Deciphering the Activity and Quiescence of High-redshift Cluster Environments: ALMA Observations of Cl J1449+0856 at z = 2

Thanks to a variety of efforts devoted to the search for distant progenitors of galaxy clusters, pushing toward z ∼ 2 and beyond (e.g., Rosati et al. 1999; Kurk et al. 2000; Mullis et al. 2005; Stanford et al. 2006, 2012; Venemans et al. 2007; Eisenhardt et al. 2008; Andreon et al. 2009; Wilson et al. 2009; Papovich et al. 2010; Gobat et al. 2011; Santos et al. 2011; Spitler et al. 2012; Muzzin et al. 2013; Wylezalek et al. 2013; Clements et al. 2014; Bleem et al. 2015; Casey et al. 2015; Strazzullo et al. 2015; Flores-Cacho et al. 2016; Wang et al. 2016; Cai et al. 2017; Daddi et al. 2017; Mantz et al. 2018), it has been possible in the past decade to significantly extend the timeline and scope of galaxy evolution studies in the densest high-redshift environments. This has eventually bridged the historically divided cluster and protocluster regimes (see recent review by Overzier 2016), at a cosmic time that is thought—and indeed turns out—to be a transformational epoch for both clusters and their galaxies. The synergy of observations at different wavelengths, including optical/near-IR (NIR) to probe stellar populations and galaxy structural properties, mid/far-IR to radio to probe star formation rates (SFRs), submillimeter for gas reservoirs, and X-ray, mid-IR, and radio for nuclear activity, has proved fundamental in exploring the many facets of cluster galaxy populations, as discussed below.

From observations of massive cluster galaxies at lower redshifts (e.g., Andreon 2006; De Propris et al. 2007; Lidman et al. 2008; Mei et al. 2009; Mancone et al. 2010; Strazzullo et al. 2010; Wylezalek et al. 2014), we expect that the epoch around z ∼ 2 corresponds to the transition from a regime of widespread, high levels of star formation in dense environments to the quiescent regime characteristic of cluster cores at z ≲ 1. Direct observations at high redshifts have in fact detected increasing levels of star formation, as well as nuclear and merging activity in distant z ≳ 1.5 groups and clusters (e.g., Hayashi et al. 2010, 2011; Hilton et al. 2010; Zeimann et al. 2012; Brodwin et al. 2013; Lotz et al. 2013; Dannerbauer et al. 2014; Ma et al. 2015; Popesso et al. 2015b; Santos et al. 2015; Tran et al. 2015; Alberts et al. 2016; Hung et al. 2016; Wang et al. 2016; Krishnan et al. 2017; Nantais et al. 2017, and references therein). At the same time, passively evolving galaxies are often found to be overrepresented, to different degrees, in the densest regions of these environments (e.g., Steidel et al. 2005; Kurk et al. 2009; Papovich et al. 2010; Snyder et al. 2012; Spitler et al. 2012; Gobat et al. 2013; Kubo et al. 2013; Strazzullo et al. 2013; Tanaka et al. 2013; Hatch et al. 2016), and examples of (generally massive) clusters with already very strongly suppressed star formation are also found up to z ∼ 2 (e.g., Strazzullo et al. 2010; Andreon et al. 2014; Newman et al. 2014; Cooke et al. 2016). Cluster selection, cluster-to-cluster variation, the intrinsically transitional phase of galaxy populations at this time, and observational difficulties have all contributed to assemble a varied, still unfinished picture that might at times still look controversial in some aspects, and sometimes difficult to reconcile with theoretical expectations (e.g., Granato et al. 2015).

Thanks to expensive—and thus still limited to a relatively small number of systems—dedicated follow-up programs, a number of recent studies have started investigating in more specific detail the properties of both quiescent and star-forming galaxies in distant (proto)cluster environments. Such studies explored a variety of aspects, including the environmental dependence of stellar ages and structure of passive populations (e.g., Gobat et al. 2013; Strazzullo et al. 2013; Tanaka et al. 2013; Andreon et al. 2014, 2016; Newman et al. 2014; Beifiori et al. 2017; Lee-Brown et al. 2017; Prichard et al. 2017; Chan et al. 2018); the environmental dependence of the specific SFRs (sSFRs), metallicities, and dust attenuation properties of star-forming galaxies (e.g., Tanaka et al. 2010; Hatch et al. 2011; Hayashi et al. 2011, 2016; Koyama et al. 2013, 2014; Kulas et al. 2013; Cooke et al. 2014; Smail et al. 2014; Kacprzak et al. 2015; Shimakawa et al. 2015, 2018; Tran et al. 2015; Valentino et al. 2015; Husband et al. 2016; Kewley et al. 2016); and the environmental dependence of cold gas reservoirs fueling star formation in dense environments (e.g., Aravena et al. 2012; Casasola et al. 2013; Emonts et al. 2013; Ivison et al. 2013; Tadaki et al. 2014; Casey 2016; Gullberg et al. 2016; Dannerbauer et al. 2017; Hayashi et al. 2017, 2018; Lee et al. 2017; Noble et al. 2017; Rudnick et al. 2017; Stach et al. 2017). Results from such investigations critically shape our understanding of galaxy population properties—and of the processes affecting galaxy evolution—in early dense environments, though the still very limited cluster galaxy samples, small number of clusters probed, and selection biases continue to preclude conclusive interpretations.

