Discovery of a Protocluster Core Associated with an Enormous Lya Nebula at z = 2.3

The ELAN MAMMOTH-1 was observed with NOEMA (ID: S18CW) centered on (α₂₀₀₀, δ₂₀₀₀) = (14^h41^m2447, + 40° 03'0967) between 2018 November to 2019 January in C configuration with 10 antennas. The total observing time is 8 hr and the on-source time is 4.6 hr. We use the 3 mm receiver and the correlator PolyFix in dual polarization mode, tuning one of the 3.9 GHz basebands on the observed frequency 104.250 GHz of the redshifted CO(3 − 2) transition.

We use the quasars 1505+428, 1504+377, and J1438+371 as phase and amplitude calibrators. The RF calibrators were 2013+370 and 3C273, the flux calibrator was MWC349. In the 3 mm band (band 1 of NOEMA), the absolute flux calibration is accurate within 10%. The calibrated visibility data were imaged with the software package MAPPING (part of GILDAS), using natural baseline weighting and the Hogbom cleaning algorithm. The final synthesized beam sizes are 23 × 16. To further check if there are missing CO intensities in a more extended area, we tapered and re-weighted the visibilities to a lower angular resolution of 32 × 32. The field of view (primary beam) for NOEMA observations at ν_obs = 100 GHz are 50''. Then we made the primary beam correction. The maximum recoverable scale at the observed frame 100 GHz is roughly 41''. The 1σ rms sensitivity of the natural-weighted image cube is 0.3 mJy beam⁻¹ per 45 km s⁻¹ channel, while the rms of the tapered-image cube is 0.4 mJy beam⁻¹ per 45 km s⁻¹ channel. We mainly focus on the small beam size (23 × 16) in the following discussion.

We imaged the observed frame 250 GHz continuum emission of the MAMMOTH-1 field with NOEMA in C configuration (ID: W19CX). The observations were carried out on 2019 December 29 and 31, using the PolyFiX correlator with the full available continuum bandwidth of 15.5 GHz in dual polarization. The flux scale was calibrated on MWC349 and the phase was checked with the calibrator 1505+428 close to our target. The total observing time is 8 hr with 5 hr on-source. The primary beam for NOEMA observations at ν_obs = 250 GHz (ν_rest ∼ 828.2 GHz) are 21''. The FWHM synthesized beam size is 087 × 065 and the position angle is 49°. The final continuum 1σ rms sensitivity in the central region of the cleaned image is 0.04 mJy beam⁻¹. Primary beam correction was applied to the final flux measurements described in Section 3.4.

The rest-frame optical deep image of the MAMMOTH-I field was obtained using the Hubble Space Telescope (HST) Wide Field Camera 3 (WFC3) with the F160W filter (ID: 14760). The observation was carried out in Cycle 24 with an exposure time of ∼2665 s; we use nine HST/WFC3 pointings to cover this central region in nine orbits to measure the detailed rest-frame optical morphology. The data reduction is conducted using Multidrizzle (Koekemoer et al. 2003) and the detailed procedures follow the descriptions in Cai et al. (2016). To optimize the output data quality, we choose a final output pixel scale of 006 instead of the initial pixel scale 013 and final pixfac parameter 0.7 (shrinking pixel area) after different trials of combinations of parameters.

We also observed the Lyα line of MAMMOTH-I with the Keck Cosmic Web Imager (KCWI) on the Keck II telescope of the W. M. Keck Observatory in Hawaii in 2018 May (seeing ∼15). The on-source exposure time is 1 hr. We used the Blue Medium Grating with the Large Slicer (slice width ∼35), resulting in a spectral resolution of 2000 and field of view of 33'' × 20'', centered on Source B, the quasar optical position given by Cai et al. (2017b). To convert the spectral images and calibration frames (arcs, flats, bias) to a calibrated data cube, we used the IDL-based KCWI data reduction pipeline. ¹⁴ 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 ¹⁵ ). Twilight flats were used for slice-to-slice flux correction, and the data were corrected for atmospheric refraction. Each object and sky frame was flux calibrated with the standard star BD28d4211. 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.

