The ALPINE-ALMA [CII] Survey:
CGM pollution and gas mixing by tidal stripping in a merging system at $z\sim 4.57$

Using the available ancillary data and catalogues we first identify the system studied in this paper to be composed of two massive and rest-frame UV bright galaxies (see an HST/ACS F814W cutout in Fig. 1a), at a mutual projected distance of $\sim$ 13 kpc: (1) $\rm VC\_9780$, a Lyman-break galaxy at $z_{\rm UV}=4.57$ (confirmed $z_{\rm[CII]}=4.573$, by ALPINE), with a stellar mass of $M_{\star}=1.1^{+0.4}_{-0.3}\times 10^{10}\leavevmode\nobreak\ {\rm M_{\odot}}$ and a star formation rate of ${\rm SFR}=38^{+29}_{-14}\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ (see Faisst et al. 2020); (2) $\rm C15\_705574$¹¹1$ID=705574$ in the COSMOS2015 catalogue (Laigle et al. 2016)., at a photometric redshift $z_{\rm phot}=4.62$ (confirmed $z_{\rm[CII]}=4.568$, by ALPINE), with $M_{\star}=1.2^{+1.0}_{-0.2}\times 10^{10}\leavevmode\nobreak\ {\rm M_{\odot}}$ and ${\rm SFR}=106^{+9}_{-65}\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ (from the COSMOS2015 catalogue²²2The physical properties of $\rm C15\_705574$ reported in the COSMOS2015 catalogue are computed adopting the photometric redshift. However, given the high accuracy of the latter ($|z_{\rm phot}-z_{\rm[CII]}|\sim 0.05$; see text) we assume $M_{\star}$ and ${\rm SFR}$ to be correct within their uncertainties. ; Laigle et al. 2016). To simplify the reading, we dub $\rm VC\_9780$ and $\rm C15\_705574$ as $\#A$ and $\#B$, respectively.

For both objects, $M_{\star}$ and SFR are measured through the spectral energy distribution (SED) modelling code LePHARE (Arnouts et al. 1999; Ilbert et al. 2006; Arnouts & Ilbert 2011) ran over the broadband photometry (including ground- and space-based imaging from the rest-frame UV to the near-IR; see Faisst et al. 2020) available for the well-studied COSMOS field (Scoville et al. 2007). From the HST/ACS F814W image, modelling the light profiles with a Sérsic function (Sérsic 1963) and assuming a Sérsic index n=1 (i.e., an exponential-disk profile), we estimate rest-frame UV effective radii of $r_{\rm e,UV}=1.0\pm 0.24$ kpc and $r_{\rm e,UV}=1.1\pm 0.3$ kpc for $\#A$ and $\#B$, respectively (see Fujimoto et al. 2020 for an accurate description of the size-fitting procedure for ALPINE galaxies).

In Fig. 1b we show a HST/WFC3 F160W cutout of the system³³3Data drawn from a recently approved mid-cycle-26 HST program (Faisst et al., in prep.). (rest-frame $\lambda_{\rm rest}\sim 2800\leavevmode\nobreak\ \AA$), while in Fig. 1c we display the UltraVista Ks-Band image from the fourth data release ($\rm DR4$⁴⁴4http://www.eso.org/rm/api/v1/public/releaseDescriptions/132; rest-frame $\lambda_{\rm rest}\sim 4000\leavevmode\nobreak\ \AA$), showing a faint elongation of $\#A$ towards the South-East (S-E; see also Sec. 3.3).

The close separation both in spatial projected distance ($\sim 13$ kpc in the HST rest-frame UV images) and redshift ($|\Delta z|=0.004$, corresponding to an absolute velocity offset of only 270 km s^-1; see Sec. 3.3) suggests that $\#A$ and $\#B$ are gravitationally bound. Moreover, the ratio of their stellar masses, $r_{\rm mass}\sim 0.9$, is very close to unity, indicating that the galaxy system is undergoing a close major merger event (e.g., Stewart et al. 2009; Lotz et al. 2010; Jones et al. 2020).

