A deep ALMA Band 3 survey of HDFS/MUSE3D: Survey description and initial results

TL;DR

This study presents a deep ALMA Band 3 survey of a $1\times1$ arcmin$^2$ field in the HDFS/MUSE3D, targeting CO transitions to probe molecular gas in galaxy groups at $z>1$ with long baselines that reach sub-arcsecond scales. The authors detail field selection, spectral tuning to cover CO(2-1) and CO(5-4) for MUSE-identified groups, and report a continuum rms around $1.4$–$1.7$ $\mu$Jy/beam with an angular resolution near $0.13''$–$0.15''$, plus a maximum recoverable scale of $~1$–$2''$. They detect CO(2-1) emission from six lines associated with the $z_{spec}=1.284$ group (and a serendipitous ALMA-114), deriving interstellar gas masses of $\sim2\times10^9$ to $9\times10^{10}\,M_\odot$ and revealing diverse morphologies and dynamical states, including offsets with rest-frame UV emission. Through a $1/V_{\rm max}$ analysis, they estimate the molecular gas mass function and cosmic molecular gas density in group environments, finding results broadly consistent with field surveys but probing deeper in molecular mass, highlighting that $\Omega_{\rm H2}$ at $z\sim1.3$ remains uncertain and potentially comparable between group and field environments. Overall, the work demonstrates the feasibility and scientific value of ultra-deep, high-resolution mm surveys to characterize ISM in galaxy groups and informs the evolution of the cosmic H$_2$ density across cosmic time.

Abstract

(abridged) After more than 10yr of ALMA operations, the community interest in conducting deep, extra-galactic, millimetre surveys resulted in varying strategic compromises between areal size and map depth to survey the sky. The current bias leans towards a galaxy population found in the field or towards rich star-bursty proto-cluster groups, both tendentiously surveyed at coarse spatial resolutions. Here, we describe a deep 3mm ALMA survey in long-baselines on a 1x1arcmin2 region in the Hubble Deep Field South, also covered by the Multi Unit Spectroscopic Explorer (MUSE) in order to assess resolved molecular gas properties in galaxies in group environments at z>1. ALMA observations comprising a 4-pointing mosaic with a single Band3 (3mm) spectral tuning were conducted to cover CO transitions from different groups identified by MUSE. This work consists in a total effective time on source of 61h in configurations with up to 15km baselines. The final data-set yields an angular resolution of 0.15"-0.2" (imaging weights dependent) and maximum recoverable scales of 1"-2". The final continuum map reaches a sensitivity of rms~2uJy/beam, allowing the detection of three sources at 3mm (only one showing multi-wavelength counterparts). Moreover, we detect six line emitters associated with CO J=2-1 at zspec=1.284, one of them previously undetected by MUSE and none detected in 3mm continuum. The inter-stellar medium gas masses range from ~2E9 to ~9E10 Msun (adopting alphaCO=4Msun/(K.km/s.pc2), including Helium). Overall, this galaxy group is quite diverse with no two galaxies alike, some showing clear physical offsets with respect to Hubble imaging tracing rest-frame ultra-violet emission. We also derive cosmic molecular gas mass densities using this sample as a reference for group environments, and we find that these yield comparable densities as the galaxy population found in field environments.

Figures (14)

• Figure 1: Comparison of the ALMA coverage (red iso-contours at primary beam attenuation levels of 0.8, 0.6, 0.4, and 0.2) with those of HDFS (WFPC2/F814W; background grey map) and MUSE3D ($1\times1\,$arcmin$^2$ blue region). Blue stars and circles indicate the positions the members of the groups at $z_{\rm spec}=1.284$ and 4.699, respectively (note that two of the high-redshift members are too close to each other to distinguish their two circles). The HDFS imaging was aligned with Gaia-DR3 reference Gaia23 making use of three stars in the field (black diamonds) at RA, Dec = 22:32:50.513, -60:34:00.94 (used for the MUSE Slow Guiding System); 22:32:56.999, -60:34:05.82 (brightest star in the field); and 22:32:50.513, -60:32:18.83. The image is oriented such that North is up, and West is right.
• Figure 2: The spectral tuning used to conduct the survey was especially chosen to target the transitions CO (2-1) and (5-4) towards the two galaxy groups identified by the MUSE 3D survey at $z=1.284$ and $4.699$, with 9 and 6 reported members, respectively (vertical lines correspond to each member). For the $z=4.699$ group, the [CI] (1-0) transition is also covered.
• Figure 3: Sensitivity (noise rms) as a function of integration time as measured in a sensitivity-ordered sub-sample of 47 EBs with a fixed range of baseline length. Upper plot shows the measured image rms (blue squares) and the theoretical rms expected from the radiometric equation and ALMA sensitivity calculator (red shaded region) versus time on source (TOS). The cyan data-point with error-bar shows the median rms and the standard deviation of the per-EB continuum map rms values. The green solid curve shows the expected rms decrease from this median data-point with a simple scaling by the square root of exposure time. Lower plot shows the same information normalized by the green solid curve in the upper plot.
• Figure 4: Assessment of the quality of the uv coverage of our combined data using the tools developed by Petry24. (Left) Observed and expected 1-D Baseline Lengths Distributions (BLDs), i.e., histogram of the baseline lengths of the visibilities for the representative frequency channel (87 GHz) of the dataset (blue) and the corresponding expected histograms for the ideal shape to achieve the most Gaussian PSF given an angular resolution range $\pm20\%$ around the nominal value of $\sim$0.13 (green and grey histograms). (Right) Plot of the ratio of observed and expected BLD in 2-D, i.e., also showing baseline orientation. A value of 1.0 indicates that observation and expectation exactly agree. Larger values indicate an over-exposure, smaller values an under-exposure.
• Figure 5: CO (2-1) emission towards B15-10. The cube used for the analysis, was imaged with natural weighting and with a smoothing of 4 channels (i.e., resulting in a spectral resolution of $\sim90\,$km/s). We present the spectrum of the detection in the top panel, while the velocity-integrated flux map (moment-0, M0) and the velocity map (moment-1, M1) are shown in the bottom left and right panels, respectively. Both maps have the same size (i.e., 3.4"-wide or $\sim$29 kpc). The moment-0 contours are overlaid on both maps (black, solid contours indicate levels at $3\times\sqrt(2)^{n}~\times~$ rms, where $n=0,\,1,\,...$, while dashed, white contours indicate $-3\sigma$). The maps are centred at the position reported by Bacon15. The top panel with the spectrum displays a filled spectrum within which the flux has been integrated and the moments have been obtained. The red vertical dashed line at zero-velocity assumes the light-weighted centre frequency, while the red dotted vertical line shows the expected frequency assuming the MUSE-derived redshift. The horizontal shaded grey region shows the $\pm1$ times the std. A thick line shows the single-component Gaussian best-fit when using the raw spectral-resolution ($\sim24\,$km/s) cube, while a dashed line that when using the 4-channel smoothed cube.