The CO(1 − 0) line data of this field was observed using the Very Large Array (VLA) in the most compact D-configuration and published in Emonts et al. (2019), which is included in the analysis of this paper. The exposure time is 14 hr on-source, with an rms noise level of 0.057 mJy beam⁻¹ channel⁻¹, for a channel width of 30 km s⁻¹. The field of view (primary beam) for VLA observations at ν_obs ∼ 34.81 GHz (ν_rest = 115.27 GHz) is 13. The synthesized beam is 26 × 23. We use these near-infrared and radio data to compare with our CO(3 − 2) observations in order to study the molecular gas distribution and kinematics and thus provide a link between the ELAN and the stellar build-up of this system.

At the center of the ELAN MAMMOTH-1, the CO(3 − 2) line flux of Source B (G2) is 0.237 ± 0.051 Jy km s⁻¹ (Figure 2). We fit the FWHM _CO(3−2) to be 370 ± 90 km s⁻¹, which is a typical CO line width found in samples of SFGs and quasars (e.g., Carilli & Walter 2013; Ueda et al. 2014). However, the line width of the CO(1 − 0) detected in Emonts et al. (2019) is much narrower, only 85 km s⁻¹. We re-analyze the previous CO(1 − 0) data, fitting it with a double Gaussian profile with no constraint applied. After subtracting the narrow Gaussian component, the line width of the new broad CO(1 − 0) component is similar to that of CO(3 − 2) . More details are presented in Section 4.2. The redshift derived from CO(3 − 2) line is z_CO = 2.3123 ± 0.0006, while the redshift derived from Lyα is z_{Ly α} = 2.329 ± 0.013 (Cai et al. 2017b). This shift may be due to resonant scattering of Lyα. The 2D Gaussian fit for the velocity-integrated image suggests line emission from an unresolved source.

Figure 2 also reveals two bright sources in CO(3 − 2) in the nearby region of Source B, labeled as Sources A and C. They were detected in CO(1 − 0) line emission down to a sensitivity level of 0.057 mJy beam⁻¹ channel⁻¹ for a channel width of 30 km s⁻¹ (Emonts et al. 2019). Adopting the naming convention from Emonts et al. (2019), Source A (G1) is the brightest detection in both CO(3 − 2) and CO(1 − 0) lines in these systems. However, it is not located inside the ELAN MAMMOTH-I (6'' away from Source B) and has its own Lyα emission (Q. Li 2021, in preparation). It is also unresolved. The line flux I_CO(3−2) is 0.298 ± 0.044 Jy km s⁻¹. Its FWHM _CO(3−2) of ∼180km s⁻¹ is similar to that of CO(1 − 0) (∼170km s⁻¹). Source C (G6) is a faint detection roughly 9'' west of Source B with a line flux of 0.245 ± 0.056 Jy km s⁻¹. The line width is FWHM _CO(3−2) = 280 ± 70 km s⁻¹, similar to the CO(1 − 0) of ∼230 km s⁻¹. The velocity offset relative to Source B is ∼510 km s⁻¹.

The other three sources (G3, G4, G5) are all >3σ detections in CO(3 − 2) and have HST optical counterparts. However, they are not detected in CO(1 − 0) . By applying the line ratio of r_3,1 = 0.52 for SFGs (Kirkpatrick et al. 2019), the 1σ rms of the VLA CO luminosity measurement of ${L}_{\mathrm{CO}(1-0)}^{{\prime} }\,\sim 0.3\times {10}^{10}$ K km s⁻¹ pc² would correspond to ${L}_{\mathrm{CO}(3-2)}^{{\prime} }\sim 1.6\times {10}^{9}$ K km s⁻¹ pc² for CO(3 − 2). The sensitivity of the VLA observations is insufficient to detect the CO(1 − 0) line from these three objects.

Now we compare CO, Lyα, and optical counterparts of this system. In Figure 3 we show the map of the CO(3 − 2) line emission from NOEMA (blue contours), CO(1 − 0) from VLA (red contours), and the Lyα from KCWI (black contours, Q. Li et al. 2021, in preparation) overlaid on the HST/WFC3 FW160W image (Z. Cai et al. 2021, in preparation). The HST image shows four optical counterparts around Source B, see Figure 3(a). The CO(3 − 2) and CO(1 − 0) peaks of Source B (G2) are almost consistent with the central HST optical source. The peak of Lyα is located at another optical counterpart. The CO peak and Lyα peak have an offset of ∼1.5'' (∼13kpc). Taking into account the accuracy of the radio interferometric positions ¹⁶ from the VLA and NOEMA, and the seeing of the Lyα observations of ∼ 10, the offset could be real.

