially probes metal-enriched gas, it is at present typically more useful as a general probe of the conditions in the CGM rather than of galactic accretion directly. For example, in an analysis of the UV metal line emission from the CGM of z = 2−4 simulated LBGs from the FIRE project, Sravan et al. (2016) showed the UV metal line emission arises primarily from gas collisionally excited by galactic winds (see Fig. 8).
Figure 8. Example simulation of the UV metal line emission from the CGM of an LBG-mass halo as a function of redshift. This simulation, from the FIRE project, includes strong stellar feedback. Colored lines show UV metal line luminosities within 1 Rvir but excluding the inner 10 proper kpc (a proxy for central galaxies). Star formation rates within 1 Rvir and gas mass outflow rates at 0.25 Rvir are plotted as gray and black lines, respectively. The UV metal line luminosities, star formation, and mass outflow rates are all strongly time variable and correlated. Peaks in CGM luminosity correspond more closely with peaks in mass outflow rates, which typically follow peaks of star formation with a time delay, indicating that energy injected by galactic winds is the primary source of CGM UV metal line emission. Adapted from Sravan et al. (2016).
3.3. X-ray Emission from Hot Halo Gas
Finally, we comment on the use of X-ray observations for probing galactic accretion. In galaxy clusters, it is well established that the hot intra-cluster medium (ICM) is primarily the result of gas shocked heated during the cluster assembly and that the ICM cooling observed in X-rays drives accretion onto galaxies (albeit with a strong apparent suppression of the cooling flows in many clusters, tentatively due to AGN feedback, e.g.; McNamara & Nulsen, 2007). But what processes do X-rays probe in lower-mass halos (e.g., Mulchaey & Jeltema, 2010; Anderson et al., 2013; Li et al., 2016)?
van de Voort et al. (2016) analyzed the X-ray emission in simulated halos from the FIRE project. As for the other FIRE simulations mentioned in this review, these simulations included stellar feedback but no AGN feedback. Figure 9 summarizes summarizes how the soft X-ray emission depends on SFR at z < 0.5, for different halo masses. The X-ray emission around dwarf galaxies is a strong function of their SFR but there is no correlation between LX around massive galaxies or galaxy groups (M500 c > 1012 M⊙). In intermediate-mass halos (M500 c ≈ 1011−12 M⊙), there is a close to linear relation between X-ray luminosity and SFR. These results indicate that diffuse X-rays primarily probe star formation-driven galactic winds in low-mass halos (see also the analytic wind models of Zhang et al., 2014).
Figure 9. Left: Soft X-ray luminosity LX (0.5-2 keV) at z = 0−0.5 as a function of SFR (averaged over 100 Myr) for zoom-in cosmological simulations with stellar feedback from the FIRE project. Crosses with the same color belong to the same galaxy at different times (halo masses can be read off from the panel on the right). Right: The power, α, of the correlation between LX and SFR (LX ∝ SFRα) for individual halos as a function of halo mass. The X-ray emission around dwarf galaxies is a strong function of their SFR, while halos with M500 c ≈ 1011−12 M⊙ exhibit a correlation close to linear. There is no correlation between LX and SFR for the most massive halos, because hot, virialized halo gas produces more X-rays than star formation-powered winds in those halos. Thus, X-ray emission is sensitive to gas accretion onto non-dwarf halos at low redshift (including Milky Way-mass halos, galaxy groups, and galaxy clusters) but primarily probes galactic winds in dwarfs. Adapted from van de Voort et al. (2016).