We present here new results from Atacama Large Millimeter/submillimeter Array (ALMA) observations of the galaxy cluster Cl J1449+0856 (hereafter Cl J1449; Gobat et al. 2011) at z = 2, complementing our previous work on its galaxy populations (Gobat et al. 2013; Strazzullo et al. 2013, 2016; Valentino et al. 2015, 2016) with a critical independent vantage point. Cl J1449 was identified as an overdensity of IRAC color-selected galaxies and spectroscopically confirmed with now ∼30 spectroscopic members (Gobat et al. 2011, 2013; Valentino et al. 2015; Coogan et al. 2018). The estimated halo mass based on its extended X-ray emission and stellar mass content is (5–7) × 10¹³ M_⊙ (Valentino et al. 2016, and references therein), placing this structure in the mass range of the average progenitors of today's typical massive clusters. For what directly concerns the results presented here, our previous work has highlighted the mixed galaxy population in this cluster, consisting of both quiescent and highly star-forming sources (as well as active galactic nuclei [AGNs]; Gobat et al. 2013; Strazzullo et al. 2013; Valentino et al. 2015, 2016). In particular, in Strazzullo et al. (2016, hereafter S16) we investigated the nature of the massive red population characterizing the cluster core, in terms of dusty massive star-forming galaxies and sources with already-suppressed star formation, based on a purely photometric analysis at optical/NIR wavelengths. On the other hand, in spite of the statistical validity of this approach, the ultimate confirmation of the nature of such sources, as well as an actual estimate of the (obscured) star formation occurring in the cluster core, remains with SFR indicators not biased by dust attenuation. Cl J1449 had been previously observed with Spitzer/MIPS (24 μm) and later also with Herschel/PACS and SPIRE, which indeed suggested potentially high levels of star formation activity right in the cluster core as already reported in the first study by Gobat et al. (2011). However, the angular resolution and depth of these data, and/or ambiguities with respect to contamination from nuclear activity, hampered the effectiveness of these observations in establishing a reliable picture of star formation and quenching in this system.

In this work based on ALMA observations, we thus focus on three main aspects: (1) the quantification of star formation occurring in the core of Cl J1449, for the first time using deep, high-resolution star formation probes not biased by dust attenuation (Sections 3 and 4.1); (2) the constraints set by these new observations on the first massive quiescent galaxies that, even as early as z ∼ 2, are a significant feature of the core of Cl J1449 (Section 4.2); and (3) the direct observation of the forming brightest cluster galaxy (BCG; Section 5), which, especially thanks to the combination with Hubble Space Telescope (HST) data, produces a remarkably detailed picture of a critical phase in the early evolution of (proto-)BCGs. Besides the scope of this paper, the ALMA observations discussed here enable the investigation of a significantly wider range of questions related to the effect of the environment on the properties and evolution of high-redshift cluster galaxies: the companion paper by Coogan et al. (2018, hereafter C18) presents in particular the dust and gas properties of the ALMA-detected cluster members and provides extensive descriptions of all submillimeter and radio observations of Cl J1449.

We assume a ΛCDM cosmology with Ω_M = 0.3, Ω_Λ = 0.7, and H₀ = 70 km s⁻¹ Mpc⁻¹. Stellar masses and SFRs are quoted for a Salpeter (1955) initial mass function (IMF). Magnitudes and colors are quoted in the AB system.

The central area of Cl J1449 was observed in ALMA Cycle 1 and 3 programs 2012.1.00885.S and 2015.1.01355.S. The Cycle 1 program obtained a band 7 mosaic probing 870 μm continuum over a ∼0.3 arcmin² region in the cluster central area. Observations were completed in 2014 December for a total on-source time of ∼2.3 hr. The probed field, offset by 8'' from the cluster center, reaches out to clustercentric distances of ∼100–200 kpc depending on the direction (see Figure 1). For comparison, the estimated cluster virial radius is r₂₀₀ ∼ 0.4 ± 0.1 Mpc (Gobat et al. 2013; Valentino et al. 2016).

