Revealing the Environmental Dependence of Molecular Gas Content in a Distant X-Ray Cluster at z = 2.51

Our JVLA observations of CLJ1001 were performed in 2015 December under the program 15B-290 (P.I.: Tao Wang). Part of the data has been already presented in W16 (including example CO(1-0) spectra), which were only used to confirm cluster members. In detail, the observations were carried out in the Ka-band with the D configuration, with an effective frequency coverage of 32.2–33.59 GHz, corresponding to z ∼ 2.43–2.58 for CO(1-0). The observations were done in excellent weather conditions with precipitable water vapor as low as 2.0 mm and wind speed 0.3–1.5 km s⁻¹, therefore the achieved system temperature (T_sys) is about or even below 40 K, while typical T_sys is about 50 K at the Ka-band in winter. The full width half power (FWHP) size of the primary beam is 13 at 32.878 GHz (z = 2.506 for CO(1-0)). We observed 3C147 for flux calibration during the full observations, and a point source J1024-0052 near our target for phase calibration during each scan loop (every ∼8 minutes). The total integration time is ∼13 hr. The data were reduced using the Common Astronomy Software Application (CASA) package (McMullin et al. 2007) with a standard pipeline. We chose 05 pixels and a spectral resolution of 30 km s⁻¹ with a natural weighting scheme for imaging. Image deconvolution was performed with a CLEAN threshold of 3σ of each cube. The resulting datacube has a synthesized beam size of ∼288 × 252 with an rms of ∼33–40 μJy beam⁻¹ per channel at the phase center.

In order to acquire a complete census of cluster (star-forming) members, we recently conducted a deep NB survey toward CLJ1001 with Subaru/MOIRCS. The NB survey employed the "CO" filter centered at 2.3 μm to identify Hα emitters at z = 2.49–2.52, combined with the already available deep K_s-band data in COSMOS from the UltraVista survey (McCracken et al. 2012; Muzzin et al. 2013; Laigle et al. 2016). With 4.4 hr of integration, we have detected 49 Hα emitters with line flux down to 1.5 × 10⁻¹⁷ erg s⁻¹ cm⁻². This corresponds to a dust-free SFR of ∼5 M_⊙ yr⁻¹ at z = 2.51 (Kennicutt 1998). Details of data reduction and star formation properties of these Hα emitters will be discussed in a forthcoming paper. Here we only use their positions to search for CO(1-0) line emissions.

We extract CO(1-0) spectra at the position of the cluster members (Hα-emitters) out to 2× FWHP of the primary beam (PB). This approach allows us to detect sources with fainter fluxes and with higher fidelity than a blind search. In total, 14 Hα-emitters are detected with a signal-to-noise ratio (S/N) > 3 (Figure 1). We measured the CO(1-0) line fluxes for each object by running a 2D Gaussian fit with CASA (IMFIT) on the velocity-integrated (moment-0) map. The velocity range used to create the moment-0 map of each object was determined so to maximize the signal-to-noise of the detection. During this process the spatial position of the targets was kept fixed based on the coordinates found from the Hubble Space Telescope (HST)/F160W (if available) or NB ancillary images, in order to minimize false detections due to noise fluctuations. Thirteen out of these 14 galaxies (except ID-14) are also covered by our Atacama Large Millimeter/submillimeter Array (ALMA) band-3 observations, and they are all detected in CO(3-2) at roughly the same velocity (Wang et al. 2018, in preparation). In the case of low S/N with CO(1-0), CO(3-2) data is combined with CO(1-0) to determine the velocity range of the line emission. For a sanity check, we have also measured directly their total fluxes in the uv-plane with GILDAS,¹⁴ a procedure that gives consistent results. The measured integrated CO(1-0) line intensities, after primary beam correction, are listed in Table 1. The 14 detections include all but 1 massive galaxy (ID-131651 in W16) with ${M}_{* }\gt {10}^{10.5}{M}_{\odot }$ within the PB, which is classified as a passive galaxy in W16 based on its rest-frame colors, suggesting that the Hα emission mostly likely originates from an (radio) active galactic nucleus (AGN), as further supported by the non-detection of CO(1-0).