Zoom In Zoom Out Reset image size

Source A (G1) has an optical counterpart in the HST image. It was also detected Lyα emission, 7'' away from the Lyα peak of the ELAN MAMMOTH-I. The CO(3 − 2) peak of Source A (G1) coincides with CO(1 − 0) . Emonts et al. (2019) reported that the CO(1 − 0) emission of Source A and Source B have extended features of ∼25 kpc and 30 kpc, respectively. However, our CO(3 − 2) observations suggest that they are unresolved, which implies that CO(3 − 2) is compact and emitted from the star-forming region within the galaxy.

Source C (G6) does not have any Lyα counterpart but shows an HST optical detection. Source C also has CO(1 − 0) emission in VLA observations (Emonts et al. 2019). The CO(1 − 0) and CO(3 − 2) lines also show that it is a point source. Its CO(3 − 2) line flux is weaker than for A and B.

The maximum recoverable scale is ∼41'' for our CO(3 − 2) observations, corresponding to the physical scale of ∼340 kpc at z ∼ 2.3. Here we tapered the beam to recover the additional emission in CGM with a beam size of 32 × 32, corresponding to a physical scale of ∼27 kpc. We assumed the same average r_3,1 = 0.6 and the same CO line width in the CGM; the sensitivity of CO(1 − 0) in VLA data is 0.057 mJy beam⁻¹ per 30 km s⁻¹ channel and the derived sensitivity of CO(3 − 2) is 0.5 mJy beam⁻¹ per 45 km s⁻¹ channel. Our NOEMA data sensitive is 0.4 mJy beam⁻¹ per 45 km s⁻¹ channel, enough to detected the CO(1 − 0) extended features in CO(3 − 2), but at this sensitivity in our NOEMA data, we did not detect diffuse CO(3 − 2) emission. This indicates that there is no diffuse CO(3 − 2) line emission across the nebula. It is different from the diffuse CO(1 − 0) emission which appears extended across a region of ∼30 kpc.

To constrain the nature of the sources of this system, here we calculate the CO(3 − 2) line luminosity of each source as ${L}_{\mathrm{CO}}^{{\prime} }=3.25\times {10}^{7}\times {I}_{\mathrm{CO}}{D}_{L}^{2}{(1+z)}^{-3}{\nu }_{\mathrm{obs}}^{-2}\ {\rm{K}}\ \mathrm{km}\ {{\rm{s}}}^{-1}\ {\mathrm{pc}}^{2}$ (Solomon et al. 1992) where I_CO is the integrated line flux in Jy km s⁻¹, D_L is the luminosity distance in Mpc and ν_obs is the observing frame CO(3 − 2) line frequency. The derived line luminosities are in the range of (2.1−7.1) × 10⁹ K km s⁻¹ pc². The CO(3 − 2)/CO(1 − 0) luminosity ratios (r_3,1) of Sources A, B, and C are 0.59 ± 0.17, 0.61 ± 0.17, and 0.54 ± 0.25, respectively. The median line ratios r_3,1 for AGN- and star-formation-dominated galaxies are 0.92 ± 0.44 and 0.52 ± 0.17 (Kirkpatrick et al. 2019). Carilli & Walter (2013) suggested the average line ratio r_3,1 of quasars is 0.97. Source B is lower than this value and lies toward the low end of the expected AGN-dominated range.

Low r_3,1 in ELANe have also been reported previously, e.g., Genzel et al. (2003) observed SMM J02399, a BAL quasar in a >140 pkpc ELAN at z ∼ 2.8 (Ivison et al. 1998; Li et al. 2019). Its r_3,1 is 0.48 ± 0.13. By contrast, the central radio galaxy MRC 1138–262 in the ELAN Spiderweb Galaxy (Lyα extended ∼200 kpc, Miley et al. 2006) shows a very high global r_4,1(${L}_{\mathrm{CO}(4-3)}^{{\prime} }/{L}_{\mathrm{CO}(1-0)}^{{\prime} })$ of 1.00 ± 0.28 (Emonts et al. 2018). Its CGM has r_4,1 = 0.45 ± 0.17, similar to SFGs.