In order to explore the larger context in which this merging event is occurring, we relied on the wealth of other spectroscopic and photometric data available in the COSMOS field to map out the large-scale galaxy density surrounding our system. The mapping was done using the Voronoi Monte Carlo (VMC) technique, first presented in Tomczak et al. (2017), Lemaux et al. (2017, 2019), and Hung et al. (2020) for use at $z\sim 1$, and adapted for studies at higher redshift ($2\leq z\leq 5$) in Cucciati et al. (2018) and Lemaux et al. (2018). This technique statistically combines both spectroscopic and photometric redshifts to generate a suite of overlapping two-dimensional density maps in fine redshift slices over an arbitrary redshift range, and it has been shown to have high fidelity, precision, and accuracy in reconstructing the galaxy density field at a variety of different redshifts and for a variety of different observational data (see Lemaux et al. 2018; Lemaux et al., in prep.). While we refer to the works cited above for more details, we briefly describe the method in the Appendix A.

Our system is located in close projected spatial and redshift proximity to the massive proto-cluster ${\rm PClJ1001+0220}$ at $z\sim 4.57$, discovered by Lemaux et al. (2018). Thus, we focus here on generating a map that spans the fiducial redshift boundaries of ${\rm PClJ1001+0220}$ ($4.53\leq z\leq 4.60$). Such a redshift range corresponds to 7.5 $h_{70}^{-1}$ proper kpc, or $\sim 3750$ km s^-1. As in Lemaux et al. (2018), we use the full VUDS and zCOSMOS (Lilly et al. 2007, 2009; Diener et al. 2013, 2015; Lilly et al., in prep.) catalogs as our primary spectroscopic catalogs, with photometric redshifts derived from fitting to the photometry presented in the COSMOS2015 catalog (Laigle et al. 2016). In addition to the VUDS and zCOSMOS catalogs, we incorporate here new Keck ii/Deep Imaging Multi-Object Spectrograph (DEIMOS, Faber et al. 2003) observations of ${\rm PClJ1001+0220}$ (see a description of the observations in Lemaux et al., in prep.) which resulted in the spectroscopic confirmation of 106 new galaxies in the region in and around ${\rm PClJ1001+0220}$.

Shown in Fig. 2 is the resultant VMC galaxy overdensity map for the north-east (N-E) region of the ${\rm PClJ1001+0220}$ proto-cluster. Also shown are the original VUDS spectral members of the system (see Lemaux et al. 2018 for more details), as well as our $\#A$ – $\#B$ system. One of the new members confirmed by the DEIMOS observations ($ID=705080$) is also shown. This galaxy, with a $M_{\star}=5.3^{+2.5}_{-2.1}\times 10^{9}\leavevmode\nobreak\ {\rm M_{\odot}}$ and a ${\rm SFR}=22.2^{+17.1}_{-8.7}\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$, is located at $\sim 35$ $h_{70}^{-1}$ proper kpc from our merging system, although it is not detected by ALMA, neither in [CII] nor in the rest-frame FIR continuum. Measuring its Ly$\alpha$ we determine a redshift $z_{Ly\alpha}=4.5769$, corresponding to a velocity offset of $\Delta v\sim 260\leavevmode\nobreak\ {\rm km\leavevmode\nobreak\ s^{-1}}$ with respect to $\#A$ and $\Delta v\sim 580\leavevmode\nobreak\ {\rm km\leavevmode\nobreak\ s^{-1}}$ with respect to $\#B$⁵⁵5We note that the actual velocity offsets are likely to be even lower, since the redshift measured through Ly$\alpha$ can be off by few hundreds of ${\rm km\leavevmode\nobreak\ s^{-1}}$ toward the red (see e.g., Shapley et al. 2003; Trainor et al. 2015; Verhamme et al. 2018; Cassata et al. 2020).. Such a close projected spatial and redshift proximity to $\#A$ – $\#B$ suggests that DEIMOS $ID=705080$ is possibly associated to them.

As can be seen in Fig. 2, our merging system at $z\sim 4.57$ is nearly coincident with the overdensity peak of the N-E sub-component of the proto-cluster ${\rm PClJ1001+0220}$, and lie within $\leq$ 385 pkpc from the barycenter of this sub-component. The average overdensity across this region is $\log(1+\langle\delta_{gal}\rangle)=1.5$, comparable to the cores of $z\sim 1$ groups or the outskirts of massive clusters (e.g., Lemaux et al. 2017; Tomczak et al. 2017). Altogether these findings indicate that the merging system presented in this paper belongs to a large-scale dense and complex proto-cluster environment. In Sec. 4 we discuss possible implications of this scenario in the interpretation of our results.

In the next section we describe the results of ALMA observations of our system.

