Discovery of a Lya emitting dark-cloud within z ~ 2.8 SMMJ02399-0136 system

TL;DR

This study uses integral field spectroscopy to discover a luminous, dark Lyα-emitting cloud near a high-redshift quasar, revealing anisotropic UV emission and complex nebular kinematics over a large spatial extent.

Contribution

First detection of a luminous, dark Lyα emitter near a z~2.8 quasar, demonstrating quasar UV emission anisotropy and complex nebular dynamics.

Findings

Lyα nebula extends over 140 kpc with high luminosity.

Discovery of a dark Lyα emitter with no optical counterpart.

Evidence of anisotropic UV radiation from the quasar.

Abstract

We present Keck/KCWI integral field spectrograph observations of the complex system surrounding SMM J02399$-$0136 (a lensed $z=2.8$ sub-mm galaxy), including an associated Ly$\rm \alpha$ nebula, a dust-obscured, broad-absorption-line quasar, and neighboring galaxies. At a 3$\sigma$ surface brightness contour of 1.6$\times$ 10$^{-17}$ erg s$^{-1}$ cm$^{-2}$ arcsec$^{-2}$, the Ly$\rm \alpha$ nebula extends over 17 arcsec ($\gtrsim 140$ physical kpc) and has a total Ly$\rm \alpha$ luminosity of $2.5 \times 10^{44} \, \rm erg \, s^{-1}$ (uncorrected for lensing). The nebula exhibits a kinematic shear of $\sim$ 1000 km s$^{-1}$ over 100 pkpc with lowest velocities east of SMM J02399$-$0136 and increasing to the southwest. We also discover a bright, Ly$\rm \alpha$ emitter, separated spatially and kinematically from the nebula, at a projected separation of $\approx$60 kpc from the quasar. This source has no clear central counterpart in deep $Hubble \, Space \, Telescope$ imaging, giving an intrinsic Ly$\rm \alpha$ rest-frame equivalent width greater than 312 $ \rm \mathring{A}$ (5-$\sigma$). We argue that this 'dark cloud' is illuminated by the quasar with a UV flux that is orders of magnitude brighter than the emission along our sightline. This result confirms statistical inferences that luminous quasars at $z>2$ emit UV radiation anisotropically. Future KCWI observations of other lines, e.g. Ly$\rm \beta$, He$\rm _{II}$, C$\rm _{IV}$, etc, and with polarimetry will further reveal the origin of the Ly$\rm \alpha$ nebula and nature of the dark cloud.

Tables

Table 1. Table 1 The properties for individual source in SMMJ02399

Source	RA	DEC	${\lambda }_{\mathrm{detected}}$	redshift	Aperture	SB$\left(Ly\alpha \right)$	L${\left(Ly\alpha \right)}^a$
	J2000	J2000	$\mathring{A}$			${10}^{-17}{\mathrm{ergs}}^{-1}{\mathrm{cm}}^{-2}{\mathrm{arcsec}}^{-2}$	${10}^{43}{\mathrm{ergs}}^{-1}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
L1	02:39:51.86	-01:35:58.15	4625.5	2.8048$\pm$0.0004	$\sim$0.7”$\times$0.6”	8.3$\pm$1.8	3.5$\pm$0.6
L2	02:39:52.04	-01:35:57.27	4621.6	2.7985$\pm$0.0004	$\sim$0.7”$\times$0.6”	8.3$\pm$1.7	3.5$\pm$0.6
L3	02:39:51.63	-01:36:04.56	4634.9	2.81259$\pm$0.00008	$\sim$1.4”$\times$1.2”	3.0$\pm$0.6	1.3$\pm$0.2
Entire nebula							25.0$\pm$3.6

Equations

$$\begin{array}{rl}{SB}_{\mathrm{Ly}\alpha }& =\frac{f_{gm}{\eta }_{\mathrm{thick}}h{\nu }_{Ly\alpha }}{{\left(1+z\right)}^4}\frac{\Phi }{\pi }\\ & =5.1\times 10^{-17}{\left(\frac{1+z}{3.8}\right)}^{-4}\left(\frac{f_{gm}}{0.5}\right){\left(\frac{R}{60\mathrm{kpc}}\right)}^{-2}\\ & \times \left(\frac{L_{\nu ,LL}}{1.2\times 10^{30}\mathrm{erg}{\mathrm{s}}^{-1}\mathrm{H}{\mathrm{z}}^{-1}}\right)\mathrm{erg}{\mathrm{s}}^{-1}\mathrm{c}{\mathrm{m}}^{-2}\mathrm{arcse}{\mathrm{c}}^{-2}\end{array}$$

