JWST PRIMER: A deep JWST study of all ALMA-detected galaxies in PRIMER COSMOS -- dust-obscured star-formation history back to z $\simeq$ 7

TL;DR

This work combines JWST PRIMER COSMOS near-/mid-infrared imaging with an archival ALMA COSMOS sample to study dust-obscured star formation back to $z\sim7$. The authors secure IR counterparts for 128 ALMA-detected galaxies, obtain redshifts (spectroscopic for 52% and robust photometric redshifts for the rest), and derive SFRs and stellar masses, finding most sources to be massive and star-forming, heavily obscured by dust. They construct a dust-dominated cosmic star-formation rate density ($\rho_{\rm SFR,IR}$) history, correct for ALMA incompleteness using the field’s massive galaxies, and show that UV-visible star formation dominates beyond $z\sim4$, though dust-enshrouded activity may still contribute up to ~20% of $\rho_{\rm SFR}$ by $z\sim8$ and ~5% by $z\sim10$. The results provide a more complete and robust view of dusty star formation across cosmic time and highlight the need for deep, complete ALMA surveys to refine the high-redshift dusty galaxy census.

Abstract

We use the deep NIRCam and MIRI imaging from the JWST PRIMER survey to study the properties of (sub)mm sources detected by ALMA in the centre of the COSMOS field, with the aim of better constraining the history of dust-enshrouded star formation. The wealth of ALMA data in this field enabled us to isolate a robust sample of 128 (sub)mm sources within the 175 sq. arcmin of the PRIMER COSMOS survey footprint, spanning two decades in (sub)mm flux density. The JWST imaging is deep/red enough to reveal secure galaxy counterparts for all of these sources. Moreover, 52% of the galaxies have spectroscopic redshifts, enabling us to refine the photo-zs for the remaining galaxies. Armed with this robust redshift information, we calculate the star-formation rates (SFR) and stellar masses of all 128 ALMA-detected galaxies, and place them in the context of other galaxies in the field. We find that the vast majority of star formation is dust-enshrouded in the ALMA-detected galaxies, with SFR ranging from ~1000 down to ~20 solar masses per year. We also find that virtually all (126/128) have high stellar masses, at all redshifts, with log(M/Msun) > 10. The unusually high quality of our sample enables us to make a robust estimate of the contribution of the ALMA-detected galaxies to cosmic star-formation rate density from z = 2 out to z = 7. Finally, to correct for the fact that the deep ALMA pointings cover < 20% of the PRIMER COSMOS area, we use our knowledge of all other massive galaxies in the field to produce a completeness-corrected estimate of dust-enshrouded star-formation rate density over cosmic time. This confirms that UV-visible star formation dominates at z > 4, but also indicates that dust-enshrouded star formation likely still made a significant contribution at higher redshifts: extrapolation of our results suggest a ~20% contribution at z = 8, and potentially still ~5% at z = 10.

Figures (21)

