DRAFT VERSION AUGUST 9, 2016Preprint typeset using LATEX style emulateapj v. 12/16/11

ALMA OBSERVATIONS OF LYMAN-α BLOB 1: HALO SUB-STRUCTURE ILLUMINATED FROM WITHIN

J. E. GEACH1 , D. NARAYANAN2,3 , Y. MATSUDA4,5 , M. HAYES6 , LL. MAS-RIBAS7 , M. DIJKSTRA7 , C. C. STEIDEL8 , S. C. CHAPMAN9 ,R. FELDMANN10 , A. AVISON11,12 , O. AGERTZ13 , Y. AO4 , M. BIRKINSHAW14 , M. N. BREMER14 , D. L. CLEMENTS15 ,

H. DANNERBAUER16,17 , D. FARRAH18 , C. M. HARRISON19 , N. K. HINE1 , M. KUBO4 , M. J. MICHAŁOWSKI20 , DOUGLAS SCOTT21 ,D. J. B. SMITH1 , M. SPAANS22 , J. M. SIMPSON20 , A. M. SWINBANK19 , Y. TANIGUCHI23 , E. VAN KAMPEN24 , P. VAN DER WERF25 ,

A. VERMA26 , T. YAMADA27

Draft version August 9, 2016

ABSTRACTWe present new Atacama Large Millimeter/Submillimeter Array (ALMA) 850µm continuum observations of the original Lyman-α Blob (LAB) in the SSA22 field at z = 3.1 (SSA22-LAB01). The ALMA map resolves the previously identified submil-limeter source into three components with total flux density S850 = 1.68± 0.06 mJy, corresponding to a star formation rate of∼150 M yr−1. The submillimeter sources are associated with several faint (m≈ 27 mag) rest-frame ultraviolet sources identifiedin Hubble Space Telescope Imaging Spectrograph (STIS) clear filter imaging (λ≈ 5850Å). One of these companions is spectro-scopically confirmed with Keck MOSFIRE to lie within 20 projected kpc and 250 km s−1 of one of the ALMA components. Wepostulate that some of these STIS sources represent a population of low-mass star-forming satellites surrounding the central sub-millimeter sources, potentially contributing to their growth and activity through accretion. Using a high resolution cosmologicalzoom simulation of a 1013M halo at z = 3, including stellar, dust and Lyα radiative transfer, we can model the ALMA+STISobservations and demonstrate that Lyα photons escaping from the central submillimeter sources are expected to resonantly scatterin neutral hydrogen, the majority of which is predicted to be associated with halo substructure. We show how this process givesrise to extended Lyα emission with similar surface brightness and morphology to observed giant LABs.

Keywords: galaxies: evolution – galaxies: high-redshift – galaxies: halos

1 Centre for Astrophysics Research, University of Hertfordshire, Hat-field, AL10 9AB, UK. j.geach@herts.ac.uk

2 Dept. of Physics and Astronomy, Haverford College, PA, 190413 Dept. of Astronomy, University of Florida, Gainesville, FL, 326084 NAOJ, Osawa, Mitaka, Tokyo 181-8588, Japan5 The Graduate University for Advanced Studies, 2-21-1 Osawa, Mi-

taka, Tokyo 181-8588, Japan6 Stockholm University, Department of Astronomy and Oskar Klein

Centre for Cosmoparticle Physics, SE-10691, Stockholm, Sweden7 Institute of Theoretical Astrophysics, University of Oslo, P.O. Box

1029 Blindern, NO-0315 Oslo, Norway8 California Institute of Technology, 1216 East California Boulevard,

MS 249-17, Pasadena, CA 911259 Dept. of Physics and Atmospheric Science, Dalhousie University, Hal-

ifax, NS B3H 4R2, Canada10 Dept. of Astronomy, University of California Berkeley, CA 9472011 UK ALMA Regional Centre Node12 Jodrell Bank Centre for Astrophysics, School of Physics and Astron-

omy, The University of Manchester, Manchester, M13 9PL, UK13 Dept. of Physics, University of Surrey, GU2 7XH, Surrey, UK14 H. H. Wills Physics Laboratory, University of Bristol, Tyndall Av-

enue, Bristol, BS8 1TL, UK15 Imperial College London, Blackett Laboratory, Prince Consort Road,

London SW7 2AZ, UK16 Instituto de Astrofísica de Canarias, La Laguna, Tenerife, Spain17 Universidad de La Laguna, Astrofísica, La Laguna, Tenerife, Spain18 Dept. of Physics, Virginia Tech, Blacksburg, VA 2406119 Centre for Extragalactic Astronomy, Department of Physics, Durham

University, South Road, Durham, DH1 3LE, UK20 Institute for Astronomy, University of Edinburgh, Royal Observatory,

Blackford Hill, Edinburgh, EH9 3HJ, UK21 Dept. of Physics & Astronomy, University of British Columbia, Van-

couver, BC, V6T 1Z1, Canada22 Kapteyn Astronomical Institute, University of Groningen, PO Box

800, 9700 AV Groningen, Netherlands23 The Open University of Japan, Wakaba, Mihama-ku, Chiba, 261-

8586, Japan24 ESO, Karl-Schwarzschild-Str. 2, D-85748 Garching, Germany25 Leiden Observatory, Leiden University, P.O. Box 9513, NL-2300 RA

Leiden, The Netherlands26 Oxford Astrophysics, Department of Physics, University of Oxford,

Keble Road, Oxford, OX1 3RH, UK27 Tohoku University, Sendai, Miyagi 980-8578, Japan