Zoom In Zoom Out Reset image size

The synthesized beam of FWHM ∼ 14 × 06 is well suited for the size of z ∼ 2 galaxies and to avoid confusion in the crowded cluster-core environment. The map has an rms sensitivity of ∼70 μJy beam⁻¹ and thus, as a reference, reaches down to a SFR of ∼40 M_⊙ yr⁻¹ (3σ; Béthermin et al. 2012, see Section 4.1) for "main sequence" (MS; Elbaz et al. 2011) star-forming galaxies at z ∼ 2, corresponding to stellar masses of ∼2 × 10¹⁰ M_⊙ (Sargent et al. 2014). The same program also obtained a single overlapping pointing in band 3 to probe CO(3–2) emission from cluster galaxies at matched depth with the 870 μm continuum, with the main goals of confirming cluster membership of sources detected in the 870 μm map and estimating gas reservoirs fueling their star formation. However, as the later acquired CO(4–3) observations discussed below are deeper (see C18), and as higher-order CO transitions are better SFR (rather than total molecular gas mass) tracers (e.g., Daddi et al. 2015; Liu et al. 2015), these CO(3–2) data are not used for the purposes of this work.

The Cycle 3 follow-up observed again this region with a band 4 pointing to probe CO(4–3) emission from cluster members. Observations were completed in May 2016, for a total on-source time of ∼2 hr. The FWHM of the primary beam is ∼41'', well matched to the 870 μm mosaic (Figure 1). The synthesized beam FWHM is ∼12 × 1''. The rms is 10 mJy km s⁻¹ over 100 km s⁻¹, corresponding to a 3σ detection limit of ∼35 M_⊙ yr⁻¹ for a MS galaxy at z = 2 with a CO(4–3) line FWHM of ∼400 km s⁻¹.

We focus here on the constraints on star formation activity and its suppression in the core region of Cl J1449 derived from the observations of 870 μm continuum and CO(4–3) line emission. We refer the reader to C18 for a full, extensive summary of the ALMA observations and for a detailed description of the measurements of the 870 μm continuum and CO(4–3) line fluxes that we use in this work. Summarizing those aspects most important to the analysis presented here, we note that C18 detected continuum sources and independently searched for spectral lines in the band 7 and band 4 observations. Then, both 870 μm continuum and CO(4–3) line fluxes were measured at the positions of all these millimeter-detected sources, as well as at the positions of all known cluster members from our previous optical/NIR studies (Gobat et al. 2011, 2013; Strazzullo et al. 2013, 2016; Valentino et al. 2015). All the analysis and results below are based on these measurements as described in C18.

The 870 μm map of the core of Cl J1449 reveals six continuum sources (S/N > 4) with F_870μm > 0.5 mJy (Figure 1), giving a projected source density in the ∼0.3 arcmin² survey field that is a factor of ∼6 higher than expected from blank-field counts (e.g., ∼1.0 ± 0.8 sources would be expected from Oteo et al. 2016). In fact, all six sources are concentrated within a circle of r ∼ 015. Four of the six 870 μm sources are brighter than 1 mJy, resulting in a projected overdensity of ∼10 (∼0.3 ± 0.3 sources expected to this flux limit from Oteo et al. 2016). We note that, as discussed below, the two brightest 870 μm sources are likely to be background interlopers. We estimate their flux magnification due to the cluster potential, assuming a spherical Navarro, Frenk & White (NFW, 1997) halo of mass 6 × 10¹³ M_⊙ and a concentration in the range c = 1–5. For the brightest, ∼6 mJy source (labeled A5 below), the estimated photometric redshift z ∼ 2.8 yields a magnification factor of ∼12% (5%–18% within the 3σ range of the photometric redshift). For the other source (A4, ∼1.9 mJy), we estimate a magnification of about 9%–20% (30%, 40%) for a source redshift from 2.5 to 3.5 (5, 7). Even for a source redshift z ∼ 7, the flux would still be brighter than 1 mJy. We thus conclude that the overdensity of millimeter-bright sources is not likely significantly affected by lensing magnification by the cluster potential.