Full text

††slugcomment: Accepted to Astrophysical Journal

1. Introduction

In the unified model of AGN emission from Antonucci (1993), the ionizing radiation propagates anisotropically from an active black hole due to the absorption from a clumpy torus. For this reason, AGNs appear differently depending on their inclination with respect to our line of sight. This model thereby unifies observations of two main classes of AGN: Type I quasars, which have both broad and narrow emission lines, and Type II quasars, which have only narrow lines (e.g. Tristram et al., 2007; Nenkova, Sirocky, Ivezić, & Elitzur, 2008). Both types of AGN are characterized by extreme ionizing luminosities (up to $\sim 10^{47}$erg s*-1*), which should reveal the surrounding gas distribution out to large distances, if illuminated (Rees, 1988; Haiman & Rees, 2001; Cantalupo et al., 2005).

Following this idea, over the last few years, a few enormous Ly$\alpha$ nebulae (ELANe) with sizes up to 500 kpc were discovered around luminous Type I and a few Type II QSOs at $z=2$–3 (e.g. Cantalupo et al. 2014; Hennawi & Prochaska 2013; Arrigoni Battaia et al. 2018; Cai et al. 2017; Hennawi, Prochaska, Cantalupo, & Arrigoni-Battaia 2015). Moreover, deep narrow band (NB) imaging and Integral Field Spectroscopy have revealed the existence of Lyman-alpha line emitters with high equivalent width ($EW_{0}>240$Å) around $z=2$–3 quasars, also known as “dark galaxies” (e.g. Cantalupo et al. 2012; Marino et al. 2017). Overall, the simultaneous and detailed studies of the kinematics, metallicity, luminosity and abundance of the ELANe and dark galaxies provide us an indispensable opportunity to understand the interactions between the galaxies hosting AGN at high redshift and the gas that surrounds and fuels them (e.g. Prochaska & Hennawi 2009; Fumagalli et al. 2011).

With these motivations in mind, we have initiated a survey of $z>2$ quasars and previously known Ly$\alpha$ nebulae with the Keck Cosmic Web Imager (KCWI; Morrissey et al., 2012), a new blue-sensitive integral field spectrograph with wavelength coverage spanning from $\sim 3500-5600$Å. Its field-of-view (FoV), spectral resolution, and capability for precise sky subtraction are optimal for studying gas on scales of $\approx 200$ kpc. Here we report on deep KCWI observations of the complex system surrounding SMM J02399$-$0136 (Ivison et al., 1998), hereafter SMM J02399, the first galaxy selected at submm wavelengths (i.e. a SMG), gravitationally amplified by the massive foreground galaxy cluster Abell 370. Typical SMGs are dust-rich and highly obscured. In particular, the Ly$\alpha$ nebula surrounding SMM J02399 was quickly found to lie at z $\sim$ 2.8, using the 4 m Canada-France-Hawaii Telescope, in part because its Ly$\alpha$ luminosity is extremely bright ($\sim 10^{44}\,\rm erg\,s^{-1}$; Ivison et al. 1998). This source resides at the high luminosity end of the WISE detected LAB in Bridge et al. (2013), which are also Ly$\alpha$ nebulae powered by SMGs and have a Ly$\alpha$ luminosity of $10^{42}-10^{44}\,\rm erg\,s^{-1}$. It was the first SMG detected in CO, revealing it to contain a massive reservoir of molecular gas ($\approx 10^{11}$ M*⊙*, Frayer et al., 1998). This gas reservoir was later shown to cover a large volume and contain multiple, merging galaxies (Ivison et al., 2010), one of which is a dusty, broad-absorption-line (BAL) quasar (Villar-Martín et al., 1999a, b; Vernet & Cimatti, 2001; Frayer et al., 2018, cf. Genzel et al. 2003). The existence of a large reservoir of diffuse molecular gas has been recently inferred by observations of the CH+ cation (Falgarone et al. in prep.), following the detections against several other SMGs (Falgarone et al., 2017).