1. INTRODUCTIONSSA22-LAB01 (z = 3.1, Steidel et al. 2000) is the most

thoroughly studied (Chapman et al. 2004; Bower et al. 2004;Geach et al. 2007, 2014; Matsuda et al. 2007, Weijmans et al.2010; Hayes et al. 2011; Beck et al. 2016) giant (100 kpc inprojected extent) Lyman-α ‘Blob’ (LAB). LABs are intrigu-ing objects: the origin of the extended Lyα emission could bedue to gravitational cooling radiation, with pristine hydrogenat T ∼ 104−5 K cooling primarily via collisionally excited Lyαas it flows into young galaxies (Katz et al. 1996; Haiman et al2000; Fardal et al. 2001). Alternatively, galactic winds, pho-toionization, fluorescence or scattering processes have beenproposed (Taniguchi & Shioya 2000; Geach et al. 2009, 2014;Hine et al. 2016; Alexander et al. 2016). In any scenario thepicture is one of an extended circumgalactic medium (CGM)that is rich in cool gas (be it clumpy or smoothly distributed),and so LABs reveal astrophysics associated with the environ-ment on scales comparable to the virial radius of the massivedark matter halos they trace. Nevertheless, the process (orprocesses) giving rise to the extended line emission remainsin question.

Cen & Zheng (2013) predicted that giant LABs shouldcontain far-infrared sources close to the gravitational centreof the halo. The presence of a central submillimetre-brightgalaxy within SSA22-LAB01 has been in debate (Chapmanet al. 2004; Matsuda et al. 2007; Yang et al. 2012), butwas recently confirmed with the solid detection of an unre-solved S850 = 4.6± 1.1 mJy submillimetre source using theSCUBA-2 instrument on the 15-m James Clerk Maxwell Tele-scope (JCMT) (Geach et al. 2014). Here we present newhigh-resolution Band 7 continuum (λobs = 850µm) observa-tions of SSA22-LAB01 with the Atacama Large Millime-ter/Submillimeter Array (ALMA). These observations resolvethe rest-frame 210-µm emission at z = 3.1, close to the peakof the thermal dust emission and therefore a good probe of

arX

iv:1

608.

0294

1v1

[as

tro-

ph.G

A]

9 A

ug 2

016

2 J. E. GEACH ET AL.

5 asec 1 asec

a b

c

1 asec

S1

Figure 1. Observations of SSA22-LAB01. (left) MUSE continuum-subtracted line image showing the Lyα emission averaged over 4976–5000Å, with contoursat levels of 0.5, 0.7 and 1×10−19 erg s−1 cm−2 Å−1. Black lines show Lyα polarization (Hayes et al. 2011) and black contours show the 1′′ tapered ALMA 850µmemission at >3σ significance; (center) zoom-in showing the ALMA map with the highest resolution image as background with the full resolution and tapered850µm contours overlaid (starting at 3σ, increasing in steps of 1σ), yellow ellipses show the FWHM of the full resolution and tapered synthesized beams; (right)HST STIS optical image of the same region as the central panel, indicating the orientation of the Keck MOSFIRE slit (§2.4) and the position of a faint companionsource we label ‘S1’. We overplot the same contours as the central panel to illustrate how the submillimeter sources associate with emission in the STIS map.

the cold and dense interstellar medium (ISM). Throughoutwe assume a cosmology with ΩΛ = 0.72, Ωm = 0.28 andh = 0.697 = H0/100 km s−1 Mpc−1. At z = 3 1′′ subtends ap-proximately 8 projected kiloparsecs (pkpc).

2. OBSERVATIONS2.1. Atacama Large Millimeter/Submillimeter Array

SSA22-LAB01 was observed in two projects (2013.1.00922.S[PI Geach] and 2013.1.00704S [PI Matsuda]) with a similarconfiguration in Band 7: the full 7.5 GHz of bandwidth cen-tred at 347.59 GHz was used to measure the continuum emis-sion at approximately the same frequency as the SCUBA-2detection of the target. A total of 36–40 12 metre anten-nas were used in the observations, with baselines spanning21-918 m, thus capable of recovering emission on scales ofup to 5′′, with a maximum resolution of 0.4′′. Antenna Tsystemperatures were approximately 100–150 K and the meanprecipitable water vapour column was 0.249 mm. Observa-tions of the phase calibrator source J2206-0031 confirm con-sistent flux scaling between the two projects in the calibrationphase. All calibration was performed using the CASA soft-ware and the visibilities from each project concatenated intoa single measurement set for which the total integration timeis 2860 seconds.

The dirty image reveals the two main components of sub-millimetre emission, and we use this information to sup-ply circular masks (each 2′′ in radius) at 22h17m26.0s,+0012′36.3′′ and 22h17m26.1s, +0012′32.4′′ (J2000) to theCASA clean task. We use a natural weighting of the visibili-ties and clean down to a threshold of 70µJy within 1000 iter-ations. We adopt multi-scale cleaning, allowing model Gaus-sian components of width 0′′ (delta function), 0.5′′, 1′′, 2′′and 3′′. The clean is run twice: at full resolution and withan outer taper of 1′′. To both maps we ‘feather’ the singledish SCUBA-2 (regridded) map to improve the recovery ofthe total flux of the source and potentially improve sensitivityto extended emission. The depth of the map is 40µJy beam−1,increasing to 90µJy beam−1 after tapering. Final maps arecorrected for primary beam attenuation for analysis. The syn-thesized beams are 0.40′′×0.38′′ (PA 23) and 1.01′′×0.98′′

(PA 112) in the feathered full resolution and 1′′ tapered mapsrespectively.