Regardless of the presence of AGNs, the ratios r_3,1 in different galaxies vary greatly from 0.4 to 0.9. At high z, the literature, such as Harris et al. (2010; z = 2.5 − 2.9 SMGs) and Aravena et al. (2010; z ∼ 1.5 normal SFGs), reports r_3,1 ∼ 0.6. Sharon et al. (2016) report that there is no statistically significant difference in the mean line ratio (r_3,1 = 0.90 ± 0.40 for both populations combined) in z ∼ 2 galaxies including both AGNs and SMGs. Furthermore, Riechers et al. (2020) report r_3,1 = 0.84 ± 0.26 in z = 2 − 3 main-sequence galaxies from the ASPECS surveys. The higher-J CO transition observations, at least J_up ≥ 4, can further reveal the excited gas and especially excitation from AGNs.

Previous continuum observation of MAMMOTH-1 using SCUBA-2 at the observed frame 350 GHz revealed bright dust-continuum emission with a flux of S_350GHz = 4.6 ± 0.9 mJy (Arrigoni Battaia et al. 2018b). However, the SCUBA-2 beam size of 15'' is too large to constrain the continuum emission from individual galaxies. All the CO-detected objects in this area could contribute to the SCUBA-2 flux. Three continuum sources are detected at >4σ in this field from our NOEMA 250 GHz (1.2mm at observed frame) map at 087 × 065 resolution. They are counterparts of the CO detections described above (Sources A, B, and C). We used 2D Gaussian fitting to measure the continuum and detect a continuum flux of 0.74 ± 0.15 mJy from Source A. The continuum emission is marginally resolved along the major axis with a deconvolved source size of (1.25 ± 0.32)'' × (0.57 ± 0.26)''. This is the brightest continuum detection in the MAMMOTH-1 field. We detect a 4σ continuum source at the position of Source B. The continuum emission is unresolved and we adopt the peak surface brightness as the total continuum flux which is 0.14 ± 0.03 mJy. We also detect continuum emission from Source C with a flux of 0.18 ± 0.05 mJy. The source is unresolved as well. The continuum detections are summarized in Table 1.

The NOEMA continuum map at 250 GHz allows us to determine the FIR luminosity and the star-formation rate. We assume a modified blackbody for optically thin thermal dust emission, with the dust temperature of 42K for Source B (AGN), and 35 K for other galaxies. We adopt an emissivity index of β = 1.6 (Beelen et al. 2006). For the non-detections, we adopt a 3σ upper limit. We estimate FIR luminosity by integrating the modified blackbody function in 8–1000 μm anchored by 250 GHz flux measurement. The derived FIR luminosity of Source A, Source B, and Source C is (13.0 ± 2.6), (5.1 ± 1.1), and (3.2 ± 0.9) × 10¹¹ L_⊙. In Figure 4, we compared these galaxies around ELAN MAMMOTH-I with other quasar samples, intermediate-z ULIRGs, normal SFGs and SMGs (Bothwell et al. 2013; Carilli & Walter 2013; Riechers 2013; Magdis et al. 2014; Cañameras et al. 2015; Daddi et al. 2015; Harrington et al. 2016; Strandet et al. 2017; Yang et al. 2017; Arabsalmani et al. 2018; Dannerbauer et al. 2019). The FIR luminosities of the galaxies in MAMMOTH-I are slightly lower than that of quasars but still consistent with normal SFGs; see Figure 4, right panel). The CO–FIR luminosity ratios for the MAMMOTH-I members are at the lower end but still follow the trend of the relation between L_IR and ${L}_{\mathrm{CO}(3-2)}^{{\prime} }$. This indicates that these galaxy members are all following the star-formation law, in which the star-formation rate traced by the IR luminosity has a tight relation with the mass of fuel traced by CO(3 − 2).