We use our ALPINE ALMA observations (see Sec. 2) of the [CII] line emission to perform a three-dimensional (3D) morpho-spectral decomposition of our merging system.

We start by extracting two spectra from large 1^′′ (diameter)-apertures centered on the rest-frame UV centroids of $\#A$ and $\#B$.

We then produce velocity-integrated maps of the [CII] emission in/around $\#A$ and $\#B$ (see Fig. 3) by collapsing the cube-channels included in the spectral range [$f_{\rm cen}-\sigma_{\rm G}$, $f_{\rm cen}+\sigma_{\rm G}$], where $f_{\rm cen}$ and $\sigma_{\rm G}$ are the central frequency and the standard deviation from a Gaussian model of the spectra. The corresponding velocity ranges are labelled in the top-left corners of both panels of Fig. 3, defined adopting a zero-velocity reference at $z_{\rm ref}=4.570$ (see next paragraphs of this Section). Although this will be discussed in detail further on by means of tailored analyses, we note that maps in Fig. 3 already suggest a disturbed morphology of the system, showing the presence of complex resolved extended emission around the sources.

We then estimate the spatial extent of the [CII]-emission arising from $\#A$ and $\#B$ by fitting a simple 2D Gaussian model to the velocity-integrated maps (Fig. 3), and use the ${\rm FWHM}_{x}\times{\rm FWHM}_{y}$ ellipses as apertures to extract the correct spectral information⁶⁶6As an alternative spectra-extraction procedure, we tested a 1^′′-aperture centred on the [CII] peaks, and the results are virtually unchanged, not affecting our interpretation..

This three-step process is needed to allow for any peculiar geometry or spatial offset between the [CII] and the rest-frame UV emission. Indeed, we note that the [CII] peak of $\#A$ is offset by $\gtrsim 0.5^{\prime\prime}$ from its HST centroid (see similar cases in e.g., Maiolino et al. 2015; Carniani et al. 2018b), with a clear elongation toward the S-E (see Fig. 3a). This [CII] tail is HST-dark, although a faint spike is possibly detected in the HST/WFC3 F160W image (see Fig. 1b). It is also broadly coincident with the rest-frame optical counterpart (see UltraVista DR4 Ks-band image in Fig. 1c), indicating a UV slope steeply rising towards longer wavelengths and therefore a significant presence of obscuring dust (see e.g., Calzetti et al. 2000), as also suggested by the $\gtrsim 4\sigma$ ALMA detection of co-spatial FIR-continuum emission (see Fig. 3a). With the currently available ALMA resolution, it is unclear whether this [CII] tail is produced by a dust-obscured star-forming region within $\#A$, or it is associated to a dusty neighbour at the same redshift.

The final [CII] spectra of $\#A$ and $\#B$ are shown in Fig. 4a (upper and lower panel, respectively), where we choose an intermediate redshift $z_{\rm ref}=4.570$ as a common zero-velocity reference for the system⁷⁷7This choice helps when comparing the single galaxy spectra with the emission line profile of the full [CII] emitting system, as discussed further on, whose centroid corresponds to a $z_{\rm ref}=4.570$.. Galaxy $\#A$, at $z_{\rm[CII]}^{\#A}=4.573$, has a [CII]-luminosity of $L_{\rm[CII]}^{\#A}=3.5\leavevmode\nobreak\ (\pm 0.3)\times 10^{8}\leavevmode\nobreak\ {\rm L_{\odot}}$, and a ${\rm FWHM}_{\#A}=190$ km s^-1, while $\#B$, at $z_{\rm[CII]}^{\#B}=4.568$ (blueshifted by $\rm\Delta v_{BA}=-270$ km s^-1 with respect to $\#A$) is broader (${\rm FWHM}_{\#B}=260$ km s^-1) and $\sim 2\times$ more luminous, $L_{\rm[CII]}^{\#B}=6.6\leavevmode\nobreak\ (\pm 0.4)\times 10^{8}\leavevmode\nobreak\ {\rm L_{\odot}}$. Interestingly, the second spectrum (see lower panel of Fig. 4a) shows two additional [CII] emitters, dubbed $\#C$ and $\#D$, spectrally offset by a few hundreds of km s^-1, and spatially close to $\#B$ (less than a $\sim 1^{\prime\prime}$-beam in projection; see velocity integrated map in the Appendix B). The line properties of $\#B$, $\#C$ and $\#D$ (total flux, centroid and FWHM) are estimated from a 3-Gaussian modelling of the spectrum. Although their integrated flux is detected at $>4\sigma$ (see also Appendix B), $\#C$ and $\#D$ are very faint ($L_{\rm[CII]}^{\#C}=7.7\leavevmode\nobreak\ \pm 1.5\times 10^{7}\leavevmode\nobreak\ {\rm L_{\odot}}$, $L_{\rm[CII]}^{\#D}=8.4\leavevmode\nobreak\ \pm 1.9\times 10^{7}\leavevmode\nobreak\ {\rm L_{\odot}}$), and narrow (${\rm FWHM}_{\#C}=40$ km s^-1, ${\rm FWHM}_{\#D}=57$ km s^-1), and are most likely small satellite galaxies (or gas clumps) orbiting around or accreting onto $\#B$.