To measure the integrated flux densities and sizes wethreshold the maps at 3σ, where σ is the root mean squarenoise measured around the sources and sum the flux enclosedby each 3σ contour. In the full resolution map there are threedistinct components: a, b and c. Integrated flux density un-certainties are estimated from the root mean squared value,scaled to the number of beams subtended by the 3σ contour.The sizes of each source in the tapered map are measured asthe maximum width of the 3σ contour, with an uncertaintyδθ = 0.6×θbeam/SNR, where θbeam is the full width half max-imum of the beam and SNR = 3.

Note that there is a discrepancy between the single dishflux and the total flux measured in the ALMA map: ∆S850 =2.9±1.1 mJy. Although this is not significant at the 3σ level,we should discuss this potential ‘missing’ flux. First, it isimportant to note that single dish flux densities in the sub-millimeter can be statistically boosted; Geach et al. (2016)measured the flux boosting as a function of signal-to-noiseratio in the SCUBA-2 Cosmology Legacy Survey (where theSCUBA-2 detection of this target originates), finding an aver-age boosting of approximately 30% for SSA22-LAB01. Thearray configuration allows us to recover emission on scales ofup to 5′′. It is unlikely that there is a 850µm emission com-ponent on scales larger than this (40 pkpc at this redshift), butgiven the sensitivity of our observations, it implies that, if real,the missing submillimetre emission is extended on scales ofat least 3′′ (25 pkpc). One possibility is that the emission isspread over a large number of faint clumps around the centralsources – an idea we explore later in the paper. Deeper ob-servations with a slightly more compact array configurationwould be beneficial to address this.

2.2. Multi-Unit Spectroscopic ExplorerIntegral field spectroscopy was performed with the Very LargeTelescope Multi Unit Spectroscopic Explorer (MUSE) instru-ment. The observations were carried out under clear or pho-tometric conditions, between November 2014 and Septem-ber 2015. We used the extended blue setting, resulting

ALMA OBSERVATIONS OF LYMAN-α BLOB 1 3

in a minimum wavelength of 4650Å. Each integration was1500 seconds, and the field was rotated by 90 between eachexposure in order to reduce fixed pattern noise from the in-tegral field units (IFUs) and residual flat-field errors. Datawere reduced with the MUSE pipeline (version 1.0.5), follow-ing standard procedures for bias subtraction, dark current re-moval, flat-fielding, and basic calibration of the individual in-tegrations. Each individual spectrum was fully post-processedto output an image of the field of view, data cubes, and pixeltables, which we examined individually. We computed shiftsbetween the individual exposures using standard centeringtools in IRAF. Finally, we used the MUSE_EXP_COMBINEtask to drizzle and stack all the individually reduced pixtables.

2.3. Hubble Space Telescope Imaging SpectrographThe Hubble Space Telescope Imaging Spectrograph (STIS)observations (project 9174, Chapman et al. 2004) used the50CCD clear filter which provides a wide response in the op-tical, with effective central wavelength 5850Å and full widthhalf maximum throughput 4410Å over a 52′′× 52′′ field-of-view. Six exposures of SSA22-LAB01 were taken in ‘LOW-SKY’ time for a total of 7020 seconds of integration. HSTLegacy Archive calibrated images were combined into a sin-gle deep co-add with 0.1′′ pixels using the DRIZZLEPAC soft-ware (version 2) from the Space Telescope Science Institute.The 5σ sensitivity limit is 27.6 mag (Chapman et al. 2004).

2.4. Keck Multi-Object Spectrometer For Infra-RedExploration

K-band (1.92–2.30µm) spectroscopy of SSA22-LAB01 wasobtained on a single slit of a multislit mask observed dur-ing science verification of the Multi-Object Spectrometer ForInfra-Red Exploration (MOSFIRE, Mclean et al. 2012) on theW. M. Keck Observatory Keck 1 10m telescope. The slitwidth was 0.7′′, resulting in a spectral resolving power ofR ∼ 3600. The total integration time was 5040 seconds, ina sequence of 28 180s exposures using an ABAB nod patternwith a nod amplitude of 3.0′′. The seeing during the observa-tion was estimated to be 0.35′′ full width half maximum. Thedata were reduced by the MOSFIRE Data Reduction Pipeline(Steidel et al. 2014).

3. ANALYSIS & INTERPRETATIONIn the following sections we analyse and interpret the ob-

servations, characterising the ALMA sources in the contextof the LAB environment. We then use a high resolution cos-mological hydrodynamic simulation and radiative transfer todevelop a hypothesis about the role of the central submillime-ter sources in contributing to the extended Lyα emission, pay-ing particular attention to creating mock observations that canbe directly compared to the data, and determining whther sys-tems resembling SSA22-LAB01 are expected in current mod-els of galaxy formation.

3.1. ObservationsAt full (0.4′′) resolution we resolve three main components

that we label ‘a’, ‘b’ and ‘c’ within the SSA22-LAB01 (Figure1). Component ‘a’ is the brightest submillimetre source, lo-cated close to the geometric centre of the LAB and is 1′′ fromcomponent ‘b’; in the tapered map these effectively blend intoa single source, and we consider a+b jointly hereafter. Com-ponent ‘c’ is located 4.5′′ to the south-east and is associated

Table 1Properties of ALMA sources in SSA22-LAB01.

Source R.A. Dec. S850(h m s) ( ′ ′′) (mJy)

SSA22-LAB01 ALMA a+b 22 17 26.01 +00 12 36.5 0.95±0.04SSA22-LAB01 ALMA c 22 17 26.11 +00 12 32.2 0.73±0.05

with strong stellar continuum emission, with a precise red-shift measured from [OIII] and Hβ lines, z = 3.1000±0.0003,placing it within the Lyα nebula (Kubo et al. 2015). Compo-nent a+b is also coincident with stellar continuum emissionof complex, clumpy morphology – it is not clear whether thisis a single bound system and we are simply detecting brightclumps of submillimeter and optical emission, or we are wit-nessing the coalescence of several independent low mass sys-tems.