Zoom In Zoom Out Reset image size

The derived total FIR luminosity of the whole nebula from the 250 GHz observations is <2.7 × 10¹² L_⊙. Arrigoni Battaia et al. (2018b) reported a bright detection at observed frame 350 GHz (850 μm), with a flux of S_{350 GHz} = 4.6 ± 0.9 mJy (with a beam size of 15'') and a 3σ upper limit of about 16 mJy at observed frame 667 GHz (450 μm). The group of six galaxies are all covered by the 350GHz SCUBA-2 beam (∼15''). The total flux at the observed frame 250 GHz is <1.42 mJy. Assuming the modified blackbody for Source B (AGN; T_dust = 42 K, β = 1.6) and other galaxies (T_dust = 35K, β = 1.6), the derived total 350GHz flux should be <3.7 mJy. The results are slightly lower than the SCUBA-2 detections. This may be due to the standard matched filter applied in the SCUBA-2 data reduction in order to increase the point source detectability (Arrigoni Battaia et al. 2018b).

Assuming a star-formation-powered emission, we estimate the star-formation rate as SFR/M_⊙ yr⁻¹ = 4.5 × 10⁻⁴⁴ × L_FIR/erg s⁻¹ (Kennicutt 1998); see Table 2. The three sources detected at 250 GHz are violent starbursts with a star-formation rate ranging between 54–224 M_⊙ yr⁻¹.

Table 2. Physical Properties of the Protocluster Core MAMMOTH-I

Source	LFIR	SFR	${L}_{\mathrm{CO}(1-0)}^{{\prime} }$	${L}_{\mathrm{CO}(3-2)}^{{\prime} }$	r3,1	Mgas
	1011 L⊙	M⊙ yr−1	1010 K km s−1 pc2	109 K km s−1 pc2		1010 M⊙
(1)	(2)	(3)	(4)	(5)	(6)	(7)
G1(A)	13.0 ± 2.6	224 ± 45	1.2 ± 0.3	7.1 ± 1.1	0.59 ± 0.17	4.3 ± 0.1
G2(B)-narrow	5.1 ± 1.1	88 ± 19	1.1 ± 0.2	6.7 ± 1.4	0.61 ± 0.17	4.0 ± 0.1
G2(B)-broad			0.7 ± 0.4		0.91 ± 0.51
G3	<2.1	<36	<0.9	3.0 ± 1.1	>0.33	2.1 ± 0.8
G4	<2.1	<36	<0.9	3.7 ± 0.9	>0.41	2.6 ± 0.6
G5	<2.1	<36	<0.9	2.1 ± 0.6	>0.24	1.5 ± 0.4
G6(C)	3.2 ± 0.9	54 ± 15	1.0 ± 0.4	5.4 ± 1.2	0.54 ± 0.25	3.6 ± 0.1

Note. Col. (1): Source name. Col. (2): FIR luminosity from 8 to 1000 μm, assuming a modified blackbody for optically thin thermal dust emission, with a dust temperature of 42K for Source B (as it is a Type II AGN) and 35K for other galaxies. We adopt an emissivity index of β = 1.6, which is the typical value found in FIR-bright quasars at z ∼ 2−4 (Beelen et al. 2006). Given upper limits are 3σ. Col. (3): star-formation rate. Here we use SFR = 4.5 × 10⁻⁴⁴ × L_FIR (Kennicutt 1998). Col. (4): CO(1 − 0) luminosity of the ELAN MAMMOTH-I from Emonts et al. (2019). Col. (5): CO(3 − 2) luminosity from our NOEMA observations. Col (6): CO(3 − 2) /CO(1 − 0) line ratio. The fitted broad and narrow components of G2 are discussed in Section 4.2, which trace the center galaxy and the extended CGM gas, respectively. Col (7): molecular gas mass, assuming a typical conversion factor for high-z galaxies of α_CO = M_H2/${L}_{\mathrm{co}}^{{\prime} }$ = 3.6 M_⊙ (K km s⁻¹ pc²)⁻¹ (e.g., Daddi et al. 2010; Genzel et al. 2010). Gas masses for G1, G2 and G6 are derived from ${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ (Emonts et al. 2019). We assume that the other three galaxies are star-forming-dominated galaxies r_3,1 = 0.52 and derive gas mass from ${L}_{\mathrm{CO}(3-2)}^{{\prime} }$.