To visualize the total [CII] emission arising from the full system we produce a velocity-integrated flux map by collapsing the spectral channels in the velocity range $[-500:+500]$ km s^-1, using a consistent $z_{\rm ref}$ as discussed above. Such a broad velocity domain is needed to safely include the four detected [CII]-emitting galaxies (see spectra in Fig. 4a). The resulting velocity-integrated [CII] map is shown in Fig. 4b. We find a very large [CII]-emitting structure extending on circumgalactic scales, with a projected diameter of $\sim 30$ kpc. Such a metal-enriched gaseous structure encloses the single galaxy components, and shows a large amount of [CII] arising from regions significantly displaced ($\gtrsim 10$ kpc) from the rest-frame UV / optical counterparts, especially along the direction orthogonal to the $\#A$ - $\#B$ axis. However, because of its 2D nature, the [CII] flux map shown in Fig. 4b suffers from the dilution of signal emitted by sources with a width much narrower than the velocity-filter adopted for the spectral integration. Therefore, to better visualize the 3D distribution of the gaseous structure we produce two position-velocity (PV) diagrams, one extracted along the $\#A$ - $\#B$ axis (Fig. 5a), and the other along the orthogonal direction, centred on $\#B$ (Fig. 5b). Both diagrams highlight the complex morpho-kinematic structure of the system, clearly showing the presence of circumgalactic tails around the interacting galaxies (see e.g., the [CII]-emitting regions labelled as ^′CGM^′ in both panels of Fig. 5).

One important piece of information is the net contribution of the four interacting galaxies themselves to the overall [CII] emission arising from the extended gaseous structure. To distinguish between the relative ^′galaxy^′ and ^′CGM^′ [CII] budgets, we extract a spectrum of the total [CII]-emitting system from a region within the 1-$\sigma$ contour of the velocity-integrated map (see grey solid line in Fig. 4b), and compare it with the sum of the single $\#A$, $\#B$, $\#C$ and $\#D$ spectra (see Fig. 6a). We note that, due to the dynamical complexity of our object, such a procedure is more accurate than extracting relative budgets directly from the velocity-integrated map, because of the limitations of the latter, as discussed above. We find that the full [CII]-emitting structure has a total luminosity of $L_{\rm[CII]}^{tot}=2.3\leavevmode\nobreak\ (\pm 0.2)\times 10^{9}\leavevmode\nobreak\ {\rm L_{\odot}}$, i.e., $\sim 2$ times larger than the sum of the [CII]-emitting galaxies. This implies that about $50\leavevmode\nobreak\ (\pm 6)\leavevmode\nobreak\ \%$ (see Fig. 6b) of the total [CII] luminosity resides between the individual components, $L_{\rm[CII]}^{CGM}=1.1\leavevmode\nobreak\ (\pm 0.2)\times 10^{9}\leavevmode\nobreak\ {\rm L_{\odot}}$, in a sort of circumgalactic and metal-rich gaseous envelope around the merging galaxies. This fraction of [CII] emission arising from the CGM is larger than what typically found around the few individual normal isolated galaxies (i.e., with not evident signs of major/minor mergers) that show a [CII]-halo (see Fujimoto et al. 2020), suggesting a link with the merging-nature of our system.

In the next section we discuss possible mechanisms responsible for such large amount of circumgalactic [CII] emission.

A natural scenario to explain the origin of the luminous gaseous envelope is that the circumgalactic [CII] traces gas stripped during gravitational encounters occurring in the dense merging system. Strong dynamical interactions between the four galaxies might therefore remove a significant amount of interstellar gas from their outer regions, polluting the CGM with chemically enriched material, and producing the complex morphology (see Fig. 3 and Fig. 4a) and kinematics (see Fig. 5) observed.