The total flux density of signal above the 3σ level in thetapered map is S850 = 1.68± 0.06 mJy; Table 1 gives the po-sitions and flux density of the two main components a+b andc. Assuming cool dust emission traces the dense ISM (Scov-ille et al. 2015), the corresponding total molecular gas massis Mgas ≈ 4×1010 M and the average molecular gas surfacedensity is Σgas ≈ 140 M pc−2. If ΣSFR scales linearly withΣgas (Wong & Blitz 2002) we estimate that the submillime-tre sources are forming stars at rates of order 100 M yr−1.We have poor constraints on the shape of the spectral energydistribution of these sources, but we can estimate their inte-grated total infrared luminosities by assuming an appropri-ate range of templates describing the thermal dust emissionof active galaxies. We adopt the Dale & Helou (2002) setof templates, characterised by a parameter α, describing thepower law index for the distribution of dust mass over heat-ing by the interstellar radation field (U): dM(U) ∝ U−αdU ,with most galaxies falling within the range α = 1–2.5; we con-servatively consider this full range when estimating the totalinfrared luminosities of the sources. Redshifting to z = 3.1and normalising to the observed 850µm flux density, we es-timate total infrared (8–1000µm) luminosities in the rangeLIR ≈ (0.2 − 1.9)× 1012L and LIR ≈ (0.2 − 1.5)× 1012Lfor for sources a+b and c respectively. The median in-frared luminosity considering the full range of templates isLIR ≈ 6.0± 0.3× 1011L and LIR ≈ 4.6± 0.3× 1011L forthe two components respectively, where the error bar reflectsthe measurement uncertainty on the integrated ALMA flux.Assuming a star formation rate calibration following Kenni-cutt & Evans (2012) these average luminosities correspond toSFR ≈ 90M yr−1 and SFR ≈ 70M yr−1 respectively, con-sistent with the ISM mass scaling value derived above, al-though the SFRs could be a factor of three higher if the galax-ies are better represented by the more ‘active’ templates of theDale & Helou library.

The emergent luminosity of Lyα emission associated withobscured star formation is LLyα = 0.05 fesc,LyαLIR (Djikstra &Westra 2010), under the assumption of Case B conditions andwhere fesc,Lyα is the escape fraction (of Lyα photons) fromthe galaxy. This corresponds to LLyα ∼ fesc,Lyα1044 erg s−1 perALMA source, similar to the total Lyα luminosity of SSA22-LAB01. From the models of Leitherer et al. (1999), the ioniz-ing flux (ν = 200–912Å) scales like L200−912≈ SFR× fesc,UV×1043 erg s−1 (for 100 Myr of continuous star formation, Solarmetallicity and Salpeter IMF). Thus, for star formation ratesof order 100 M yr−1 and modest high escape fractions, the

4 J. E. GEACH ET AL.

02

46

810

1214

19800 20000 20200 20400 20600

4850 4900 4950 5000

λobserved (A° )

λrest (z = 3.1, A° )

Dis

tanc

e al

ong

slit

(arc

seco

nds)

S1

ALMA c

[OIII][OIII]Hβ

Figure 2. Near-infrared spectrum of ALMA source c and companion ob-tained with Keck MOSFIRE. The background image shows the two dimen-sional spectrum, slightly smoothed for clarity. Strong atmospheric emissionlines are masked. The nod pattern employed results in a positive (red) andnegative (blue) signal in the co-added spectrum offset by 3′′. The one dimen-sional spectra for each source reveal Hβ and [OIII] emission lines, with the5007Å [OIII] line separated by 250 km s−1 between ALMA source ‘c’ andS1, which are separated by an angular distance of 2.1′′ (17 pkpc).

ALMA sources appear to be viable ‘power sources’ for ex-tended Lyα emission through photoionization. The next ques-tion follows naturally: what is the nature of the environmentaround these central star-forming sources?

Like other giant LABs (e.g. Prescott et al. 2012), SSA22-LAB01 is thought to trace a massive dark matter halo of orderMh ≈ 1013M; this is close to the mass where cold flows arenot expected to survive inside the virial radius at this epoch(Dekel et al. 2009). Recently, moving mesh hydrodynamicsimulations suggest cold streams are assimilated into the hothalo gas inside the virial radius even down to halo masses ofMhalo ≈ 1011.5M (Nelson et al. 2013), suggesting that coldflows are not sufficient to explain the extended Lyα luminos-ity of giant LABs. Following arguments put forward by Hayeset al. (2011), Geach et al. (2014) and Beck et al. (2016), wepostulate that extended line emission in SSA22-LAB01 couldarise from Lyα photons generated by central star formation inthe submillimetre sources, escaping and resonantly scatteringin neutral gas associated with a large ensemble of lower-massgalaxies in the same potential well (e.g. Cen & Zheng 2013;Francis et al. 2013). This is consistent with recent spectropo-larimetry observations of SSA22-LAB01 that favour a sce-nario in which Lyα photons undergo long flights from centralsources before scattering in HI in the CGM (Hayes et al. 2011;Beck et al. 2016, although cf. Trebitsch et al. 2016). In Fig-ure 1 we show the polarization vectors of Hayes et al. (2011)that describe a circular pattern roughly centered on the sub-millimeter sources.