• Figure 1: The complete 175.3 arcmin$^2$ coverage provided by the JWST PRIMER imaging within the COSMOS field. The NIRCam-only coverage (dark grey) is 66.0 arcmin$^2$, and the MIRI-only coverage (light grey) is 33.1 arcmin$^2$, with 76.2 arcmin$^2$ of overlapping imaging provided by both instruments (grey). The positions of the 128 ALMA sources in our final sample within this footprint (as specified in Section \ref{['sec:construction']}) are indicated by the red circles. The near/mid-infrared counterparts of all of these ALMA sources are securely detected by the NIRCam and/or MIRI imaging provided by PRIMER, enabling a complete study of their redshift distribution and physical properties.
• Figure 2: The distribution of 870$\rm{\mu}$m flux density for the 128 ALMA sources in our final sample (red). For sources where the only (or highest S/N) detection in the A$^3$COSMOS catalogue was obtained at a wavelength other than 870${\rm \mu}$m, the flux density was converted to a common observed wavelength of 870$\rm{\mu}$m assuming that the (sub)mm spectral energy distribution has a form $\propto \nu^3$ (see text). The 17 sources detected by AS2COSMOS (light grey) and the 7 sources detected at 2 mm by Ex-MORA (dark grey; 6 in common with A2SCOSMOS) included in our sample are over-plotted on the histogram. Not only is our ALMA sample $\sim 10$ times larger than previous (sub)mm samples studied within the same footprint, but it also bridges the luminosity gap between the brightest sources (e.g., those uncovered with SCUBA-2) and the fainter (sub)mm detections resulting from targeted ALMA follow-up of distant optical/infrared selected galaxies 2022MNRAS.515.3126I.
• Figure 3: The distribution of the angular separation between the position of each ALMA source in the PRIMER COSMOS footprint and its nearest near-infrared neighbour in the JWST (red) or UltraVISTA (black) imaging. Both distributions peak at an angular separation $< 0.2$ arcsec, reflecting the high positional accuracy of all catalogues. The first local minimum of the ALMA-NIRCam cross-matching distribution occurs at 0.25 arcsec (red vertical dashed line) while that for the ALMA-UltraVISTA cross-matching distribution appears at a somewhat larger radius of 0.65 arcsec (black vertical dashed line), as expected due to the limitations of ground-based seeing. These minima were used to define the search radius for near-infrared counterparts to the ALMA sources in the JWST and UltraVISTA catalogues respectively.
• Figure 4: Image stamps in all the available JWST, UltraVISTA and HST bands for 4 of the sources in our sample (ID 095, ID 053, ID 007, and ID 034). For each of the 4 sources, 16 stamps are shown, ranked in order of decreasing wavelength with filter names marked at the top of each panel, and with the position of the highest-S/N ALMA detection marked by the red cross. ALMA contours are illustrated in the longest wavelength JWST image, F770W, starting from 3.5 $\sigma$ with 1.5-$\sigma$ spacing for ID 095 and 3-$\sigma$ spacing for the rest. These illustrate different but also common features of the typical galaxies in the sample. ID 095, which lies at a redshift of only $z=0.726$ is detected with HST at optical wavelengths (albeit only just at F435W). It is also securely detected in all UltraVISTA near-infrared bands, but the vastly improved angular resolution provided by NIRCam is key to revealing its spiral disk morphology. ID 053, with $z=1.552$ is dark on HST F435W to HST 814W images, but has a consistent disk-like morphology across all infrared bands. ID 007, at $z=2.086$ appears to have an HST detection but the longer-wavelength imaging reveals that this does not correspond to the centre of the galaxy; it is either a blue clump or a separate object, and the true mass-dominant galaxy is not revealed until wavelengths longward of 2 $\mu$m. Finally, ID 034 is a higher-redshift ' HST-dark' galaxy, with $z_{\rm phot} \simeq 3.55$, and is basically not detected at all until 2 $\mu$m. Despite these differences caused by a mix of redshift and the degree of dust attenuation, all 4 of the example objects shown here, and in fact all of the galaxies in our sample are detected (and resolved) with relative ease in the longest wavelength NIRCam and/or MIRI 7.7$\mu$m imaging.
• Figure 5: An example RGB image of one of the ALMA-detected galaxies in our sample (ID 100), created from the JWST PSF-homogenised NIRCam imaging: F444W (red), F277W (green), and F115W (blue). The position of the ALMA detection is marked by the white cross. While there is no ambiguity over the correct near-infrared counterpart, a nearby very blue object/component can be seen $\simeq 1$ arsec to the east. Such a source is too distant to contaminate the NIRCam photometry, but could potentially contaminate ground-based photometry, especially at optical (i.e., rest-frame UV) wavelengths. A total of 26 sources were revisited and subjected to manual (rather than automated) photometry to ensure reliable colours and total flux densities. RGB images of all sources in the sample are presented in Appendix \ref{['fig:stamp1']}.