A SUBMILLIMETER GALAXY ILLUMINATING ITS CIRCUMGALACTIC MEDIUM: Lyα SCATTERING IN A COLD, CLUMPY OUTFLOW

Lyman-α Blobs (LABs) are ∼100 kpc scale radio-quiet Lyα emission line nebulae with integrated luminosities up to L_Lyα ∼ 10³⁷ W. LABs were first discovered during the course of deep optical narrowband imaging of the Small Selected Area 22^hr (SSA22) field (Steidel et al. 2000), which was known to contain a significant overdensity of Lyman-break Galaxies (LBGs) at z = 3.09 (Steidel et al. 1998), although Francis et al. (1996) and Keel et al. (1999) had also previously detected extended Lyα emission around luminous galaxies at z = 2.4 (both in high density environments) and extended Lyα emission was reported to be associated with some of the first high-z submillimeter galaxies (SMGs, e.g., Ivison et al. 1998). Subsequent narrowband (and broadband; see Prescott et al. 2012, 2013) surveys of the SSA22 "protocluster" and elsewhere have since revealed a substantial sample of these high redshift nebulae with a range of Lyα luminosities and extents (e.g., Matsuda et al. 2004, 2011; Yang et al. 2009; Erb et al. 2011; Bridge et al. 2013), and often inhabiting dense environments (e.g., Prescott et al. 2008).

Since their discovery, the nature of LABs has been in debate. It was recognized early on that the galaxy halo cooling budget would contain a significant cold component (White & Frenk 1991). Low metallicity gas at T ∼ 10⁴–10⁵ K cools most efficiently via Lyα emission through collisional excitation (Fabian & Nulsen 1977; Katz & Gunn 1991; Katz et al. 1996; Haiman et al. 2000), so one train of thought is that extended Lyα emission implies the presence of potentially pristine cold gas in the circumgalactic medium (CGM). Taking this idea further, many cosmological hydrodynamical simulations suggest that the baryonic growth of massive (M_h > 10¹² M_☉) galaxies at z > 2 (and for lower-mass galaxies at all epochs) is dominated by a mode of "cold accretion" in the form of narrow filamentary streams that penetrate the hot, virially shocked halo gas (Katz et al. 2003; Kereš et al. 2005, 2009; Dekel et al. 2009). Nevertheless, the predicted cold flows have so far eluded unambiguous observation and some recent simulations seem to refute the existence of cold flows which survive to the disk altogether (Nelson et al. 2013). Nevertheless, LABs have been touted as the closest we have come to identifying the filamentary cold mode accretion phenomenon in nature.

While the cosmological gas simulations can track the history of cold gas easily enough, the predicted fluxes of the cooling Lyα emission are far more uncertain, with the dependence of flux on the complex interplay of turbulence, radiative transfer, and local sources of ionizing radiation being taken into account (e.g., Rosdahl & Blaizot 2012). The prescriptions for Lyα radiative transfer through the gas are notoriously complex due to the fact that it is a resonant line (Neufeld 1990) and sensitive to assumptions about the subgrid physics governing the transport of Lyα photons through astrophysical media, as well as the many other possible sources of Lyα photons in and around galaxies. As a result, different simulations that apply bespoke models of Lyα radiative transfer, aiming to reproduce the properties of observed LABs, predict a wide range of physical properties (Lyα luminosities can vary by an order of magnitude) and key observables, such as Lyα extents and surface brightnesses (Faucher-Giguère et al. 2010; Goerdt et al. 2010; Rosdahl & Blaizot 2012).