The STIS imaging reveals many faint ( fλ ≈10−19 erg s−1 cm−2 Å−1) clumps of UV emission surroundingthe submillimetre sources (Figure 1, Chapman et al. 2004;Matsuda et al. 2007, Uchimoto et al. 2008, see also Prescottet al. 2012). There is no spectroscopically confirmed redshiftfor source a+b, and, unlike source c, it is not unambiguouslyassociated with a STIS counterpart. Instead, there are anumber of STIS clumps at (and around) the position of a+b,and the optical colors of these are broadly consistent witha z = 3.1 redshift (Kubo et al. 2015). In addition, there isa possible HI absorption feature in the Lyα spectrum at

10 kpc 1 asec

Figure 3. Predicted ultraviolet and submillimetre emission of a Mh ≈1013M halo at z ≈ 3 from a hydrodynamic simulation with realistic radia-tive transfer (§3.2). (left) predicted emission in the HST STIS 50CCD bandwith zero noise. Black contours show the predicted 850µm emission, loga-rithmically spaced at 0.25 dex, starting at 0.1µJy. Both the STIS and submil-limeter image has been slightly smoothed for clarity; (right) shows the sameemission, but as would be observed with an identical set-up to the real obser-vations. Background scaling of the mock HST STIS image and 1′′ taperedALMA contours are identical to Figure 1 (we only show the tapered contoursfor clarity here). Only the brightest handful of neighbouring STIS sources aredetectable, with flux densities of fλ & 2× 10−20 erg s−1 cm−2 Å−1. Formally,we detect five STIS sources above a 3σ detection threshold, which is similarto the observations after considering contamination from projections (§3.3).

this location (Weijmans et al. 2010; Geach et al. 2014);thus, source a+b is unlikely to be a chance projection of asubmillimetre source at a different redshift. In the eventthat this ALMA source is not associated with the LAB, thenobviously the main impact on our analysis and interpretationis that there will be approximately a factor two fewer Lyαand UV photons generated by central star formation.

If some (a fraction are likely to be chance projections) ofthe other faint STIS sources represent low mass satellites ofthe ALMA sources, then the cold gas associated with themcould be responsible for scattering Lyα photons emergingfrom the central active galaxies. A key test is actually con-firming ‘membership’ of the STIS sources, and to this endwe have obtained near-infrared spectroscopic confirmation ofone of these faint companion sources ‘S1’ (Figure 2), detect-ing the [OIII] line at z = 3.0968. S1 is 2.1′′ (17 pkpc) fromALMA source ‘c’ with a line-of-sight velocity offset of just250 km s−1. Although not a complete survey by any means,this observation does support the view that the central ALMAsources are accompanied by a population of low-mass satel-lites. With this empirical evidence in hand, we now considerwhether the same observations are predicted by galaxy for-mation and evolution models. In particular, we run a zoomedcosmological hydrodynamical simulation coupled with radia-tive transfer schemes to produce mock observations that canbe directly compared to the data. The main objective is totest whether systems similar to SSA22-LAB01 arise in cur-rent galaxy formation schemes.

3.2. SimulationsWe simulate the formation of a galaxy in a massive (Mh ≈

1013M) halo (by z = 2) utilising a cosmological zoom tech-nique with the hydrodynamics code GIZMO (Hopkins etal. 2014). We employ a pressure-entropy formulation ofsmoothed particle hydrodynamics, which conserves momen-tum, energy, angular momentum, and entropy. We first run acoarse large (144 Mpc3) dark matter-only simulation to iden-tify a halo of interest. This simulation is run at a rela-tively low mass resolution Mdm = 8.4× 108M. At z = 2,we select a massive (2.03× 1013 M) halo, and re-simulate

ALMA OBSERVATIONS OF LYMAN-α BLOB 1 5

100 kpc

Observed (MUSE)

Figure 4. Predicted surface brightness profiles of extended Lyα emis-sion compared to observations. The first panel shows the MUSE ob-servation of SSA22-LAB01 with an isophotal contour at µLyα = 2.2 ×10−18 erg s−1 cm−2 arcsec−2 used to originally classify LABs (Matsuda et al.2004). Additional panels show, at the same scale, the Lyα surface bright-ness maps predicted by our simulation (no noise has been added), smoothedwith a Gaussian beam of FWHM 1′′ to mimic ground-based seeing. Thesemaps are generated by projecting the simulation volume along each of thesix faces of the cube. The central star-forming galaxies (analogous to theALMA sources) are at the center of the cube, and so it is clear that large scalescattering in the halo substructure can form LABs that are highly asymmetricabout the ‘power’ sources, and produce emission line halos with a range ofmorphology. Note also the fainter extended emission predicted by the modelon scales larger then the single isophote shown.

the evolution of this halo at significantly higher resolution(Feldmann et al. 2016), including baryons with particle massmbar = 2.7×105 M. The initial conditions are generated withthe MUSIC code (Hahn et al. 2011) and the minimum bary-onic, star and dark matter force softening lengths are, respec-tively, 9, 21 and 142 proper parsecs at z = 2.

Gas cools using a cooling curve that includes both atomicand molecular line emission, with the neutral ISM broken intoatomic and molecular components following an analytic al-gorithm describing the molecular fraction in a cloud basedon the gas surface density and metallicity (Krumholz, Mc-Kee & Tumlinson 2009). Star formation occurs in moleculargas and follows a volumetric relation ρ? = ρmol/tff, where ρmolis the volume density of molecular gas and tff is the freefalltime. Star formation is restricted to gas that is locally self-gravitating following the algorithms of Hopkins, Narayanan& Murray (2013). Once stars have formed, they impact theISM via a number of feedback channels. In particular, weinclude models for momentum deposition by stellar radia-tion and supernovae, photoheating of HII regions, and stellarwinds (see Hopkins et al. 2014 for details). GIZMO tracks 11metal species (H, He, C, N, O, Ne, Mg, Si, S, Ca, and Fe)with yields from Type 1a and Type II supernova followingIwamoto et al. (1999) and Woosley & Weaver (1995).