In this scenario, our observations could be a high-$z$ analogue of local studies of circumgalactic stripped [CII] in local groups. For instance Appleton et al. (2013) detect a broad [CII] emission arising from an extended shocked filament ($\sim 35$ kpc) produced by supersonic collisions between galaxies in the compact local group Stephan’s Quintet. They observe very large [CII]/Polycyclic Aromatic Hydrocarbon (PAH) and [CII]/FIR ratios, which cannot be accounted for by [CII] emission originating from photo-dissociation regions (PDRs; Stacey et al. 1991, 2010) only, and suggest that circumgalactic [CII] can be excited by the dissipation of mechanical energy (turbulence and shocks) injected by strong dynamical interactions. Such a mechanism is expected to be more frequent and more disruptive at high-$z$ because of (i) higher merger rates (e.g., Conselice 2006; Wetzel et al. 2009; Lotz et al. 2011) and (ii) generally higher gas fractions (e.g., Santini et al. 2014; Dessauges-Zavadsky et al. 2017; Schinnerer et al. 2016; Scoville et al. 2017; Tacconi et al. 2018), and might also explain previous similar observations of extended [CII] emission around multi-component systems at $z\sim 5$ (see ${\rm HZ6}$ and ${\rm HZ8}$ in Capak et al. 2015; Faisst et al. 2017). We will further explore this hypothesis by extending similar morpho-spectral decomposition analyses presented in this paper to other major/minor merging systems in the ALPINE survey (see a discussion on ALPINE mergers in Le Fèvre et al. 2019 and Jones et al. 2020) in future works.

Interestingly, since as discussed in Sec. 3.2 our merging system is close to an overdensity peak of the proto-cluster ${\rm PClJ1001+0220}$ at $z\sim 4.57$ (see Lemaux et al. 2018), the merger-induced CGM pollution might be a natural channel to feed and enrich the nascent proto- intra-cluster-medium (ICM; see e.g., Cucciati et al. 2014, Tozzi et al. 2003, and a recent review by Overzier 2016). Thus, our results could be some of the first observational evidence at high redshift that mergers and gas stripping are key mechanisms in driving the build-up and the chemical evolution of the ICM, as predicted by models and numerical simulations (see e.g., Gnedin 1998; Toniazzo & Schindler 2001; Tornatore et al. 2004; De Lucia et al. 2004; Kapferer et al. 2007). Also, the peculiar location of our system within a large-scale overdense structure suggests that it sits in proximity of a cosmic web node where several filaments from the IGM might be feeding into it. Shocks are predicted to occur as the intergalactic gas accretes onto massive galaxies and mixes with their CGM (see e.g., Birnboim & Dekel 2003; Kereš et al. 2009), and they likely contribute to the [CII] excitation. To quantitatively assess this scenario we will need new deep observations of the large-scale environment tracing the hydrogen gas, representing the bulk of the intergalactic streams, that can be traced e.g., through Ly$\alpha$ with MUSE (which has a field of view of $\sim 400$ kpc at this redshift).