Inspired by these earlier results, we observed the system with KCWI tuned to the Ly$\alpha$ nebula and directed to cover it and its neighboring galaxies and quasar. To our surprise, the data reveal a new, luminous Ly$\alpha$ source – the primary focus of this paper, which is structured as follows. We describe our KCWI observation and data reduction process in §2. In §3, we present the detection of the $\gtrsim 100$ kpc Ly$\alpha$ nebula powered by this system and the discovery of a new dark cloud illuminated by the nearby BAL QSO. Finally, we discuss the Ly$\alpha$ nebula and the physical interpretations of the newly discovered dark galaxy in §4, and give a brief summary in §5. Throughout this paper, we assume a flat cosmological model with $\Omega_{\Lambda}=0.7,\Omega_{m}=0.3$ and $H_{0}=70\,\rm km\,s^{-1}\,\rm Mpc^{-1}$ which implies the physical scale is $\approx 7.85\,\rm kpc$ per arcsec at the redshift of SMM J02399$-$0136. We follow the naming convention presented in Ivison et al. (2010): L1 marks the BAL quasar; L2 is a low-surface-brightness companion to the East of L1; L2SW is a massive, extremely luminous and dust-obscured starburst galaxy, East-South-East of L1 and South-West of L2; L1N is a faint, compact northern component, visible in post-SM2 HST imaging.

2. Observations and Data Reduction

2.1. KCWI observations

The KCWI observations of SMM J02399 were carried out on UT 2017 October 21 (seeing $\sim$ 1.5”) using KCWI on the Keck II telescope of the W. M. Keck Observatory in Hawaii. We used the BM1 grating and the medium slicer (slice width $\sim 0.7^{\prime\prime}$) which yields an IFU datacube with FoV of $20^{\prime\prime}\times 16.8^{\prime\prime}$ (pixel scale of $0.3^{\prime\prime}\times 0.7^{\prime\prime}$) centered on J023951.88$-$013558.0, the quasar optical position given by Ivison et al. (1998). The grating was tilted to give a central wavelength of 4620Å and provides a spectral resolution, $R\approx 4000$. The good wavelength coverage is $\sim$ 4230 - 5010Å . The total on-source exposure time is four 10-min exposures, each dithered by $\sim 0.6^{\prime\prime}$. For sky observations, we used an ‘offset-target-field’ to construct the sky datacube. The offset-target has a different redshift and is located $\sim 2$ degrees from SMM J02399. It is a compact point-source at the Ly$\alpha$ wavelength of SMM J02399.

To convert the spectral images and calibration frames (arcs, flats, bias) to a calibrated datacube, we used the IDL-based KCWI data reduction pipeline***Available at https://github.com/kcwidev/kderp/releases/tag/v0.6.0. Basic CCD reduction is performed on each science frame to obtain a bias-subtracted, cosmic-ray-cleaned and gain-corrected image. The continuum flat images are employed for CCD response corrections and pixel-to-pixel variations. We used a continuum-bar image and an arc image (ThAr) to define the geometric transformations and wavelength calibration, generating a rectified object data cube (see the pipeline documents†††https://github.com/kcwidev/kderp/blob/master/AAAREADME). Twilight flats were used for slice-to-slice flux correction, and the data was corrected for atmospheric refraction. Each object and sky frame was flux calibrated with the standard star, Hiltner 600. For the sky frame, we first masked the point-source of the offset-target, and then estimated the sky level at each wavelength channel by the median of unmasked sky pixels. Then, for each channel, we subtract the sky from SMM J02399. For each exposure we found the QSO centroid to measure the offsets between exposures, and then performed a weighted mean with inverse-square variance weighting to construct the final data cube.