To model the ALMA and STIS observations we employ thedust radiative transfer code POWDERDAY which computes thestellar spectral energy distributions (SEDs) of the star parti-cles (constrained by their ages and metallicities) and propa-gates this radiation through the dusty ISM. The gas particlesare projected onto an adaptive grid with an octree-like mem-ory structure (using YT, Turk et al. 2011), and the dust massis assumed to be a constant fraction (40%) of the total metalmass in a given cell (Dwek et al. 1998; Watson 2011). Pho-tons are emitted in a Monte Carlo fashion and absorbed, re-radiated, and scattered by dust; this process is iterated uponuntil the dust temperature has converged. The stellar SEDs arecalculated at run time for each stellar particle utilising FSPS(Conroy & Gunn 2010). We employ the Padova iscochronesand assume a Kroupa IMF. The dust radiative transfer em-ploys HYPERION (Robitaille et al. 2011). We assume a Wein-gartner & Draine (2001) dust size distribution, with RV = 3.15.To create mock observations we consider the observed frame850µm and optical emission predicted by POWDERDAY andHYPERION projected onto a celestial grid at the position ofSSA22-LAB01.

To simulate the ALMA observations we use the CASA taskSIMALMA, setting the simulation parameters to match the realobservations (§2.1), including the same array configuration.The resulting maps have nearly identical synthesized beamand 1σ noise as the data. For the STIS imaging we convolvethe predicted observed frame UV-through-optical emissionwith the 50CCD throughput curve to generate single imagesprojected along the same view angles as the 850-µm images.These images are then convolved with the STIS point spreadfunction and gridded to an identical pixel scale as the obser-vations. Finally, we create a noise image by removing >1.5σfeatures from the real STIS image and replace those pixelswith values randomly drawn from the remainder of the pix-els in the image. The noise-free, PSF-convolved flux image isthen added to the noise image to create the mock observation.

Finally, we consider Lyα radiative transfer of centrally gen-erated photons through the simulation volume. We use thecode TLAC (Gronke et al. 2014, 2015) with the default coreskipping parameter x = 3. The central 400 pkpc of simu-

6 J. E. GEACH ET AL.

lation volume is interpolated onto a regular grid, and wetreat two central star-forming galaxies (SFRs 189 M yr−1 and115 M yr−1) as sources of Lyα photons. To compute the Lyαluminosity emitted by the cells, we use the relation betweenstar formation rate and Lyα luminosity (applicable to solarmetallicity and a Kroupa IMF)

LLyα/ergs−1 = 1.87×1042× (SFR/M yr−1) . (1)

The photons are emitted from a random point inside the cor-responding cell with an emission frequency drawn from a dis-tribution depending on the thermal velocity of the gas in thecell (Gronke et al. 2014, 2015) using the ‘peeling’ algorithm(Yusef et al. 1984; Zheng et al. 2002; Laursen 2010). We fol-low the prescriptions of Laursen (2010) for the Small Mag-ellanic Cloud (Pei et al. 1992; Weingartner & Draine 2001;Gnedin et al. 2008) when calculating the dust content in thegrid, and a full description of this method can be found in thatwork. We compute an effective dust density nd at every cellas

nd = (nHI + fionnHII)(Z/〈ZSMC〉), (2)

where nHI and nHII are the neutral and ionised hydrogen densi-ties respectively at every cell. Z and 〈ZSMC〉 denote the metal-licity at every cell and the SMC average respectively, andfion accounts for the dust content in the ionised gas (Laursen2010). We assume that fion is 1% and note that 〈ZSMC〉 is 0.6dex below the solar value.

3.3. Comparing data and simulationsFigure 3 shows the synthetic ALMA and STIS observa-

tions. We do not expect perfect fidelity with the observa-tions, but we can note some important similarities. As inthe observation, the simulated halo contains two main star-forming galaxies with SFRs 189 and 115 M yr−1 respec-tively; these are well within the range of SFRs we estimatein §3.1. Both galaxies are submillimeter emitters, but onlyone has a flux density high enough to be detected in ourmock ALMA map, with S850 ≈ 0.5 mJy, within a factor oftwo of the observations. In the simulation, both galaxiesare bright STIS sources, whereas this is only true of one ofthre observed ALMA components. The two central galax-ies are surrounded by several STIS-detectable clumps (within∼5′′ or about 40 pkpc) that are of similar flux to those ob-served. Formally, we detect five sources at the >3σ levelin the synthetic STIS image28, with flux densities fλ ≈ (0.2–4)×10−19 erg s−1 cm−2 Å−1. This is compared to eight sourceswith fλ ≈ (0.2–2)×10−19 erg s−1 cm−2 Å−1 over the same areain the real image using identical detection criteria (Figure 1).In the real data some of these sources could be chance pro-jections, and we quantify this contamination by running thedetection on the wider STIS image to measure the averagesurface density of sources in the same flux range. In a random5′′ radius aperture we would expect 4±1 sources by chance,so the predicted abundance of STIS sources in the simulationis in good agreement with the observations.