The left panel of Figure 1 identifies all ALMA 870 μm selected sources discussed in this work (labeled A1 to A6), while the right panel identifies the HST-selected (F140W) red sources identified in S16 (labeled H1 to H13; note that H6 and A6—and H10 and A5—correspond to the same galaxy, as indicated). Of the six 870 μm sources, two have spectroscopic redshifts measured in the Gobat et al. (2013) HST grism follow-up of Cl J1449: a foreground galaxy at z ∼ 1.3 (A3, thus ignored henceforth), and the cluster member A6 = H6. None of the other 870 μm sources have an optical/NIR spectroscopic redshift determination. The HST counterparts to A1, A2, and A4 are very faint¹⁸ and were thus not included in our previous studies of galaxy populations in Cl J1449. A5 was included in our previous work and was deemed to be likely an interloper at z_phot ∼ 2.8 (Strazzullo et al. 2013).

On the other hand, A1 and A2 both show a highly significant detection of CO(4–3) line emission (Figure 2 and C18), securely confirming their cluster membership. However, no lines are detected for the two brightest 870 μm sources, A4 and A5. In fact, as discussed at length in C18, in spite of their high 870 μm fluxes, no lines are detected for these sources in any of our data sets probing CO(4–3), CO(3–2), and CO(1–0) at the cluster redshift, as well as bright millimeter lines ([C i](2–1), C ii, CO transitions up to CO(7–6)) over a significant fraction of the 1 < z < 9 range. Nonetheless, several redshifts remain unprobed, notably including the range around the photometric redshift of A5 (C18). Therefore, we do not currently have confirmation of the redshift of A4 and A5. We note that the likelihood of observing two such bright sources unrelated to Cl J1449 in the small field probed is extremely low: as discussed in more detail in C18, these sources might in principle still be cluster members with very recently and rapidly suppressed star formation, with the lack of CO(4–3) emission being potentially reconciled with their bright 870 μm continuum by the gas and dust tracing star formation on different timescales. Nonetheless, given their large 870 μm fluxes and thus expected very bright CO line emission compared to the depth of our observations (Figure 3), we currently conclude that, at face value, the most likely explanation is that A4 and A5 are interlopers. Among the six bright 870 μm sources, only A1, A2, and A6 are thus confirmed to belong to the cluster.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.1. Dusty Star Formation in the Cluster Core

CO(4–3) emission is detected at >3σ for a total of seven cluster galaxies, including previously known members (H1, H3, H6 = A6, HN7, H13) and those newly confirmed by the detection of the CO line itself (A1, A2; Section 3). All but one of these (H3) also have a >2.5σ 870 μm detection (Figures 2, 3). Figure 3 shows the IR luminosities L_IR of these sources as estimated from the 870 μm continuum flux (L_IR,870μm) or the CO(4–3) line flux¹⁹ (L_IR,CO43).

The IR luminosity estimate L_IR,870μm was derived from the measured 870 μm flux using the average MS and starburst (SB) SEDs from Béthermin et al. (2012) at z = 2. The L_IR,CO43 estimate was derived from the measured CO(4–3) line flux by assuming the CO spectral line energy distributions (SLEDs) (in particular, the CO(5–4)/CO(4–3) line ratio) of ULIRGs²⁰ (intended to represent SBs) and star-forming BzK galaxies (intended to represent MS sources; from Papadopoulos et al. 2012; Daddi et al. 2015, respectively) and the CO(5–4) versus L_IR relation from Daddi et al. (2015). We consider the adopted assumptions on the CO(5–4)/CO(4–3) line ratio as the most appropriate for galaxies in this sample, but we also show for comparison in Figure 3 the effect on the L_IR,CO43 estimate of a range of different assumptions on the CO SLED, including those measured for the Milky Way (inner region; Fixsen et al. 1999), SMGs (Bothwell et al. 2013), and the Papadopoulos et al. (2012) model (see discussion in Daddi et al. 2015, and references therein). The CO(5–4)/CO(4–3) line ratios from these different determinations are affected to different degrees by measurement uncertainties (see the original papers for details), but for what concerns this work we note that the impact of the different plausible line ratios (including their uncertainties) on our L_IR,CO43 estimate is clearly subdominant with respect to the scatter in the CO(5–4) versus L_IR relation, as well as to the typical measurement errors of CO(4–3) line fluxes in this work.