Zoom In Zoom Out Reset image size

Table 1.  Physical Properties of the CO(1-0)-detected Cluster Members in CLJ1001

ID	IDa (W16)	zCO	log M*	log LIR	FWHM	${L}_{\mathrm{CO}(1-0)}^{{\prime} }$	${\alpha }_{\mathrm{CO}}$	Mgas	tdep
			(M⊙)	(L⊙)	(102 km s−1)	(1010 K km s−1 pc2)	(M⊙/(K km s−1 pc2))	(${10}^{10}{M}_{\odot }$)	(Gyr)
1	130949	2.503	11.36 ± 0.15	12.55 ± 0.14	5.0 ± 0.5	2.3 ± 0.2	4.06	9.2 ± 0.9	0.26
2	130901	2.507	11.35 ± 0.15	12.04 ± 0.21	6.8 ± 0.9	1.8 ± 0.3	4.06	7.4 ± 1.1	0.68
3	131079	2.514	11.13 ± 0.15	11.88 ± 0.20	2.8 ± 1.2	0.6 ± 0.1	4.08	2.4 ± 0.3	0.32
4	130933	2.501	11.06 ± 0.15	12.09 ± 0.16	6.9 ± 1.6	0.6 ± 0.1	4.08	2.6 ± 0.5	0.21
5	130359	2.508	11.03 ± 0.15	12.44 ± 0.13	2.4 ± 0.4	2.7 ± 0.4	4.08	11.1 ± 1.5	0.40
6	131077	2.494	10.93 ± 0.15	12.87 ± 0.13	5.5 ± 0.4	4.9 ± 0.4	4.09	20.2 ± 1.7	0.27
7	132044	2.505	10.90 ± 0.15	12.08 ± 0.18	6.8 ± 1.6	2.8 ± 0.8	4.09	11.3 ± 3.5	0.94
8	130891	2.513	10.83 ± 0.15	12.62 ± 0.13	3.4 ± 0.2	3.2 ± 0.3	4.10	13.3 ± 1.4	0.32
9	131661	2.500	10.80 ± 0.15	11.97 ± 0.19	0.9 ± 0.4	0.5 ± 0.1	4.10	1.9 ± 0.3	0.20
10	131904	2.506	10.80 ± 0.15	11.78 ± 0.25	6.9 ± 3.7	1.0 ± 0.3	4.10	4.2 ± 1.1	0.70
11	132627	2.506	10.73 ± 0.15	12.27 ± 0.16	6.0 ± 1.9	12.7 ± 4.2	4.10	52.2 ± 17.4	2.83
12	130842	2.515	10.67 ± 0.15	11.10 ± 0.20	0.9 ± 0.3	0.3 ± 0.1	4.11	1.4 ± 0.3	1.07
13	⋯	2.505	10.67 ± 0.15	12.31 ± 0.20	5.3 ± 0.8	2.0 ± 0.3	4.11	8.4 ± 1.3	0.41
14	129444	2.515	10.54 ± 0.15	11.95 ± 0.25	1.4 ± 0.3	13.4 ± 2.7	4.12	55.0 ± 11	6.17

Note. ^aIDs are from the K_s-selected catalog in Muzzin et al. (2013).