What are these faint neighbouring sources? In the simu-lation, these STIS clumps are the brightest of a populationof star-forming sub-halos (that are apparent in the left panelof Figure 3) that ‘swarm’ the central star-forming galaxies

28 We detect sources using SEXTRACTOR version 2.19.5 (Bertin &Arnouts 1996) with a detection threshold of 3 continguous pixels above 3×the r.m.s. in the local background. No filtering is applied, and we use a backsize of 32 pixels.

galaxy, bombarding it over a ∼1 Gyr period in a phase of hi-erarchical growth that accompanies the accretion of recycledgas previously heated by stellar feedback at an earlier epoch(Narayanan et al. 2015). So the simulation can broadly repro-duce the ALMA and STIS observations, and we can interpretthis as a demonstration that, as far as can be determined by thedata, the halo population observed in SSA22-LAB01 is con-sistent with the prediction of this particular model of galaxyevolution for a halo of equivalent mass. What of the extendedLyα emission?

Considering the two main star-forming galaxies sources asLyα emitters (with Lyα luminosities scaled from their SFRs,Equation 1), we find that 58% of centrally generated photonsescape the simulation box, and 16% of those scatter at leastonce in HI in the CGM. The majority (77%) of the HI mass inthe central 400 pkpc of the simulation volume is tied to sub-halos, and so it follows that the majority of the Lyα scatteringoccurs in and around satellites, which can be treated as coldclumps within the hot halo gas. Figure 4 shows the projectedLyα surface brightness maps of the simulated halo comparedto the latest integral field observations from MUSE. Resonantscattering in the satellite population gives rise to extendedLyα emission with surface brightness distributions similar toobserved LABs (µLyα ≈ 10−18 erg s−1 cm−2 arcsec−2). The ob-served Lyα surface brightness and morphology is highly ori-entation dependent Lyα sources – the emission need not besymmetric about the central sources that are the source of theLyα photons. This is simiular to the situation in real LABs,for example, the next largest LAB in SSA22, SSA22-LAB02contains an x-ray luminous AGN (Geach et al. 2009; Alexan-der et al. 2016) that is significantly offset from the peak andgeometric center of the LAB. This has important implicationsfor correctly identifying potential luminous galaxy counter-parts to LABs, the apparent absence of which in some systemshas previously been argued in favour of cold mode acceretion(e.g. Nilsson et al. 2006, but see also Prescott et al. 2015).

4. SUMMARYIt is uncontroversial that the intergalactic medium is rich in

cool baryons at z ≈ 3, and recent observations have demon-strated how gas in the cosmic web can be detected in Lyαemission on scales much larger than 100 pkpc via illuminationby nearby quasars (Cantalupo et al. 2014; Martin et al. 2015).Deep stacking experiments have also demonstrated that dif-fuse, extended Lyα emission appears to be a generic featurewithin 100 pkpc of star-forming galaxies at high-redshift, po-tentially linked to scattering in a clumpy CGM (Steidel et al.2011). However it has not been clear how cold gas is actuallydistributed within dark matter haloes and how it relates to thegrowth of central galaxies.

Our new ALMA observations have resolved two activegalaxies at the heart of SSA22-LAB01, with a total star for-mation rate of order 150 M yr−1, and there is evidence tosuggest that these are surrounded by a retinue of lower masssatellites, the brightest of which are detectable in deep HSTSTIS imaging. Our self-consistent hydrodynamic simulationand mock observations lead us to postulate that cold gas as-sociated with the halo substructure – much of which is pre-dicted to be associated with the satellites – acts as a scatter-ing medium for Lyα photons escaping from the central star-forming galaxies, and demonstrate that this can give rise tolarge Lyα nebulae similar to observed LABs. Although res-onant scattering is unlikely to be the only contributor to theextended Lyα emission, we argue that it may be a dominant

ALMA OBSERVATIONS OF LYMAN-α BLOB 1 7

process. We suggest that deep, high resolution Lyα observa-tions offer a route to mapping the distribution and kinematicsof the satellite populations – or, more generally, sub-structure– of distant massive haloes.

ACKNOWLEDGEMENTSWe thank the referee for a constructive report that has im-

proved this work. J.E.G. is supported by a Royal SocietyUniversity Research Fellowship. Partial support for D.N. wasprovided by NSF AST-1442650, NASA HST AR-13906.001and a Cottrell College Science Award. M.H. acknowledgesthe support of the Swedish Research Council, Vetenskap-sradet and the Swedish National Space Board (SNSB), andis Fellow of the Knut and Alice Wallenberg Foundation. H.D.acknowledges financial support from the Spanish Ministry ofEconomy and Competitiveness (MINECO) under the 2014Ramón y Cajal program MINECO RYC-2014-15686. A.V.is supported by a Research Fellowship from the LeverhulmeTrust. The authors acknowledge excellent support of the UKALMA Regional Centre Node. This letter makes use of thefollowing ALMA data: ADS/JAO.ALMA#2013.1.00704S,ADS/JAO.ALMA#2013.1.00922.S. ALMA is a partnershipof ESO (representing its member states), NSF (USA) andNINS (Japan), together with NRC (Canada) and NSC andASIAA (Taiwan), in cooperation with the Republic ofChile. The Joint ALMA Observatory is operated by ESO,AUI/NRAO and NAOJ. This work was supported by theALMA Japan Research Grant of NAOJ Chile Observatory,NAOJ-ALMA-0086.