Another possible contributing scenario is that a fraction of the circumgalactic metal-enriched gas traced by the extended [CII] envelope is expelled from the internal ISM of galaxies through galactic outflows. Evidence of star formation-driven winds has been recently observed at $z>4$ (see Sugahara et al. 2019; Faisst et al. 2020; Ginolfi et al. 2020; Cassata et al. 2020), and repeated episodes of outflows are thought to be responsible for the origin of diffuse [CII]-halos ($\sim 20$ kpc) detected around isolated massive star-forming high-$z$ galaxies (see Faisst et al. 2017; Fujimoto et al. 2019; Ginolfi et al. 2020; Pizzati et al. 2020; Fujimoto et al. 2020). However, a rough estimate suggests that outflows might not be the dominant mechanism for injecting the large amount of circumgalactic cold gas observed in our system (see Fig. 4b and Fig. 6). Using the $L_{\rm[CII]}-M_{\rm gas}$ relation calibrated by Zanella et al. (2018) and the [CII] luminosity arising from the CGM (i.e., about $50\%$ of the total $L_{\rm[CII]}$ (see last paragraph of Sec. 3.3) we obtain a rough estimate of the dense mass of in the [CII] gaseous envelope, $M_{\rm gas}^{\rm CGM}\sim 3\times 10^{10}\leavevmode\nobreak\ {\rm M_{\odot}}$. Then, using typical mass outflow rates of $\dot{M}_{\rm out}\sim 30\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ (as measured by Ginolfi et al. 2020 in the sub-sample of highly star-forming ALPINE galaxies), and assuming that both $\#A$ and $\#B$ significantly contribute to gas-expulsion⁸⁸8 We assume that the contribution from the satellite galaxies $\#C$ and $\#D$ to gas-pollution through outflows is negligible. Indeed, from their $L_{\rm[CII]}$, using the $L_{\rm[CII]}-{\rm SFR}$ relation calibrated by Schaerer et al. 2020, we find ${\rm SFRs}\lesssim 10\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$, and galaxies with such a low SFR at high-$z$ have been reported to not show signatures of prominent outflows (e.g., Ginolfi et al. 2020), nor signs of significant amount of circumgalactic emission (e.g., Ginolfi et al. 2020; Fujimoto et al. 2020). , we find that outflows would require at least $\sim$ 0.6 Gyr to expel the estimated amount of dense gas in the CGM. We note that this estimate is very conservative, since it is derived under the unlikely assumptions that (i) re-accretion does not occur, and (ii) both $\#A$ and $\#B$ have constantly had during their past growth a sufficient SFR to produce powerful outflows.

A third complementary process would be that part of the extended [CII] emission is induced by the star formation of several small satellites, which are unresolved by our ALMA observations and faint in the rest-frame UV HST images (the few light spots around $\#A$ and $\#B$ in the HST maps, see Fig. 1a and Fig. 1b, may belong to this category). To estimate the efficiency of this mechanism we estimate the possible diffuse SFR produced by faint satellites by integrating the rest-frame UV light within an aperture defined as the 1$\sigma$ level of the [CII] full structure (see grey line in Fig. 1a and Fig. 4b), and then subtracting the UV flux arising from regions defined as the blue ellipses (see Sec. 3.3 and Fig. 4b) associated with the individual galaxy components $\#A$, $\#B$, $\#C$ and $\#D$. Using a standard Kennicutt & Evans (2012) calibration, we estimate a total diffuse UV-based ${\rm SFR}_{\tiny{CGM}}^{\tiny{UV}}=13\pm 5\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$, which is about one order of magnitude lower than expected assuming a $L_{\rm[CII]}-{\rm SFR}$ relation⁹⁹9We used three different calibrations, which give consistent results: (i) ${\rm SFR}_{\tiny{CGM}}^{\tiny{[CII]}}\sim 129\pm 46\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ from Schaerer et al. (2020), calibrated for galaxies at $4<z<6$; (ii) ${\rm SFR}_{\tiny{CGM}}^{\tiny{[CII]}}\sim 124\pm 40\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ from De Looze et al. (2014), calibrated for local galaxies; (iii) ${\rm SFR}_{\tiny{CGM}}^{\tiny{[CII]}}\sim 125\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$ from the model of Lagache et al. (2018). Errors on the SFR are computed using the uncertainties on the parameters of the relations. (see e.g., De Looze et al. 2014; Lagache et al. 2018; Schaerer et al. 2020; see a similar calculation in Fujimoto et al. 2019), ${\rm SFR}_{\tiny{CGM}}^{\tiny{[CII]}}\sim 125\leavevmode\nobreak\ {\rm M_{\odot}\leavevmode\nobreak\ yr^{-1}}$. This suggests that, unless the diffuse SFR was exceptionally dust-obscured, which is disfavoured by the non-detection of ALMA FIR-continuum (see also Fudamoto et al. 2020, showing that low-mass objects are not expected to have high $L_{\rm IR}/L_{\rm UV}$), small faint satellites do not significantly account for the detected large circumgalactic [CII] emission.

Altogether, gas stripping by dynamical interactions and (with a possibly relative lower efficiency) galactic outflows and small faint satellites jointly contribute to the gas mixing and the metal enrichment of the CGM around high-$z$ merging systems. To resolve these mechanisms and quantitatively assess their relative efficiencies, we urge (i) deeper/ higher-resolution ALMA FIR-lines mapping and (ii) future rest-frame optical emission line observations with JWST, to be interpreted with tailored zoom-in cosmological simulations (e.g., Kohandel et al. 2019; Pallottini et al. 2019; Graziani et al. 2020).