Figure 3 shows these L_IR,870μm and L_IR,CO43 estimates and the related uncertainties as follows. Black error bars show statistical uncertainties from flux measurement errors. The double symbol adopted for all sources highlights the systematic uncertainties in estimating IR luminosities from the 870 μm continuum assuming a MS or SB SED, or from CO(4–3) line fluxes assuming the BzK or ULIRG SLED, as indicated. The thick gray error bar and shaded area along the bisector show the estimated intrinsic scatter of the adopted scaling relations, that is, the scatter in SED shape (dust temperature) and in the CO(5–4) versus L_IR relation (concerning L_IR,870μm and L_IR,CO43, respectively). Hatched regions in the figure mark IR luminosities below a corresponding reference 3σ limit estimated by assuming the following: for L_IR,870μm, the 3σ limit of the 870 μm map (Section 2) and the MS SED; and for L_IR,CO43, the 3σ limit of the band 4 observations at field center (Section 2), a line width of 400 km s⁻¹, the BzK CO(5–4)/CO(4–3) line ratio, and the CO(5–4) versus L_IR relation. While these are shown as an indication, the measurements, errors, and upper limits shown for the individual sources account for their actual position within the band 4 primary beam FWHM and highlight the systematics due to the SED or CO SLED choice as discussed, though upper limits for L_IR,CO43 still assume a line width of 400 km s⁻¹.

As Figure 3 shows, the L_IR,CO43 and L_IR,870μm IR luminosity estimates are typically consistent within the estimated uncertainties. The two obvious exceptions are A4 and A5, which both have high L_IR,870μm from the bright 870 μm flux but no CO emission, leading to an inconsistent upper limit on L_IR,CO43 even when accounting for the estimated uncertainties. As discussed above, we therefore conclude that these sources are in fact interlopers. For all other sources, the consistency of L_IR,870μm and L_IR,CO43, besides ensuring cluster membership of the 870 μm detections, also confirms the reliability of the SFR estimates.

Summing the derived infrared luminosities of cluster members within the probed ∼0.08 Mpc² (proper, at z = 2) region yields a total L_IR ∼ (4.3 ± 0.5) × 10¹² L_⊙ (the error corresponding to the range obtained from L_IR,CO43 with both ULIRG and BzK SLEDs, and L_IR,870μm with a MS SED²¹ ), corresponding to a total SFR of ∼700 ± 100 M_⊙ yr⁻¹ (adopting the Kennicutt [1998] calibration). This yields an overall projected SFR density of ∼(0.9 ± 0.1) × 10⁴ M_⊙ yr⁻¹ Mpc⁻² and a SFR volume density of ∼(1.0 ± 0.1) × 10⁴ M_⊙ yr⁻¹ Mpc⁻³ within the probed region²² (over the probed fraction of the virial volume, given the estimated cluster virial radius and assuming that the cluster is spherical). Again, these estimates assume that the two brightest 870 μm sources A4 and A5 are interlopers: A5 for itself would otherwise contribute a SFR ∼ 1000 M_⊙ yr⁻¹.

The total unobscured SFR of the ALMA-detected cluster members as estimated from the rest-frame UV luminosity L_UV is <20 M_⊙ yr⁻¹. Given the SFR threshold reached by these observations, the high L_IR/L_UV of the resulting ALMA-detected sample further highlights how galaxy populations in this cluster core are unusually skewed toward very reddened sources (see also Figure 2; further discussion in Section 4.1.1 and in C18). For comparison, the total unobscured SFR of all cluster galaxies within the same region is estimated to be in the range 100 ± 20 M_⊙ yr⁻¹, after correcting for incompleteness using the field UV luminosity functions from Parsa et al. (2016) and Alavi et al. (2016).

The measured SFR density is obviously orders of magnitude higher than the field average at the same redshift (e.g., Madau & Dickinson 2014), as observed in various kinds of other high-redshift structures (e.g., Clements et al. 2014; Dannerbauer et al. 2014; Santos et al. 2015; Tran et al. 2015; Wang et al. 2016). This is in fact largely due to the concentration of (star-forming) galaxies within the small volume of the dense cluster core, rather than to individual galaxies having particularly high SFRs. In fact, the overall sSFR in the ALMA-probed field is ~1.1 ± 0.6 Gyr⁻¹, compared to a field average at z = 2 of $\sim {1.9}_{-0.9}^{+1.9}$ Gyr⁻¹ (Madau & Dickinson 2014). As shown below (Figure 5) and also discussed in C18, the SFRs of individual sources are generally consistent with MS levels, with the possible exception of the two brightest sources A1 and A2 having higher SFRs.

