SCUBADive II: Searching for $z>4$ Dust-Obscured Galaxies via F150W-Dropouts in COSMOS-Web

Sinclaire M. Manning, Jed McKinney, Katherine E. Whitaker, Arianna S. Long, Olivia R. Cooper, Caitlin M. Casey, Rafael C. Arango-Toro, Jaclyn B. Champagne, Nicole E. Drakos, Andreas L. Faisst, Maximilien Franco, Ghassem Gozaliasl, Santosh Harish, Hossein Hatamnia, Christopher C. Hayward, Michaela Hirschmann, Jeyhan S. Kartaltepe, Anton M. Koekemoer, Daizhong Liu, Georgios E. Magdis, Henry Joy McCracken, Jason Rhodes, Brant E. Robertson, Margherita Talia, Francesco Valentino, John R. Weaver, Jorge A. Zavala

First listed 2025-05-14 | Last updated 2025-10-03

Abstract

The relative fraction of obscured galaxies at $z>4$ compared to lower redshifts remains highly uncertain as accurate bookkeeping of the dust-obscured component proves difficult. We address this shortcoming with SCUBADive, a compilation of the JWST counterparts of (sub-)millimeter galaxies in COSMOS-Web, in order to further analyze the distribution and properties of massive dust-obscured galaxies at early times. In this paper, we present a subset of SCUBADive, focusing on 60 ``dark'' galaxies that dropout at 1.5\micron. Motivated by JWST observations of AzTECC71, a far-infrared bright F150W-dropout with $z_{\rm phot}=5.7^{+0.8}_{-0.7}$, we complete a systematic search of F150W-dropouts with SCUBA-2 and ALMA detections to find more candidate high redshift dusty galaxies. Within our subsample, 16 are most similar to AzTECC71 due to fainter F444W magnitudes ($>24$\,mag) and lack of counterparts in COSMOS2020. Despite high star formation rates ($\langle$SFR$\rangle=450^{+920}_{-320}$\,\mdot\,yr$^{-1}$) and large stellar masses ($\langle$log$_{10}$(\mstar)$\rangle=11.2^{+0.5}_{-0.6}$\,\mdot) on average, these galaxies may not be particularly extreme for their presumed epochs according to offsets from the main sequence. We find that heavily obscured galaxies, which would be missed by pre-JWST optical imaging campaigns, comprise $\gtrsim20$\% of galaxies across mass bins and potentially contribute up to 60\% at the very high mass end (log$_{10}$(\mstar/\mdot)$>11.5$) of the $z>4$ stellar mass function.

Short digest

SCUBADive II systematically compiles 60 F150W-dropout, far-IR–bright galaxies in COSMOS-Web by cross-matching JWST non-detections at 1.5 μm with SCUBA-2 and ALMA counterparts, motivated by the benchmark AzTECC71. A core subset of 16 AzTECC71-like sources shows faint F444W (>24 mag) and no COSMOS2020 IDs. SED fits give ⟨SFR⟩=550+500−360 M⊙ yr⁻1 and ⟨log10(M⋆)⟩=11.2+0.5−0.4, with offsets indicating they are not unusually elevated above the star-forming main sequence. Crucially, such heavily obscured systems constitute roughly 20% across mass bins and could account for up to 60% of the z>4 high-mass end (log10 M⋆>11), quantifying a major, previously missed component of the early galaxy census.

Key figures to inspect

• Selection and dropout diagnostic: figure showing F150W non-detections versus detections at longer NIRCam bands (e.g., F277W/F444W), illustrating the F150W-dropout criterion and the cut that isolates the 16 AzTECC71-like sources with F444W>24 mag.
• Postage-stamp mosaics with ALMA/SCUBA-2 overlays: NIRCam panels (F115W/F150W non-detections contrasted with F277W/F356W/F444W) plus ALMA contours to verify counterpart associations, astrometry, and the FIR–NIR alignment; include the AzTECC71 benchmark for reference.
• Representative SED fits: plots showing the photometric-z posteriors and IR–to–NIR SEDs that yield high SFRs and M⋆ for individual F150W-dropouts, highlighting the faint rest-optical and strong FIR peaks that drive the z>4 solutions.
• Main-sequence offsets: Δlog(SFR) versus M⋆ at z>4 for the sample, demonstrating that most sources lie near the star-forming main sequence rather than as extreme outliers.
• Impact on the stellar mass function: fraction of heavily obscured galaxies versus M⋆ showing ~20% across bins and rising to ~60% above log10(M⋆/M⊙)>11, indicating how much mass density is missed without F150W-dropout selection.