2.2. Ancillary Data

CO(3–2) observations of SMM J02399 were obtained by Genzel et al. (2003) with a synthesized beam of $5.2\times 2.4$ arcsec2. We use these data, kindly provided by L. Tacconi, to compare the cool molecular gas emission with that traced by Ly$\alpha$. SMM J02399 is strongly magnified by Abell 370, which was imaged as part of the HST Frontier field survey (HST-14038, PI: J. Lotz). It was observed with three broadband filters: ACS/WFC F435W, F606W and F814W, for total exposure times of 12, 6 and 29 hr, respectively. We retrieved the publicly available reduced data‡‡‡ http://archive.stsci.edu/ and produced an average image, weighted by the exposure time. The F435W filter covers the Ly$\alpha$ emission of SMM J02399 and thereby provides an independent estimate of its spatial extent.

3. Results

3.1. Deep Narrow-band Imaging at Ly$\alpha$

Fig. 1 shows our continuum-subtracted, pseudo-narrow band image, spanning $\approx 30$Å with central wavelength $\lambda_{\rm NB}=\lambda_{\rm Ly\alpha}(1+z_{L1})\approx 4025$Å, and Gaussian smoothed spatially by $2^{\prime\prime}$. The $3\sigma$ depth flux density of this image in a 1.7 arcsec2 aperture §§§In this paper, we used two kinds of apertures. One ($\sim 1.7\rm\,arcsec^{2};1.4\arcsec\times 1.2\arcsec$) matches the seeing of the observations and it is used to subtract spectral features and measure the continuum in the HST image. The second ($\sim 6\rm\,arcsec^{2}$) is defined by the $5\sigma$ contour at the L3 location (see Fig. 1) and is used to measure its total/average properties. This second aperture is used to calculate the observed (Section 3.1 para.2) and inferred results (in Section 4.1). is $4\times 10^{-18}\rm erg\,s^{-1}cm^{-2}$Å*-1*, corresponding to a surface brightness of $1.6\times 10^{-17}\rm erg\,s^{-1}cm^{-2}arcsec^{-2}$. In the following discussion, we refer only to observed quantities (i.e. without lensing corrections for magnification by Abell 370 ¶¶¶The magnification factor of L1, L2, L3 are almost the same ($\sim$2.4$\times$ for each). (Ivison et al., 2010).) The peak surface brightness in the region encompassing L1 and L2 is $\approx 1.2\times 10^{-16}\rm\,erg\,s^{-1}\,$ $\rm cm^{-2}\,arcsec^{-2}$. To a 3$\sigma$ surface brightness contour, this nebula has a lensing-uncorrected size of $\gtrsim 17$ arcsec (i.e. $\gtrsim 140$ physical kpc) and has a total Ly$\alpha$ luminosity of $2.5\times 10^{44}\,\rm erg\,s^{-1}$ .

Fig. 1 also reveals a bright, previously unreported source approximately $7.5^{\prime\prime}$ to the SW of L1. We refer to this source as L3 and measure a Ly$\alpha$ luminosity of $1.3\times 10^{43}\rm\,erg\,s^{-1}$ within an aperture of 6 arcsec2. The peak surface brightness of L3 is $3.0\times 10^{-17}\rm\,erg\,s^{-1}\,cm^{-2}\,arcsec^{-2}$, similar to Ly$\alpha$ blobs surveyed by Matsuda et al. (2004, 2011), as well as the extended Ly$\alpha$ nebulae detected around $z\sim 2$ high-redshift radio galaxies and radio-loud quasars (e.g. Villar-Martín et al. 2007; Heckman et al. 1991).

Fig. 2a shows the HST imaging around L3, comprised of ACS/WFC F435W, F606W and F814W deep images and contours for the CO and Ly$\alpha$ emission. CO(1-0) and CO(3-2) observations suggest there is a large gas reservoir in this system, and the peak location of which coincides with the Ly$\alpha$ emission at L1. We have also confirmed that there is no continuum (stellar) counterparts corresponding to L3 in the SDSS, WISE and 2MASS catalogs ∥∥∥SDSS catalog: Alam et al. (2015); Abazajian et al. (2009)

WISE catalog: Cutri & et al. (2012, 2013)

2MASS catalog: Cutri et al. (2003); Skrutskie et al. (2003) , nor in the deeper surveys of PanSTARRS, DES, DECaLS. ******The Pan-STARRS1 Surveys: Chambers et al. (2016)

DES DR1: Abbott et al. (2018)

DECaLS: http://legacysurvey.org/dr7/ Examining Fig. 2b, we derive a $5\sigma$ upper limit to the continuum $4.0\pm 0.8\times 10^{-20}\rm\;erg\,s^{-1}\,cm^{-2}$ Å*-1* from the HST F435W image in a 1.7 arcsec2 aperture. The Ly$\alpha$ rest-frame equivalent width is therefore greater than 312Å, which follows the scenario of a dark galaxy††††††The definition of ‘dark galaxy’ is defined as a Ly$\alpha$ emitter with rest-frame EW greater than 240Å , following the definition of Cantalupo et al. (2012). illuminated by a luminous AGN (Cantalupo et al., 2012).