The cold flow picture is complicated further when one considers the effects of the presence of a luminous source (a starburst galaxy or one hosting an active galactic nucleus, AGN) embedded within the LAB, which appears to be a common situation (e.g., Francis et al. 1996; Keel et al. 1999; Dey et al. 2005; Geach et al. 2005, 2007, 2009; Colbert et al. 2006; Webb et al. 2009; Rauch et al. 2011; Cantalupo et al. 2012). The typical bolometric luminosities of the embedded sources imply an energy balance heavily weighted towards the galaxy, such that only relatively weak coupling, through feedback, would be required to explain the Lyα luminosities (Geach et al. 2009). Again, there is further uncertainty in the type of feedback that would give rise to such extended Lyα emission: proposals have included emergent (supernovae-driven) superwinds which shock heat cold gas swept up in the outflow (Taniguchi & Shioya 2000), photoionization of the CGM by escaping ultraviolet continuum radiation from massive stars or the accretion disks of an AGN, or photoionization via inverse Compton (IC) scattering of cosmic microwave background (or locally generated far-infrared) photons, which are energetically up-scattered by a relic population of relativistic electrons (e.g., Scharf et al. 2003; Fabian et al. 2009).

Geach et al. (2009) rule out IC scattering as a viable LAB formation mechanism using a deep X-ray stack in SSA22, arguing that photoionization by a central AGN (which may be present in up to 50% of LABs) is probably more important, even with fairly modest escape fractions likely to be applicable to the often heavily obscured sources. Recently, Hayes et al. (2011) presented evidence that at least some of the extended Lyα emission must be in the form of scattered Lyα radiation, with Lyα photons generated within a central galaxy and scattering in clouds of H i at large (∼50 kpc) galactocentric radius. The absence of an obvious luminous counterpart in some LABs (e.g., Nilsson et al. 2006; Smith & Jarvis 2007; Smith et al. 2008) has been inferred to be a sign that in some cases at least, cooling radiation alone is sufficient to explain the extended Lyα emission. However, the constraints on the level of AGN or starburst activity are often too poor to rule out the involvement of feedback energy completely.

In practice, it seems likely that a mixture of gravitational cooling and feedback processes are at play in LABs, each imparting an ambiguous observational signature in the Lyα emission which is further muddied by observational limitations (namely, angular and spectral resolution, and poor sensitivity to low surface-brightness features). In any case, it is important to note that both the cooling and "heating" models require a reasonable reservoir of gas in the circumgalactic medium, so that whatever process dominates the Lyα emission, LABs provide a rare opportunity to investigate the astrophysics at the galaxy/intergalactic medium interface close to the formation epoch of massive galaxies, z ≈ 2–3.

The archetypal giant LAB, SSA22-LAB1, is the most scrutinized specimen of this class of object, but it is perhaps also the most controversial. Chapman et al. (2001, 2004) reported the detection of a very bright submillimeter galaxy at the center of the nebula with a 850 μm (James Clerk Maxwell Telescope [JCMT]/SCUBA) flux of ∼17 mJy, implying a galaxy with hyperluminous levels of activity which Geach et al. (2007) identified with a Spitzer-IRAC source likely to be the counterpart. However, it appears that this measurement overestimated the submillimeter emission. Matsuda et al. (2007) observed SSA22-LAB1 at higher resolution (2'') using the SubMillimeter Array (SMA), at a 1σ depth of 1.4 mJy beam⁻¹ at 880 μm. The target was not detected, and the conclusion was that the 850 μm emission reported by Chapman et al. must be extended on scales of several arcseconds and resolved out in the SMA map. The mystery deepened when Yang et al. (2012) reported no detection of SSA22-LAB1 at 870 μm in an APEX/LABOCA observation at a similar resolution to the original SCUBA observation (using the on/off chopping technique in photometry mode). The reported 3σ upper limit of S₈₇₀ < 12 mJy is clearly at odds with the SCUBA detection. Most recently, in an AzTEC/ASTE 1.1 mm map of the field, Tamura et al. (2013) report only a marginally significant 2.7σ (1.9 mJy) flux enhancement at the position of SSA22-LAB1, again in tension with the original SCUBA observations.