Download table as:  ASCIITypeset image

The use of CO(1-0) avoids the uncertainty in the CO excitation, and is the most extensively used way for obtaining the total molecular gas mass, $M({{\rm{H}}}_{2})$. The conversion involves the integrated CO emission intensity (${L}_{\mathrm{CO}(1-0)}^{{\prime} }$) and a conversion factor ${\alpha }_{\mathrm{CO}}$ (Bolatto et al. 2013), through $M({{\rm{H}}}_{2})={\alpha }_{\mathrm{CO}}{L}_{\mathrm{CO}}^{{\prime} }$, with the CO line luminosity ${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ derived following Solomon & Vanden Bout (2005). We determine ${\alpha }_{\mathrm{CO}}$ for the cluster galaxy sample following Genzel et al. (2015) and Tacconi et al. (2018). The same mass–metallicity relation used in Genzel et al. (2015) is also applied to determine the metallicity for the CO-detected galaxies in our sample, which is close to solar given their large stellar masses. As a result, the derived ${\alpha }_{\mathrm{CO}}$ is close to the Milky value for this sample (Table 1).

The combination of the CO line luminosity (${L}_{\mathrm{CO}(1-0)}^{{\prime} }$), tracing total molecular gas, and the total infrared (IR) luminosity (${L}_{\mathrm{IR}}$), tracing newly formed stars, provides a crucial constraint on the SFE, with ${L}_{\mathrm{IR}}$/${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ ∝ SFR/M_gas ≡ SFE. Moreover, the use of ${L}_{\mathrm{IR}}$/${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ as an approximation of SFE allows us to not have to account for the different prescriptions for the CO-to-H₂ conversion factor, enabling direct comparisons between different samples. As shown in W16 (also see, e.g., Bussmann et al. 2015), five cluster members are detected at 870 μm with ALMA, for which we derive their IR luminosities, L_IR, by fitting the full-IR SED. For the other galaxies without 870 μm detection, we derive L_IR based on their 24 μm fluxes (Muzzin et al. 2013) and 3 GHz (Smolčić et al. 2017) radio continuum by using the average IR SED templates for galaxies at z ∼ 2.5 (Schreiber et al. 2018) and far-infrared (FIR)-radio relation (Delhaize et al. 2017). We have verified this approach through comparisons of IR-SED derived and 3 GHz derived ${L}_{\mathrm{IR}}$ for four out of the five ALMA-detected sources (excluding ID-4, which is a radio AGN), which are in good agreement. Only one source (ID-12) has neither 24 μm nor 3 GHz detections, for which we derive L_IR based on its SFR estimated from extinction-corrected Hα following Kennicutt (1998). The best-estimated L_IR for cluster members are listed in Table. 1.

Figure 2 presents the comparison of ${L}_{\mathrm{IR}}$ and ${L}_{\mathrm{CO}(1-0)}^{{\prime} }$  between member galaxies in CLJ1001 and other galaxy populations in high-z clusters (z ∼ 1.5–2) and field. The CLJ1001 galaxies exhibit a large variety in their SFE as traced by ${L}_{\mathrm{IR}}$/${L}_{\mathrm{CO}(1-0)}^{{\prime} }$, including members with high, starburst-like SFE and also members with SFE even below MS-like galaxies. To examine whether this large variety in SFE is solely driven by their different star formation modes (starburst versus MS), we further show the variation of ${L}_{\mathrm{IR}}$/${L}_{\mathrm{CO}(1-0)}^{{\prime} }$ as a function of their distance to the MS, which is defined as ΔMS = SFR/SFR_MS with SFR_MS for each galaxy derived using the MS relation at z = 2.5 (Schreiber et al. 2015) at the same stellar mass (the right panel of Figure 2), which is nearly identical to the MS relation used in Tacconi et al. (2018). Consistent with field galaxies, a general trend toward increasing SFE with enhanced star formation activity (relative to the MS) is observed in CLJ1001. However, in contrast to field galaxies, a large dispersion of SFE is present for these cluster galaxies. Most prominently, a population of cluster galaxies with MS-like SFR ($| {\rm{\Delta }}\mathrm{MS}| \lt 0.5$) exhibit significantly different SFE compared to field galaxies. This suggests that the large variation in SFE for these cluster galaxies is not driven by the variation in SFR; instead, some other mechanisms, most likely related to the dense environment, may play an important role.