Here we also estimated the $5\sigma$ continuum upper limits from both F606W and F814W imaging in an aperture of 1.7 arcsec2, obtaining $1.9\times 10^{-20}\rm\;erg\,s^{-1}\,cm^{-2}$ Å*-1* and $9.7\times 10^{-21}\rm\;erg\,s^{-1}\,cm^{-2}$ Å*-1*, respectively. Then we calculated that the $5\sigma$ upper limit of the rest-frame UV SFR is $0.3\,\rm yr^{-1}$ (Kennicutt & Evans, 2012). This SFR is not high enough to give rise to the bright Lyman-alpha emission.

3.2. Ly$\alpha$ Kinematics

Our KCWI datacube enables a study of Ly$\alpha$ emission throughout the complex system surrounding SMM J02399. Fig. 3 shows the flux-weighted centroid velocity $v_{\rm Ly\alpha}$ relative to $z_{\rm L1}$ and the velocity dispersion $\sigma_{v}$. A gradient in $v_{\rm Ly\alpha}$ is evident as one traverses from L2 to L3 through the nebulae ranging from $-300$ to $+800\,\rm km\,s^{-1}$. In contrast, the velocity dispersion is roughly constant at $\sigma_{v}\approx 500\rm km\,s^{-1}$.

We extract 1D spectra from the final datacube for L1, L2, L1N, L2SW and L3 through a aperture of 1.7 arcsec2 (Fig. 2c and Fig. 3c). The Ly$\alpha$ emission of L3 is well-modeled by a single Gaussian with central wavelength $\lambda_{\rm L3}^{\rm Ly\alpha}=4634.9\pm 0.1$Å corresponding to $z_{\rm L3}=2.81259\pm 0.00008$. The Gaussian model also gives a a velocity dispersion of $\sigma_{\rm L3}^{\rm Ly\alpha}=150\pm 6\rm km\,s^{-1}$ (or FWHML3 = $353\pm 15\rm km\,s^{-1}$).

In Fig. 3c, except for L3, the Ly$\alpha$ emission from the nebula shows both broad and narrow components, confirming the previous long-slit spectra (see also Vernet & Cimatti, 2001; Ivison et al., 2010). The nebular emission cannot be well described by a single Gaussian. The figure also emphasizes that L3 is kinematically distinct from the larger nebula; its centroid is offset by several hundred $\rm km\,s^{-1}$ and it exhibits no broad component.

The Ly$\alpha$ emission spectrum of the nebula also shows evidence for two weak absorption features on both the red and blue sides of the primary emission. Intriguingly, line observations of the CH+ cation, a specific tracer of turbulent dissipation (Falgarone et al., 2017), reveal broad emission lines ascribed to shocks and a broad absorption line (FWHM = 600 $\rm km\,s^{-1}$) in the direction of L2SW (Falgarone et al., in prep.). Ly$\alpha$ and CH+ emission of this system both show broad line-emission ascribed to shocks and narrower lines that may imply that L1 and L2 are encompassed by a large and massive HI cloud.

4. Discussion

4.1. Illumination of the dark cloud (L3)

In the previous section, we reported on the discovery of L3, a luminous Ly$\alpha$ emitter with an extremely high Ly$\alpha$ equivalent width ($W_{\rm Ly\alpha}>312$Å) and a FWHM of $\approx 350\rm km\,s^{-1}$. Cantalupo et al. (2012) reported on a sample of sources with $W_{\rm Ly\alpha}>240$Å surrounding the ultra-luminous quasar HE0109$-$3518 which lack detectable continuum counterparts in deep broad-band imaging. They termed these sources “dark galaxies” and argued they were illuminated by the quasar as no star-formation could power such high $W_{\rm Ly\alpha}$, i.e. the observed Ly$\alpha$ is fluorescent radiation powered by the incident, ionizing photons. The properties of L3 are consistent with this dark-galaxy scenario, except that the observed UV emission from L1 is orders of magnitude lower than HE0109$-$3518. Therefore, we propose a scenario wherein the dust-obscured BAL quasar is emitting UV radiation anisotropically, with bright UV emission emitted transverse to our line of sight (Fig. 4).