By comparison with the cluster mass ${M}_{\mathrm{halo}}\sim (5\mbox{--}7)\times {10}^{13}$ M_⊙, the estimated L_IR-derived SFR within the probed volume gives a lower limit (that is, not correcting for the part of the virial volume left unprobed by our ALMA observations) to the total SFR density SFR/M_halo of ∼1300 ± 400 M_⊙ yr⁻¹/10¹⁴ M_⊙, after a small correction of the total IR luminosity for the >3σ L_IR,870μm sample down to L_IR = 10⁷ L_⊙ assuming the Popesso et al. (2015a) group luminosity function at z ∼ 1.6, which should be the most appropriate for this system (see discussion in Popesso et al. 2015a). At face value, this is in line with the Popesso et al. (2015b) prediction at z = 2 for massive groups, although we remind that this is a lower limit and it is currently not possible to reliably estimate the overall contribution of the cluster outskirts (we note though that, in the available observations, essentially all the measured IR luminosity is contributed by the very central cluster region, r ≲ 100 kpc; see Figure 1). The derived lower limit to SFR/M_halo lies at the upper edge of the Alberts et al. (2016) measurements at z ∼ 1.4 (accounting for the different IMF and marginal correction to the same L_IR limit), in agreement with the expected further increase out to z = 2, though we also note that the Alberts et al. (2016) clusters have larger estimated halo masses in the range (2–5) × 10¹⁴ M_⊙ and thus are expected to have lower SFR/M_halo (e.g., Webb et al. 2013; Popesso et al. 2015b). Indeed, results from the lower halo mass sample of Alberts et al. (2014) would give significantly higher SFR/M_halo in the same redshift range (see discussion in Alberts et al. 2016), also higher than our lower limit measured here. Similarly, although our lower limit tends to be higher than measurements by, e.g., Smail et al. (2014), Ma et al. (2015), and Santos et al. (2015) on clusters of very different masses (8 × 10¹³ M_⊙ to 5 × 10¹⁴ M_⊙) at z ∼ 1.5–1.6 (SFR/M_halo overall in the range ∼500–1000 M_⊙ yr⁻¹/10¹⁴ M_⊙), it would be fully in line with these measurements for a SFR density evolution similar to that predicted by, e.g., Geach et al. (2006). For comparison, the ∼3400 M_⊙ yr⁻¹ observed within the 80 kpc core of the similarly massive (M_halo ∼ 8 × 10¹³ M_⊙) Wang et al. (2016) cluster at z = 2.5 result in a lower limit SFR/M_halo > 4000 M_⊙ yr⁻¹/10¹⁴ M_⊙.

4.1.1. Color Distribution of ALMA-detected Cluster Galaxies

Despite the poor statistics due to the very small field probed and relatively small number of massive star-forming cluster members, the ALMA-detected sample in this region appears unusually skewed toward very red (F105W–F140W, dust-uncorrected) sources. We show in Figure 4 (bottom panel) the color distribution of subsamples of the ALMA-detected sources in Cl J1449 with different stellar mass and SFR thresholds, as indicated. The adopted stellar mass thresholds log(M/M_⊙) = 10.1 and 10.5 correspond to the lowest mass of the ALMA-detected cluster members and to a mass above the mass completeness limit of the S16 sample, where our formal 3σ limit on CO(4–3) based SFR probes essentially all of the 1σ range of the MS (Figure 5). The adopted SFR thresholds correspond to the nominal 3σ and 5σ limits of the CO(4–3) observations in the assumptions discussed in Section 4.1.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 (top panel) shows for comparison the color distribution of field galaxies at z_phot = 2 ± 0.3 with the same stellar mass and SFR limits, from a control field in GOODS-S. The control field is the same as used in S16, and the adopted measurements are described there in more detail (e.g., their Section 2). We briefly remind here that we used the Guo et al. (2013) photometry, as well as stellar masses, photometric redshifts, and model SEDs from Schreiber et al. (2015) and Pannella et al. (2015). We note that both stellar mass and SFR estimates are derived from SED fitting for the field samples, while we use the CO(4–3) based SFRs for cluster galaxies. The selection of the field comparison samples cannot thus be considered as properly equivalent to the selection of the cluster samples, because of the initial field sample selection (Guo et al. 2013) and the obvious biases between the different (CO(4–3) vs. SED-based) SFR estimates adopted. In this respect, we further note that for the purpose of estimating SFRs for the field sample, a constant star formation history was assumed (model SEDs synthesized with Bruzual & Charlot 2003) for all sources, allowing for a wide dust attenuation range (A_v = 0–6, assuming the Calzetti et al. [2000] attenuation law). SED-based SFRs derived from this modeling have been shown to agree with L_IR-based estimates within a ∼0.2 dex scatter (e.g., Pannella et al. 2015, for the same SED analysis as used here).

