The Connection between Dusty Star-Forming Galaxies and the First Massive Quenched Galaxies

TL;DR

This work uses a re-calibrated L-Galaxies semi-analytic model to investigate the connection between dusty star-forming galaxies (DSFGs) and high-redshift massive quiescent galaxies (MQs). The authors show that the majority of MQs at z>2 were DSFGs at some point, with a strong correlation between current stellar mass and the historical maximum sub-mm flux $S_{870, m max}$, and with peak sub-mm emission typically occurring before the MQs’ quenching. Quenching is driven primarily by AGN feedback triggered by an early major merger that also fuels rapid SMBH growth, while subsequent mass growth for MQs proceeds mainly through dry mergers; most DSFGs, in contrast, grow secularly and quench later. Not all DSFGs become MQs, and the brightest DSFGs quench earlier, underscoring diverse evolutionary pathways within the DSFG population. The results provide a coherent, testable framework aligning with JWST findings and offering concrete predictions for MQ progenitor properties and quenching timescales.

Abstract

High-redshift (z > 2) massive quiescent (MQ) galaxies provide an opportunity to probe the key physical processes driving the fuelling and quenching of star formation in the early Universe. Observational evidence suggests a possible evolutionary link between MQs and dusty star-forming galaxies (DSFGs; or submillimetre galaxies), another extreme high-redshift population. However, galaxy formation models have historically struggled to reproduce these populations - especially simultaneously - limiting our understanding of their formation and connection, particularly in light of recent JWST findings. In previous work, we presented a re-calibrated version of the L-Galaxies semi-analytic model that provides an improved match to observationally-inferred number densities of both DSFG and MQ populations. In this work, we use this new model to investigate the progenitors of MQs at z > 2 and the physical mechanisms that lead to their quenching. We find that most MQs at z > 2 were sub-millimetre-bright ($S_{870}$ > 1 mJy) at some point in their cosmic past. The stellar mass of MQs is strongly correlated with the maximum submillimetre flux density attained over their history, and this relation appears to be independent of redshift. However, only a minority of high-redshift DSFGs evolve into MQs by z = 2. The key distinction between typical DSFGs and MQ progenitors lies in their merger histories: MQ progenitors experience an early major merger that triggers a brief, intense starburst and rapid black hole growth, depleting their cold gas reservoirs. In our model, AGN feedback subsequently prevents further gas cooling, resulting in quenching. In contrast, the broader DSFG population remains sub-millimetre-bright, with star formation proceeding primarily via secular processes, becoming quenched later.

Figures (11)

• Figure 1: The star formation rate (SFR)–stellar mass ($M_{\star}$) relation at $z = 2$, 3, and 4 is shown in the first, second, and third panels, respectively, using the re-calibrated version of L-Galaxies from yo25a. The coloured dots represent bright DSFGs with $S_{870} \geq 1 \,\rm{mJy}$, colour-coded by their $S_{870}$ flux density (see colour-bar), derived using the scaling relations presented by rachel23. The red dashed lines indicate the thresholds in specific star formation rate (sSFR) and stellar mass used to separate quenched and massive galaxies, respectively, following the definition of carnall20. The pink line and colored area correspond to the 'main sequence' of galaxies in our model and its $1\sigma$ dispersion. Overall, bright DSFGs are located in the massive, star-forming region of the diagram and lie on, or just above, the 'main sequence'. As expected, the massive, quenched region is less populated at higher redshifts.
• Figure 2: The evolution of the $S_{870}$ flux density of massive quiescent galaxies (MQs) selected at $z = 2$, 3, and 4 is shown in the first, second, and third panels, respectively. At each selected redshift, MQs are divided into four stellar mass bins of width $0.4\,\rm{dex}$, starting from a lower limit of $\log(M_{\star}/\rm{M_{\odot}}) = 10.6$. The colored regions and lines represent the $16^{\rm{th}}$, $50^{\rm{th}}$, and $84^{\rm{th}}$ percentiles of MQs within each stellar mass bin. Overall, the $S_{870}$ evolution of MQs exhibits a gradual increase with decreasing redshift, reaching a peak before undergoing a rapid decline at later times. Notably, the most massive MQs consistently display the highest $S_{870}$ peak values.
• Figure 3: The distribution of the maximum $S_{870}$ flux density reached by each modelled galaxy across its entire formation history, $S_{870,\mathrm{max}}$ (first row), and when it occurred (second row; redshift distribution), for massive quiescent galaxies (MQs) selected at $z = 2$, $3$, and $4$ (from left to right, respectively). The minimum $S_{870,\mathrm{max}}$ reached by any of the MQs is indicated by the vertical dashed black line.
• Figure 4: The maximum value of $S_{870}$, $S_{870,\mathrm{max}}$, reached across the formation history of massive quiescent galaxies (MQs), as a function of stellar mass at redshifts $z = 2$, $3$, and $4$ (from left to right). The data points are coloured according to the redshift at which $S_{870,\mathrm{max}}$ occurred, $z(S_{870,\mathrm{max}})$. A strong correlation is found between the stellar mass of MQs and their $S_{870,\mathrm{max}}$, with a dispersion that also correlates with $z(S_{870,\mathrm{max}})$. At fixed stellar mass, MQs that reached $S_{870,\mathrm{max}}$ at earlier times tended to exhibit brighter $S_{870}$ flux densities.
• Figure 5: The fraction of massive quiescent galaxy (selected at $z = 2$, $3$, and $4$; first, second, and third panels, respectively) progenitors that exceed $S_{870}$ flux densities of $1\,\rm{mJy}$ (dark blue), $3\,\rm{mJy}$ (purple), and $5\,\rm{mJy}$ (orange), versus redshift. At the redshift at which the distributions peak, a significant fraction of MQ progenitors exceed $S_{870}=1\,\rm{mJy}$. For example, $50\%$, $66\%$, and $67\%$ of MQs selected at $z=2$, $3$, and $4$, respectively, exceeded $S_{870}=1\,\mathrm{mJy}$ at $z \sim 3.4$, $4.3$, and $5.5$. These numbers are lower for higher $S_{870}$ thresholds: at their peaks, $\sim 3\%$ of MQs selected at $z=2$, $\sim 6\%$ at $z=3$, and $\sim 5\%$ at $z=4$ exceed $3\,\rm{mJy}$.