We may test the Ly$\alpha$ fluorescence hypothesis by comparing the observed Ly$\alpha$ flux with that predicted for optically thick gas illuminated by the unobscured ionizing radiation of L1. To estimate the unobscured luminosity of L1, consider the following analysis. L1 is detected by the WISE satellite with a $W2$ magnitude at $\approx 4.6\mu$m of $15.2\pm 0.1$ mag from ALLWISE source catalog. Adopting the Type-I QSO template from Richards et al. (2006) and scaling to this $W2$ measurement we may estimate $F_{\nu,LL}$, with $\nu_{LL}$ the frequency at the HI Lyman limit. We estimate $L_{\nu,LL}$ of L1 to be $\log_{10}(L_{\nu,LL}/\rm erg\,s^{-1}\,Hz^{-1})=30.07$ and note that this model also reproduces the observed $W1$ flux. In contrast, we estimate the obscured UV luminosity at $\nu_{LL}$ along our sightline to be only $\log_{10}(L_{\nu,LL}/\rm erg\,s^{-1}\,Hz^{-1})=27.74$, using the Type-II QSO template from Polletta et al. (2007) and the QSO ultraviolet template from Lusso et al. (2015).

We further assume: (1) the redshift offset between L1 and L3 is due to peculiar motions rather than the Hubble flow, such that the physical separation between the two is comparable to the projected offset; (2) L3 is an optically thick gas cloud with radius of $R_{\rm L3}\approx 10\,\rm kpc$, estimated from the NB image (Fig. 1); (3) the optically thick cloud converts a fraction $\eta_{\rm thick}=0.66$ of the ionizing photons from the quasar to Ly$\alpha$ photons emitted at a uniform brightness (Gould & Weinberg, 1996; Hennawi & Prochaska, 2013); (4) L3 is ‘half-moon’ illuminated with a geometric reduction factor of $f_{\rm gm}=0.5$ consistent with radiative transfer simulations (Cantalupo et al., 2005; Kollmeier et al., 2010).

From these assertions, we derive the Ly$\alpha$ surface brightness (Hennawi & Prochaska, 2013),

[TABLE]

where $\Phi$ (phot s*-1* cm*-2* ) is the ionizing photon number flux, and $R$ is the distance of L3 from L1. Integrating this surface brightness over the observed size of L3, we estimate $f_{\rm L3}^{Ly\alpha}=3.2\times 10^{-16}\,\rm erg\,s^{-1}\,cm^{-2}$. The observed flux of Ly$\alpha$ from L3 is $f_{\rm L3,obs}^{Ly\alpha}=1.8\times 10^{-16}\,\rm erg\,s^{-1}\,cm^{-2}$, corresponding to surface brightness of $SB_{\rm L3,obs}^{Ly\alpha}=3.0\times 10^{-17}\rm erg\,s^{-1}\,cm^{-2}\,arcsec^{-2}$. This derived value is consistent with the observed flux of Ly$\alpha$ from L3.

Adopting fluorescence as the origin of the Ly$\alpha$ emission, we further estimate a cool gas mass of $M_{\rm cool}\sim 1.7\times 10^{10}M_{\odot}$ using Equation 8 in Cantalupo et al. (2012), assuming a clumping factor C = 1 and T = 2$\times 10^{4}$K. We have also used CLOUDY ionization modeling (Ferland, 1996) with the same assumptions above, and found that the best parameter combinations to match LLyα,L3 is ($\rm N_{H},Z,n_{H}$)$\sim$ ($10^{22}\rm cm^{-2},0.01Z_{\odot},0.1cm^{-3}$).Future metal line observations may provide more precise limits. Assumed H density of L3 is $\sim$ 0.1 cm*-3* according to our best CLOUDY model, the collapse timescale tff is $\sim$ 1.6 $\times$ 108 years. Considered the cosmic correction, tff from earth is $\sim$ 6.1 $\times$ 108 years, $\sim$ 4.5% of the universe age. And considered ‘dark matter’, tff will be shorter. The integrated opacity due to Thompson scattering of the ionized cloud is negligible given its very low cross-section ($0.665\times 10^{-24}\rm cm^{2}$) even given our relatively large total electron column density of $\rm N_{e}\sim 10^{22}\rm cm^{-2}$. Even a very luminous quasar will give an insignificant flux compared with what we observed.