The green and blue/orange histograms in Figure 4 (top panel) refer to, respectively, samples including all galaxies or only galaxies classified as star-forming based on their rest-frame UVJ colors (Williams et al. 2009). While the blue/orange histograms are thus the main reference for the expected color distribution, the green histograms are shown for comparison to account for misclassification of dusty star-forming galaxies as quiescent sources; as the bulk of UVJ-quiescent galaxies are expected to be actually quiescent, this is a conservative comparison sample in this respect, as it maximizes the fraction of red sources.

We show Figure 4 as an indication of our rough expectations for the colors of a mass- and SFR-selected sample of (field) galaxies at the cluster redshift. Considering 1000 realizations of galaxy samples of the same size as the cluster samples shown in Figure 4, randomly drawn from the corresponding field distribution, returns a fraction of F105–F140 > 1.3 sources larger than that measured in the cluster samples in <2% of the realizations at worse (for the log(M/M_⊙) > 10.1, SFR > 60 M_⊙ yr⁻¹, or generally <0.5%).²³ In spite of the caveats outlined above deriving from the nonequivalent selection of the cluster and field samples due to the different adopted SFR indicators, and of the very small number statistics of the ALMA-detected sample in the cluster, the comparison with the first-order expectations from the field sample suggests at face value that very obscured sources are more prevalent than in the field. If confirmed, this would point toward environmental effects possibly related to merger-driven star formation episodes (see, e.g., C18 and references therein) and/or differences in star formation histories in the cluster environment.

4.2. The ALMA View of the Red Cluster Galaxy Population

Figure 6 highlights the complex of multiple, likely interacting components located close to the center of the extended X-ray emission and galaxy overdensity in Cl J1449, identified as the forming cluster proto-BCG (Gobat et al. 2011), and including sources H1, H4, H5, and A1. We note that although other cluster members (H2, H6) have stellar masses consistent with the individual masses of H1, H4, and H5 (see S16 and Figure 5), the configuration of the H1, H4, H5, A1 complex discussed below, as well as its location with respect to the galaxy overdensity and X-ray emission, is clearly much more suggestive, as compared to H2 or H6, of the site of main formation of the future cluster BCG. We stress for clarity that we identify this galaxy complex as a whole as the forming proto-BCG and that we do not observe in Cl J1449 any galaxy already exhibiting the peculiar features of BCGs.

Zoom In Zoom Out Reset image size

Within a radius of r = 15 we identify the two massive quiescent sources H4 and H5, the massive star-forming galaxy H1 with potentially sub-MS star formation activity, and the optically faint and very red, millimeter-brightest cluster member A1, which is forming stars at a few hundred solar masses per year (Sections 3, 4.1, and 4.2). Indeed, Spitzer/MIPS 24 μm imaging already suggested this level of star formation associated with the proto-BCG (with obvious uncertainties related to the probed rest-frame wavelength and poor angular resolution; Gobat et al. 2011). The combination of HST color imaging and ALMA observations proves that the bulk of the star formation in the proto-BCG is actually occurring in the optically faint, seemingly minor component A1, rather than in the optically bright source H1 as originally thought (Gobat et al. 2013; Strazzullo et al. 2013). Additional faint components and tails, whose cluster membership and properties cannot be reliably determined with the current observations, are observed within r ≲ 15 from the proto-BCG (Figure 6).

These observations are strongly suggestive of an actively forming BCG still assembling its stellar mass through star formation and merging. This is in line with results from the few first studies on distant BCGs, suggesting a strong increase in the fraction of highly star-forming systems (Webb et al. 2015b; McDonald et al. 2016; Bonaventura et al. 2017), with L_IR > 10¹² L_⊙ sources likely approaching  ≳50% of BCGs toward z ∼ 2. As also discussed in previous work (Webb et al. 2015a, 2015b; Kubo et al. 2016; Bonaventura et al. 2017), gas-rich mergers might play a significant role in this phase of BCG evolution, as suggested by our observations as well (see also the related discussion on merger-driven star formation in Cl J1449 at large in Coogan et al. 2018). Besides the ongoing star formation activity, we stress, though, that a significant fraction of the stellar mass in the proto-BCG complex has already evolved to a seemingly quiescent phase. The depletion timescale estimated in C18 for the most actively star-forming component A1 is of order 100 Myr. The overall picture is qualitatively reminiscent of hierarchical model renditions (e.g., De Lucia & Blaizot 2007; Tonini et al. 2012), though the observations presented here alone are obviously not sufficient to discuss the details of such modeling. Given the estimated (baryonic) masses of A1, H1, H4, and H5 and their relative projected distance, an approximate estimate of the merger timescale would be of order a few hundred million years (e.g., Lotz et al. 2011, the orbital timescale giving a lower limit of ∼100 Myr), with an estimated total stellar mass of the resulting BCG ∼ 3 × 10¹¹ M_⊙.