In this paper, we present new JCMT/SCUBA-2 (Holland et al. 2013) observations of SSA22-LAB1, taken as part of the SCUBA-2 Cosmology Legacy Survey (Geach et al. 2013; Roseboom et al. 2013) and interpret the results in the context of extremely deep Lyα SAURON integral-field observations (Weijmans et al. 2010) and the recent Very Large Telescope/UV FOcal Reducer and low dispersion Spectrograph (FORS) Lyα polarimetry observations of Hayes et al. (2011). The remainder of the paper is organized as follows. In Section 3, we describe the SCUBA–2 observations. Section 4 presents the main result and we interpret and discuss this in context with other observations in Section 5. We conclude the paper with a hypothesis for the origin of SSA22-LAB1, and propose further questions in Section 6. A ΛCDM cosmology defined by the parameters measured with the Wilkinson Microwave Anisotropy Probe (seven year results including baryonic acoustic oscillation and Hubble constant constraints; Komatsu et al. 2011) is assumed throughout: h = 0.70, Ω_m = 0.27, and $\Omega _\Lambda =0.73$.

The SSA22 field has been observed as part of the JCMT SCUBA–2 Cosmology Legacy Survey (S2CLS). A 30 arcmin diameter map, centered on 22^h17^m363, +00°19'227, has been obtained (using multiple repeats of the PONG scanning pattern). A total of 105 observations were made between 2012 August 23, and 2013 November 29 in conditions when the zenith optical depth at 225 GHz was in the range 0.05 < τ₂₂₅ < 0.1 with a mean of 〈τ₂₂₅〉 = 0.07. The beam-convolved map reaches a 1σ depth of 1.1 mJy beam⁻¹ with an integration time of ∼3000 s per 4 arcsecond pixel. Data are reduced using the smurf makemap pipeline (Chapin et al. 2013), following the procedures described in more detail in Geach et al. (2013). Flux calibration is performed using the best estimate of the flux conversion factor based on observations of hundreds of standard calibrators (Dempsey et al. 2013); this absolute flux calibration is expected to be accurate at the 15% level.

A submillimeter source with S₈₅₀ = (4.6 ± 1.1) mJy (error bar is instrumental noise only) is detected at 22^h17^m260, +00°12'375. This is within 15 of a red Spitzer-IRAC continuum counterpart source (S₈ = 7.6 ± 2.2 μJy, S₈/S_4.5 = 1.3 ± 2.2), close to the center of the Lyα halo which Geach et al. (2007) identified as the central galaxy. To investigate the veracity of the submillimeter detection, we create two "half maps" from the SCUBA-2 data, whereby each half map represents a random 50% of the individual scans used to make the total map. At the sky position of the detection reported above, we measure significant flux densities of (4.8 ± 1.6) mJy and (4.5 ± 1.5) mJy (i.e., 3σ detections with consistent flux densities in independent halves of the integration), which lend credence to the submillimeter detection of SSA22-LAB1.

Chapman et al. (2004) also present a tentative VLA 1.4 GHz radio continuum source and Owens Valley Radio Observatory (OVRO) CO(4–3) detection consistent with this position (although Yang et al. 2012 failed to detect CO(4–3) in SSA22-LAB1 with IRAM PdBI). The SCUBA-2 map in the vicinity of SSA22-LAB1 is shown in Figure 1. The mid-infrared colors of the IRAC counterpart suggest that star formation dominates the energy budget of this source (Colbert et al. 2011). Clearly, the new 850 μm flux density is lower than the S₈₅₀ = 16.8 ± 2.9 mJy reported by Chapman et al. (2001). Formally, the flux density ratio of the initial SCUBA measurement and the new SCUBA-2 flux density is 3.7 ± 1.1; the origin of this disparity remains unclear.