Zoom In Zoom Out Reset image size

In order to gain further insights into the origin of the gas and star formation properties of the cluster galaxies, we examine the relation between these properties and their positions in the cluster. Specifically, we employ the line-of-sight velocity versus clustercentric radius phase-space diagram to illustrate the relative distribution of member galaxies within the cluster. The phase-space diagram characterizes the accretion state of cluster member galaxies, which minimizes projection effects of their 2D positions with respect to the cluster center (see, e.g., Noble et al. 2013). In this diagram, galaxies that are recently accreted to the cluster tend to have large relative velocities and/or large clustercentric radius, which are offset from the central virialized region. As shown in the left panel of Figure 3, we observe a clear trend of decreasing gas content (${\mu }_{\mathrm{gas}}={M}_{\mathrm{gas}}/{M}_{* }$) with proximity to the cluster core. This is more clearly illustrated in the right panel of Figure 3, in which we plot ${\mu }_{\mathrm{gas}}$ versus ${k}_{b}=(| {\rm{\Delta }}v| /{\sigma }_{\mathrm{cluster}})\times (R/{R}_{200{\rm{c}}})$ for the cluster galaxies. The parameter k_b converts the phase-space diagram into one dimension (Noble et al. 2013). Galaxies with lower k_b are more closely bounded to the cluster, hence are likely accreted at earlier times. Figure 3 reveals clearly a trend that galaxies with high gas fraction (relative to the MS) have entered the cluster more recently than the gas-poor members. This remains true even when the normalized gas fraction, ${\mu }_{\mathrm{gas}}/{\mu }_{\mathrm{gas},{\rm{\Delta }}\mathrm{MS}}$, is adopted. While galaxies in the outskirts of the cluster exhibit a large scatter in their gas fraction (compared to field galaxies), galaxies in the cluster center (${k}_{b}\lesssim 0.1$) show exclusively a deficit of molecular gas. The transition between the gas-rich and gas-poor populations takes place at around k_b ∼ 0.1. This rapid transition may suggest that whatever environmental effects are involved, this process must be very efficient in reducing the gas content of cluster galaxies.

Zoom In Zoom Out Reset image size

We further present the variation of SFR and SFE as a function of k_b in Figure 4, showing that both SFR and SFE also varies with clustercentric radius. While galaxies in the outskirts of the cluster exhibit a large scatter in their SFR, most member galaxies in the center tend to fall below the MS. On the other hand, these galaxies in the center show a significant enhancement in their SFE compared to those in the outskirts and field galaxies. Despite their low gas fraction, their normal or suppressed SFR suggests that their enhanced SFE is mainly caused by their deficit of molecular gas (instead of an enhanced SFR). This indicates that the suppression on the SFR from the dense environment is likely delayed compared to that on the gas content. This high SFE ensures that most of these galaxies will likely consume all of their gas in a short timescale. Their gas depletion time (t_dep = 1/SFE) is around t_dep ∼ 0.4 Gyr, a factor of two shorter than field galaxies with the same ΔMS (Figure 2). This timescale is comparable to the cluster dynamical time (approximated by the crossing time), ${t}_{\mathrm{dyn}}\sim {R}_{200{\rm{c}}}/{\sigma }_{\mathrm{cluster}}\sim 0.5\,\mathrm{Gyr}$, suggesting that most of these cluster galaxies may consume all of their gas within a single orbit around the cluster center, and form a passive cluster core by z ∼ 2.

Zoom In Zoom Out Reset image size

It should be noted that because the sensitivity of the CO(1-0) observation decreases toward larger radius from the cluster center (phase center), only gas-rich systems can be detected at large radii. However, as shown in Figure 1, our CO(1-0) detected sample comprise a mass-complete sample of cluster member galaxies; i.e., we are not missing massive star-forming yet gas-poor galaxies up to 2× FWHM. Hence, our result is not affected by this observational bias.