ALMA observations of the 870 μm continuum and CO(4–3) line emission in the central region of Cl J1449 have significantly improved our understanding of galaxy populations in this cluster core. Crucially, CO(4–3) follow-up secured spectroscopic confirmation of optically faint, millimeter-bright cluster members, while questioning the membership of the two brightest 870 μm sources in the field.

The 870 μm continuum and CO(4–3) line emission yield a total estimated SFR within the probed ∼0.08 Mpc² region of ∼700 ± 100 M_⊙ yr⁻¹, resulting in a projected SFR density of ∼(0.9 ± 0.1) × 10⁴ M_⊙ yr⁻¹ Mpc⁻² and a SFR volume density five orders of magnitude larger than in the field at the same redshift. The inferred lower limit (that is, not correcting for the missing SFR from the portion of the virial volume not probed by the ALMA observations) on the SFR density per halo mass is SFR/M_halo ≳ 1300 ± 400 M_⊙ yr⁻¹/10¹⁴ M_⊙, which at face value is consistent with extrapolations from lower-redshift observations predicting high SFR densities in massive (yet sub-10¹⁴ M_⊙) halos at this redshift (see discussion in Section 4.1). In spite of its relatively significant SFR density, the core of Cl J1449 seems nonetheless far from the >3000 M_⊙ yr⁻¹ observed within a similar (or smaller) clustercentric distance in the similarly massive Cl J1001+0220 at z = 2.5 (Wang et al. 2016). Three of the five most massive galaxies (H2, H4, H5; log(M/M_⊙) ∼ 11 ± 0.1) in the core of Cl J1449 are seemingly quiescent sources remaining undetected nominally at least 2σ below the MS (the other two being the MS star-forming galaxy H6 and the possibly suppressed H1; see Section 4.2, Figure 5), compared to a star-forming galaxy fraction of 80% observed at M ≥ 10¹¹ M_⊙ in the core of Cl J1001+0220 (Wang et al. 2016).

The combination with previously available HST imaging (and grism spectroscopy) critically enhances the interpretation of these observations. Although there is generally a close correspondence between the millimeter- and optical/NIR-inferred pictures of most cluster galaxies discussed here, there are particular sources that can only be really understood by comparing the two. The bright ALMA sources A1 and A2 are associated with optical counterparts otherwise deemed comparatively minor components in the HST NIR imaging. Figure 2 underlines that, if deep high-resolution color imaging were not available, these ALMA detections would appear as dust and gas significantly offset from stellar emission in the nearest optical/NIR counterpart. The HST imaging reveals instead the faint, extremely red components perfectly matching the submillimeter emission, and probably related to very recent or ongoing merging events (see further discussion in C18).

The HST and ALMA synergy also provides in this cluster another striking snapshot of the early evolution of forming BCGs (e.g., Webb et al. 2015a; Kubo et al. 2016, see Section 5), with a seemingly multiple-merger system of quiescent and highly star-forming components likely assembling the future BCG.

The results based on ALMA observations presented here extend the reach of our previous studies in Cl J1449, drawing quantitative details in a picture combining, right into the cluster core, a star formation activity approaching a thousand solar masses per year, the first massive quiescent cluster-core galaxies, and the ongoing formation of the BCG through merging of already quiescent and still vigorously star-forming components.

Largely based on observations from ALMA programs 2012.1.00885.S and 2015.1.01355.S and HST programs GO-11648 and GO-12991. We thank the anonymous referee for a constructive report improving the presentation of this work. R.T.C. acknowledges support from STFC G1687 grant ST/N504452/1. F.V. acknowledges the Villum Fonden research grant 13160 "Gas to Stars, Stars to Dust: Tracing Star Formation across Cosmic Time." A.R. acknowledges support by INAF grant PRIN-2017. N.A. is supported by the Brain Pool Program, which is funded by the Ministry of Science and ICT through the National Research Foundation of Korea (2018H1D3A2000902). A.C. acknowledges PRIN MIUR 2015 "Cosmology and Fundamental Physics: Illuminating the Dark Universe with Euclid." H.D. acknowledges financial support from the Spanish Ministry of Economy and Competitiveness (MINECO) under the 2014 Ramón y Cajal program MINECO RYC-2014-15686. This paper makes use of the following ALMA data: ADS/JAO.ALMA#2012.1.00885.S, ADS/JAO.ALMA#2015.1.01355.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. Partly based on observations made with the NASA/ESA Hubble Space Telescope, obtained from the Data Archive at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555.