Zoom In Zoom Out Reset image size

The revised 850 μm flux density is consistent with the LABOCA upper limit S₈₇₀ < 12 mJy (Yang et al. 2012) and at the 1σ 1.4 mJy beam⁻¹ limit of the higher-resolution (2'' beam) SMA map of Matsuda et al. (2007), the 850 μm emission would only have to be distributed on a scale of a few arcseconds to fall below the SMA detection rate. Still, the galaxy-integrated 850 μm emission implies a vigorously star-forming galaxy. With no other constraints on the shape of the far-infrared spectral energy distribution, we can only make a rough estimate of the bolometric (total infrared) luminosity. Assuming a nominal single modified blackbody spectrum with β = 2 and T = 20–30 K (Magnelli et al. 2012), normalized to S₈₅₀ = 4.6 mJy (the data do not allow a constraint on either the emissivity or temperature), implies a log (L_IR/L_☉) ≈ 11.8–12.6 where the infrared luminosity is integrated over λ = 40–120 μm (rest frame) and includes a bolometric correction term of 1.91 to account for hot dust emission at mid-infrared wavelengths not modelled by the single blackbody (Helou et al. 1988; Dale et al. 2001; Magnelli et al. 2012).

The integrated luminosity estimated above should be considered with a caveat: the 15'' resolution of the SCUBA-2 map naturally results in a submillimeter flux measurement that is integrated over the equivalent of ∼120 kpc at z = 3.1, thus encompassing (potentially) several individual sources contributing to the 850 μm emission. It is already well known that a significant fraction of submillimeter galaxies (≈35% and potentially up to 50%) detected with coarse resolution instruments break up into several components when observed at higher resolution, as is now possible with sensitive interferometers (e.g., Hodge et al. 2013). Given the negative k-correction in the submillimeter bands on the Rayleigh–Jeans side of the cool dust emission, chance alignments of luminous galaxies across a wide range of epochs could give rise to multiplicity, however, given the likelihood that many high-z submillimeter sources are either physically large, clumpy systems or undergoing mergers means that single detections which resolve into multiple components do not preclude them being considered as part of the same physical system. Obviously, obtaining higher-resolution submillimeter continuum imaging of SSA22-LAB1 is now a goal, since that will address this question.

There are some hints that we might expect the SCUBA-2 850 μm detection to be extended: the failure to detect continuum emission in the 2'' resolution SMA map of Matsuda et al. (2007) with a 3σ upper limit of 4.2 mJy beam⁻¹ suggests that the 850 μm flux is distributed on a scale of at least >2''. At the position of the IRAC counterpart, SSA22-LAB1a identified by Geach et al. (2007), an HST–STIS image reveals several distinct (but extremely faint) ultraviolet components spread over several arcseconds (Chapman et al. 2004). For the sake of argument, in the following, we will refer to the central source as a single ULIRG-class galaxy (since, even if the emission was distributed on scales of several arcseconds, this is far smaller than the Lyα extent of the LAB), but the reader should remember that at all times we are referring to the total infrared luminosity sampled by the JCMT beam, and this could encompass several distinct—but physically associated—sources.

The detection of a luminous galaxy in the core of SSA22-LAB1 is further evidence that Lyα Blobs are generally associated with luminous young galaxies, be it an obscured starburst or an AGN (Geach et al. 2009). SSA22-LAB1 is unique in that it has been studied in far greater detail than any other LAB, and therefore offers the best opportunity to understand the astrophysics of the LAB phenomenon. Here, we revisit the observations of SSA22-LAB1 by Weijmans et al. (2010) and Hayes et al. (2011) and relate these to the confirmation of a luminous central source.

5.1. Spatial Distribution and Spectral Properties of the Lyman-α Emission

Weijmans et al. (2010) present very deep WHT/SAURON integral field unit (IFU) spectral observations of SSA22-LAB01 at 4810 Å <λ < 5350 Å, combining the data with the previous observations of Bower et al. (2004) to form a data cube with a total integration time of nearly 24 hr. The IFU data provide a much better probe of the structure of the Lyα halo than narrowband imaging alone (Steidel et al. 2000; Matsuda et al. 2004).