Download table as:  ASCIITypeset image

MAMMOTH-I resides in the density peak of the large-scale structure BOSS1441, which is one of the most overdense fields discovered to date with an LAE density 12 times higher than that of a random field density on a 15 cMpc scale. We discovered a remarkable galaxy concentration at a redshift of z ∼ 2.3, containing six gas-rich galaxies spectroscopically confirmed through the CO(3 − 2) transition in the central ∼100 kpc region. The high density of massive galaxies and velocity dispersion of this overdensity suggest that it could be embedded in a collapsed, cluster/group-sized halo. In this section, we further explore the total halo mass of this galaxy group.

Galaxy cluster velocity dispersion provides a reliable estimate of the cluster mass (Evrard et al. 2008; Munari et al.2013; Saro et al. 2013; Wang et al. 2016). The cluster redshift is z = 2.308, determined by the weighted average of the CO(3 − 2) redshifts of these six galaxies. The galaxy proper velocities v_i are then derived from their redshifts z_i by v_i = c(z_i − z)/(1 + z) (Danese et al. 1980). The line-of-sight velocity dispersion σ_v is the square root of the weighted sample variance of proper velocities (Beers et al. 1990; Ruel et al. 2014) which is estimated to be σ_v = 320 km s⁻¹. We assume that only the inner portion of this protocluster is virialized. Using the relation between velocity dispersion and total mass suggested in Evrard et al. (2008), ${\sigma }_{\mathrm{DM}}{(M,z)={\sigma }_{\mathrm{DM},15}(h(z){M}_{200c}/{10}^{15}{M}_{\odot })}^{\alpha }$, we derive a total halo mass of ELAN MAMMOTH-I of M_200c ∼ 10^13.1 M_⊙ (by using the canonical value of σ_DM,15 ∼ 1083km s⁻¹ and the logarithmic slope α ∼ 0.33617). It is an upper limit if the system has not yet virialized. The previous studies of AGNs, SMGs, and bright LABs show that they are expected to live in halos of 10^12–14 M_⊙ (e.g., Yang et al. 2010; White et al. 2012; Wang et al. 2016; Wilkinson et al. 2017). The total halo mass of the ELAN MAMMOTH-I is in a good agreement with that of AGNs and SMGs.

Previous work on protocluster cores based on sensitive observations of CO and ionized carbon (Miller et al. 2018; Oteo et al. 2018; Gómez-Guijarro et al. 2019) reported total molecular gas masses of ∼ 10¹¹ M_⊙ and a total halo mass as high as ∼ 10¹³ M_⊙. We use ${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ from Emonts et al. (2019) to derive the cold molecular gas mass M_H2 for Source A, Source B, and Source C. The gas masses are in the range of (3.6–4.3) × 10¹⁰ M_⊙. Here we assume a typical conversion factor for high-z galaxies of α_CO = M_H2/${L}_{{co}}^{{\prime} }$=3.6 M_⊙ (K km s⁻¹ pc²)⁻¹ (e.g., Daddi et al. 2010; Genzel et al. 2010). For the other three galaxies without detections of CO(1 − 0) we assume a gas excitation similar to typical SFGs with r_3,1 = 0.52. Their gas masses derived from ${L}_{\mathrm{CO}(3-2)}^{{\prime} }$ in our NOEMA observations are in the range of (1.5–2.6) × 10¹⁰ M_⊙, which is listed in Table 2. The total molecular gas mass in the MAMMOTH-1 protocluster is 1.8 ± 0.1 × 10¹¹ M_⊙. In summary, the gas, dust, and stellar properties of MAMMOTH-I are all comparable to these protocluster cores mentioned before, indicating that MAMMOTH-I should be the progenitor of a galaxy cluster.

Finally, we note that due to the influence of large-scale structure in and around clusters (White et al. 2010), there are uncertainties in the estimation of the mass for an individual cluster based on velocity dispersion. We are also aware that the sample used to estimate the velocity dispersion only includes SFGs. The quiescent galaxies would also change the estimate of the velocity dispersion (Wang et al. 2016). As we cannot rule out completely the existence of quiescent galaxies of this galaxy cluster, we are planning to confirm more member galaxies spectroscopically to further improve the accuracy of the velocity dispersion estimation.