Numerical studies indicate that the dark galaxy phase corresponds to gas-rich galaxies prior to efficient star formation (e.g. Cantalupo et al. 2005, 2012; Kollmeier et al. 2010; Marino et al. 2017; Francis & Bland-Hawthorn 2004; Hennawi et al. 2009). A few dedicated dark galaxy surveys using deep narrow-band imaging and the Multi Unit Spectroscopic Explorer (MUSE) (e.g. Cantalupo et al. 2012,Marino et al. 2017) found that the dark clouds reside in the mass range M${}_{\rm gas}\sim 10^{\rm 8-10}$ M*⊙, and in the Ly$\alpha$ luminosity range of $10^{\rm 41-43}$ erg s-1*. The dark cloud discovered here is consistent with these properties.

Note that there are three galaxies within $5\arcsec\times 5\arcsec$ from L3 in the deep HST images, which we indicate in panel b of Fig. 2 as G1, G2, and G3. The broadband colors of G1 and G3 are consistent with Lyman break galaxies at $z=3$, though with large uncertainties given the faintness of these galaxies. Even if we allow that G1 and G3 are partially powering L3 and adopt a UV flux from those sources, then the EWLyα estimated for L3 decreases to 209 - 270 Å. Nevertheless, the Ly$\alpha$ kinematics of L3 has a FWHM of $353\pm 15\rm km\,s^{-1}$ which is relatively quiescent. Also, the emission peak has a large spatial offset($\sim$2.0”) from both galaxies. These evidences support that L3 is not likely to be powered by either G1 or G3.

Indeed, we cannot rule out a scenario where L3 is powered by gravitational infall, but our current data supports that L3 is more likely to be externally illuminated. Future deep polarimetric observations may rule out such scenarios.

4.2. Ly$\alpha$ tracing anisotropic QSO emission

The dimensions, luminosity, and kinematics of the Ly$\alpha$ nebula surrounding SMM J02399 are reminiscent of the MAMMOTH-1 nebula which extends to 450 kpc (Cai et al., 2017) and is also powered by a Type-II AGN (Arrigoni Battaia et al., 2018). Cai et al. (2017) argued that the powering mechanism of MAMMOTH-1 could be a combination of photoionization and shocks due to an AGN outflow (e.g. Harrison et al. 2014). SMM J02399 is another nebula hosting a highly obscured system, and may have a similar powering mechanism. For the full nebula around SMM J02399  we measure a total Ly$\alpha$ luminosity of $2.5\times 10^{44}$ erg s*-1* (uncorrected for lensing). The AGN bolometric luminosities is $2.5\times 10^{46}$ erg s*-1* estimated by the UV luminosities (1450$\mathring{A}$) with $L_{bol}=4.2\nu L_{\nu},1450\mathring{A}$ (Runnoe et al., 2012). Considered the lensing magnification, the BAL QSO has a similar bolometric luminosity ($\sim 10^{46}$ erg s*-1*) with a normal QSO at z=3. To compare with MAMMOTH-1, we subtracted the Ly$\alpha$ PSF. The PSF is constructed using a full-width-half-maximum that is equal to the seeing of $1.5^{\prime\prime}$. The PSF amplitude is determined from the Ly$\alpha$ luminosity of L1 (in an aperture of $\sim 1.7$ arcsec2) convolved with a Moffat kernel. We find that the central PSF luminosity constitutes only 4.4% of the entire nebula, which is similar to the MAMMOTH-1 measurement of 4%. In the scenario involving collimation, previous studies have already shown that L1 may have a jet/outflow betrayed by the radio morphology, and L2 is likely explained as the shock-excited region or a reflection nebula (Ivison et al. 2010; Frayer et al. 2018). High-resolution CO observations with ALMA and JVLA, and red-sensitive observations of extended and broad metal-line emission (e.g. C${}_{\textrm{\scriptsize IV}}$) (Cai et al., 2017), could provide decisive evidence for the outflow scenario.