Aside from Lyα emission associated with the two Lyman Break Galaxies (C11 and C15), there are three main regions of diffuse Lyα emission, labeled R1, R2, and R3 by Weijmans et al. These contain no ultraviolet continuum source, but the two brightest components, R2 and R3 (L_Lyα = 2.3 × 10³⁶ W and 2.7 × 10³⁶ W, respectively), approximately straddle the peak of the 850 μm emission that we have identified with SCUBA-2. R1 is best described as a lower surface brightness structure extending north from R2 towards C15. Is there a physical motivation for associating the diffuse Lyα emission with the central submillimeter source? In Figure 2 (adapted from Weijmans et al. 2010), we present the average Lyα spectra extracted from 2'' radius apertures at the positions R1, R2, and R3 indicated in Figure 1. All the lines are broad, with full width zero intensity widths up to ∼2000 km s⁻¹, and poorly fit by single Gaussian dispersion profiles. As Weijmans et al. show, a double-Gaussian fit is a better model of the line profiles in all three cases (a single Gaussian combined with a Voigt absorption profile also provides an adequate fit to the data). In Figure 2, we show the best-fitting double Gaussian profiles, centered in the rest frame of each component. R2 and R3 are within 180 km s⁻¹ of each other, and both profiles have a dominant red-skewed peak. In contrast, R1 has a slightly more symmetric profile and is separated from R2 by approximately 300 km s⁻¹.

Zoom In Zoom Out Reset image size

For resonant line radiation transfer through an optically thick slab, the emergent line profile is symmetric and double peaked with a minimum at Δv = 0 due to scattering at the line resonance (Neufeld 1990). In an H i cloud, the separation of the peaks is set by the gas temperature and column density, which determine the typical frequency change a photon can experience through scattering. If a large-scale velocity gradient is present, then the line profile becomes asymmetric with either the red or blue peak dominating, depending on the direction of the velocity gradient with respect to the resonant frequency: outflows generally result in a prominent red peak, whereas infall results in a dominant blue peak (z < z_em Lyα Forest absorption suppresses the blue wing whatever the situation). Turbulent motions within the medium will also broaden the emergent line, and asymmetric line profiles can also occur due to foreground absorption of Lyα photons by clumps, or screens, of dust or H i (as is proposed for the line profile observed in LAB2; see Wilman et al. 2005).

It is interesting that the line profiles of R2 and R3 both have dominant red peaks. Given their spatial distribution relative to the SMG, one could interpret this as evidence of a biconical outflow driven by the central ULIRG. Photons from the central source that are back-scattered from the interior, rear wall of the expanding shell will be redshifted out of resonance with the front wall, while photons forward scattered from the front wall remain in resonance with the expanding shell. Note that the bipolar appearance does not necessarily require that the outflow is bipolar: the emergent structure could be equally well explained if ionizing photons only escaped from the central source along preferential directions.

5.2. Polarized Lyα Emission: Scattering of Photons from a Central ULIRG?

5.3. A Coherent Model of SSA22-LAB1

6.1. A Model of LAB Formation: A Combination of "Velvet Rope" and "Bouncer" Feedback?

6.2. Final Comments

We are grateful to the anonymous referee, whose comments have improved the clarity of this paper. We thank Matthew Hayes for providing the polarization data shown in Figure 1 and for helpful discussions, Anne-Marie Weijmans for providing the SAURON data cube, and Rob Crain, Mark Dijkstra, and Claude-André Faucher-Giguère for useful discussions. J.E.G. acknowledges support from the Royal Society by way of a University Research Fellowship. J.S.D. acknowledges the support of the European Research Council via an Advanced Grant, and the contribution of the EC FP7 SPACE project ASTRODEEP (Ref. No. 312725). The James Clerk Maxwell Telescope is operated by the Joint Astronomy Center on behalf of the Science and Technology Facilities Council of the United Kingdom, the National Research Council of Canada, and (until 2013 March 31) the Netherlands Organisation for Scientific Research. Additional funds for the construction of SCUBA-2 were provided by the Canada Foundation for Innovation.