Extended CO(1 − 0) emission has been found on the scale of tens of kpc around high-z massive galaxies or in protoclusters (e.g., Emonts et al. 2014, 2016; Dannerbauer et al. 2017). For ELAN MAMMOTH-I, Emonts et al. (2019) also reported that half of the cold molecular gas traced via the CO(1 − 0) transitions stretches on ∼30 kpc into the CGM, appearing to be a wide tail of gas in Source A(G1) and an extended reservoir of cold gas in Source B(G2). Extended high-J CO emission in the CGM has also been reported in some metal-line nebulae. Ginolfi et al. (2017) detected a large structure of molecular gas reservoir traced by CO(4 − 3) in an [O iii] nebula, extending over 40 kpc. Strikingly, with our NOEMA observations, we did not find any evidence of extended CO(3 − 2) emission.

This could be due to the fact that the critical density of CO (J, J − 1) scales roughly with n_crit ∝ J³. The ground-transition CO(1 − 0) has an effective critical density of only several 100 cm⁻³ and the J = 1 level of CO is substantially populated down to T ∼ 10 K. However, these values increase by an order of magnitude or more for the high-J transitions, like CO(3 − 2) and higher. Compared to CO(1 − 0), the CO(3 − 2) is more likely a tracer of the star-forming region or compact gas around an AGN. As in such regions, the gas is warmer and denser compared to that in the CGM. CO molecules tend to populate at high-J levels, and the optical depth also increases.

In the CGM around Sources A and B, the molecular gas is not excited to give strong CO(3 − 2) emission. The ELANe Slug (z = 2.282) and Jackpot (z = 2.041) also do not show any extended CO(3 − 2) emission (Decarli et al. 2021), whereas, for the ELAN around the Spiderweb Galaxy, CO(4 − 3) (and [C i]) are detected across ∼50 kpc, comprising ∼ 30% of the total flux. The Spiderweb Galaxy has a massive, cold molecular gas reservoir in the CGM that is roughly twice as luminous as that seen in CO(1 − 0) in MAMMOTH-I (Emonts et al. 2016, 2019). In addition, the Spiderweb Galaxy emits jets of relativistic particles visible in radio observations and has a metal-enriched outflow (Pentericci et al. 1997; Nesvadba et al. 2006). By contrast, MAMMOTH-1 does not have a radio jet and does not show any clear features of outflow in [C iv] and [He ii] observations (S. Zhang et al. 2021, in preparation). These could explain the reason why CO(3 − 2) around ELAN MAMMOTH-I is only associated with galaxies but not with the CGM.

In Figure 5 we revisit the CO(1 − 0) analysis of Source B from Emonts et al. (2019) and compare its CO(3 − 2) and CO(1 − 0) transitions. The CO(3 − 2) line shows a broad line profile with FWHM of ∼350 km s⁻¹. It is comparable to the typical line width of SMGs (e.g., Bothwell et al. 2013; Carilli & Walter 2013; Goto & Toft 2015) which implies that CO(3 − 2) traces the molecular gas from the star-forming regions within the galaxy. By comparison, while the CO(1 − 0) signal in Source B is dominated by a narrow CO(1 − 0) line (FWHM of 85 km s⁻¹) outside the central galaxies, at the location of the peak of the CO(3 − 2) emission there appears to be an additional weak, broad component. This broad CO(1 − 0) component is detected only at the 3.2σ level when minimizing the contribution of the narrow component (rightmost panel of Figure 5). However, its properties are remarkably similar to the CO(3 − 2) spectrum in Source B (G2) with FWHM = 440 ± 160 km s⁻¹, z = 2.3132 ±0.0009, I_CO(1−0) = 0.029 ± 0.015 Jy km s⁻¹, and r_3,1 = 0.9 ±0.5. The line ratio value is closer to AGN-dominated molecular gas emission. Therefore, the VLA data are consistent with the presence of two CO(1 − 0) features in Source B (G2), namely a CGM component of dynamically cold gas with very low velocity dispersion and excitation conditions (previously described in Emonts et al. 2019) as well as a weak counterpart to the CO(3 − 2) emission of molecular gas associated with the central galaxies. However, for the latter, Figure 5 shows that ambiguities in line-fitting remain and hence deeper CO(1 − 0) observations are needed before drawing any firm conclusions.

Zoom In Zoom Out Reset image size