REFERENCES

Alexander, D. M., et al. 2016, MNRAS in press, 2016arXiv160100682ABeck, M., Scarlata, C., Hayes, M., Dijkstra, M., Jones, T. J. 2016, ApJ, 818,

138Bower, R. G., Morris, S. L., Bacon, R., Wilman, R. J., Sullivan, M.,

Chapman, S., Davies, R. L., de Zeeuw, P. T., Emsellem, E. 2004,MNRAS, 351, 63-69

Cantalupo, S., Arrigoni-Battaia, F., Prochaska, J. X., Hennawi, J. F., Madau,P. 2014, Nature, 506, 63

Cen, R., Zheng, Z. 2013, ApJ, 775, 112Chapman, S. C., Scott, D., Windhorst, R. A., Frayer, D. T., Borys, C., Lewis,

G. F., Ivison, R. J. 2004, ApJ, 606, 85Conroy, C., Gunn, J. E. 2010, Astrophysics Source Code Library.

http://adsabs.harvard.edu/abs/2010ascl.soft10043CDale, D. A., Helou, G. 2002, ApJ, 576, 159Dekel, A., Birnboim, Y., Engel, G., Freundlich, J., Goerdt, T., Mumcuoglu,

M., Neistein, E., Pichon, C., Teyssier, R., Zinger, E. 2009, Nature, 457,451

Dijkstra, M., Westra, E. 2010, MNRAS, 401, 2343Dwek, E., 1998, ApJ, 501, 643Fardal, M. A., Katz, N., Gardner, J. P., Hernquist, L., Weinberg, D. H., Davé,

R. 2001, ApJ, 562, 605Feldmann, R., Hopkins, P. F., Quataert, Eliot, Faucher-Giguére, C-A., Keres,

D. 2016, MNRAS, 458, L14Geach, J. E., Smail, Ian, Chapman, S. C., Alexander, D. M., Blain, A. W.,

Stott, J. P., Ivison, R. J. 2007, ApJ, 655, L9Geach, J. E. et al. 2009, ApJ, 700, 1Geach, J. E., et al. 2014, ApJ, 793, 22Gnedin, N. Y., Kravtsov, A. V., Chen, H.-W. 2008, ApJ, 672, 765Gronke, M., Dijkstra, M. 2014, MNRAS, 444, 1095Gronke, M., Bull, P., Dijkstra, M. 2015, ApJ, 812, 123Hahn, O. & Abel, T. 2011, MNRAS, 415, 2101Haiman, Z., Spaans, M., Quataert, E. 2000, ApJ, 537, L5Hayes, M., Scarlata, C., Siana, B. 2011, Nature, 476, 304Hine, N. K., et al. 2016, MNRAS in press 2016arXiv160504912HHopkins, P. F., Narayanan, D. & Murray, N., 2013 [MNRAS,432, 2657]Hopkins, P. F. 2014, arXiv/1409.7395Iwamoto, K., et al., 1999, ApJS, 125, 439Katz, N., Weinberg, D. H., Hernquist, L. 1996, ApJS, 105, 19Krumholz, M. R., McKee, C. F., Tumlinson, J. 2009, ApJ, 699, 850Kubo, M., Yamada, T., Ichikawa, T., Kajisawa, M., Matsuda, Y., Tanaka, I.

2015, ApJ, 799,38Laursen, P. 2010, Ph.D. Dark Cosmology Centre, Niels Bohr Institute

Faculty of Science, University of Copenhagenhttp://adsabs.harvard.edu/abs/2010PhDT.......254L

Martin, D. C., et al. 2015, Nature, 524, 192Matsuda, Y., et al. 2004, AJ, 128, 569Matsuda, Y., Iono, D., Ohta, K., Yamada, T., Kawabe, R., Hayashino, T.,

Peck, A. B., Petitpas, G. R. 2007, ApJ, 667, 667McLean, I. S., et al. 2012, Ground-based and Airborne Instrumentation for

Astronomy IV. Proceedings of the SPIE, 8446, article id. 84460J, 15 pp.Narayanan, D., et al. 2015, Nature, 525, 496Nelson, D., Vogelsberger, M., Genel, S., Sijacki, D., Keres, D., Springel, V.,

Hernquist, L. 2013, MNRAS, 429, 3353Nilsson, K. K., Fynbo, J. P. U., Moller, P., Sommer-Larsen, J., Ledoux, C.,

2006, MNRAS, A&A, 452, 23Pei, Y. C., 1992, ApJ, 395, 130Prescott, M. K. et al. 2012, ApJ, 752, 86Robitaille, T. P., 2011, A&A, 536, A79Scoville, N. et al. 2015, eprint arXiv:1511.05149Steidel, C. C., Adelberger, K. L., Shapley, A. E., Pettini, M., Dickinson, M.,

Giavalisco, M. 2000, ApJ, 532, 170Steidel, C. C., Bogosavljevic, M., Shapley, A. E., Kollmeier, J. A., Reddy, N.

A., Erb, D. K., Pettini, M. 2011, ApJ, 736, 160Steidel, C. C., et al. 2014, ApJ, 795, 165Taniguchi, Y., Shioya, Y. 2000, ApJ, 532, L13Turk, M. J., et al. 2011, ApJS, 192, 9Watson, D., 2011, A&A, 533, 16Weijmans, A-M., Bower, R. G., Geach, J. E., Swinbank, A. M., Wilman, R.

J., de Zeeuw, P. T., Morris, S. L. 2010, MNRAS, 402, 2245Weingartner, J. C., Draine, B. T. 2001, ApJ, 548, 296Woosley, S. E., Weaver, T. A., 1995, ApJS, 101, 181Wong, T., Blitz, L. 2002, ApJ, 569, 157Yang, Y., et al. 2012, ApJ, 744, 178Yusef-Zadeh, F., Morris, M., White, R. L. 1984, ApJ, 278, 186Zheng, Z., Miralda-Escudé, J. 2002, ApJ, 578, 33