We note that this Ly$\alpha$ nebula is not the only one which contains multiple embedded galaxies and lacks a clear continuum sources at the peak of the Ly$\alpha$ extended emission. Currently, $\sim 10$ Ly$\alpha$ nebulae (also known as Ly$\alpha$ blobs, LAB) are reported as powered by obscured sources. Examples of these are the LABs in Prescott et al. (2012), Bridge et al. (2013), and a few LAB in the SSA22 field (e.g. Matsuda et al. 2004, 2007). With an extent of $>$140 kpc and a luminosity of $2.5\times 10^{46}$ erg s*-1*, our nebula is one of the largest and most extended among these systems, and intriguingly hosts the most luminous powering source. Regarding the previous samples of Ly$\alpha$ nebulae and LABs, the SMM J02399 nebula is one of the most luminous discovered to date, and also has multiple components. Also, this is the only system containing a proto-galaxy (a dark galaxy by the definition of Cantalupo et al. 2012), indicating a highly asymmetric ionization.

Fig. 4 illustrates the anisotropic emission of ionizing radiation from L1 that we envisage with L1 unobscured in the direction of L3. The UV photons ionize the outer layers of the cool gas cloud to produce Ly$\alpha$ emission. Along our line of sight it is a typical dust obscured BAL quasar with substantial foreground absorption, supporting unified models of AGN. This confirms the primary conclusion from a series of papers that have studied the anisotropic clustering of optically thick gas transverse to and along the line of sight to quasars (Hennawi & Prochaska, 2007, 2013, e.g.). Based on the same assumption in §4.1, here we estimate a lower limit of the open angle of L1 conservatively as $\Omega/4\pi>0.007$ ($f_{obscured}=1-\Omega/4\pi<0.993$), with the half-angle $\theta>9.5^{\circ}$. The opening angel is estimated using the linear size of L3 and the distance between L1 and L3, assuming L1 and L3 is at the same redshift, and that the redshift offset is due to the kinematics not Hubble flow. Further evidence supporting the anisotropic emission hypothesis is the BAL nature of L1, which has been constrained by the moderate continuum polarization with VLT/FORS1 (Vernet & Cimatti, 2001). BAL QSOs are expected to radiate as Type-I QSOs only through patches free of dust and dense outflowing gas (e.g., Ogle et al. 1999). On the other hand, the SMG (L2SW) close to L1 also can be a plausible source to illuminate L3, which dust cocoon may not necessarily be homogenous.

5. Concluding Remarks

In this paper, we present the Keck/KCWI IFU observations of a Ly$\alpha$ blob powered by SMM J02399 at z $\sim$ 2.8. With KCWI, we discover a dark cloud that we argue is illuminated by the dust-obscured QSO of the system. This implies strong, anisotropic UV radiation from the QSO, which was also suggested by previous polarimetry observations. The future red/blue sensitive IFS, Keck Cosmic Reionization Mapper (KCRM)/KCWI can further reveal the properties and kinematics of other lines, such as Ly$\beta$, HeII, CIV, to further reveal the nature of this and similar systems. We are also pursuing a KCWI survey of Type-II AGN to study the population of Ly$\alpha$ nebulae in a set of fully obscured sources.

Acknowledgement: This work was supported by National Key Program for Science and Technology Research and Development (grant 2016YFA0400703) and the National Science Foundation of China (11721303). We are thankful for the supports from the National Science Foundation of China (NSFC) grants No.11373008, 11533001. QL gratefully acknowledge financial support from China Scholarship Council. ZC acknowledge the supports provided by NASA through the Hubble Fellowship grant HST-HF2-51370 awarded by the Space Telescope Science Institute. , which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. EF gratefully acknowledges funding from the European Research Council, under the European Community’s Seventh framework Programme, through the Advanced Grant MIST (FP7/2017-2022, No 787813). The data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation.