Probing the Star Formation Main Sequence down to 10$^{7} M_\odot$ at $1 < z < 9$

TL;DR

The paper addresses how the star-formation main sequence (MS) behaves for very low-mass galaxies across $1 \leq z \leq 9$ by constructing a mass- and SFR-complete sample through a joint analysis of pre-JWST and JWST surveys. It employs a unified Dense Basis framework with nonparametric SFHs to re-fit $M_\star$ and SFR consistently across all surveys, enabling robust MS measurements down to $M_\star \sim 10^{7.6}\,M_\odot$ at $z\sim1$ and $M_\star \sim 10^{8.8}\,M_\odot$ at $z\sim9$. The results show an intrinsic MS slope of $\beta \sim 0.7$–$0.8$ up to $z\sim5$, with a possible steepening at low masses around $M_\star \sim 10^{9.5}\,M_\odot$ that appears independent of redshift, and an intrinsic scatter of $\epsilon \sim 0.2$–$0.3$ dex. These findings imply evolving star-formation efficiency and gas content in low-mass galaxies, highlight a disk-formation threshold, and underscore the importance of complete samples and flexible SFH modeling for interpreting the MS across cosmic time.

Abstract

The Main Sequence of Star-Forming Galaxies (SFGMS or MS) is a fundamental scaling relation that provides a global framework for studying galaxy formation and evolution, as well as insight into the complex star formation histories (SFHs) of individual galaxies. In this work, we combine large-area pre-JWST surveys (COSMOS2020, CANDELS), which probe high-$M_\star$ sources (${>10^9\,M_\odot}$), with SHARDS/CANDELS FAINT and JWST data from CANUCS, CEERS, JADES, and UNCOVER, to obtain a high-$z$, star formation rate (SFR) and stellar mass ($M_\star$) complete sample spanning both high- and low-$M_\star$ regimes. Completeness in both $M_\star$ and SFR is key to avoiding biases introduced by low-mass, highly star-forming objects. Our combined data set is 80% complete down to $10^{7.6}\,M_\odot$ at $z\sim1$ ($10^{8.8}\,M_\odot$ at $z\sim9$). The overall intrinsic MS slope (based on the SFR$_{100}$ and $M_\star$ derived with Dense Basis and nonparametric SFHs) shows little evolution up to $z\sim5$, with values $\sim0.7 - 0.8$. The slope in the low-$M_\star$ regime becomes steeper than that in the high-$M_\star$ end at least up to $z\sim5$, but the strength of this change is highly dependent on the assumptions made on the symmetry of the uncertainties in $M_\star$ and SFR. If real, the steepening suggests reduced star formation efficiency or declining gas content with decreasing $M_\star$. The transition between the low-$M_\star$ regime and the canonical MS occurs around $10^{9.5}\,M_\odot$, independent of $z$. This critical value may coincide with the assembly of galaxies' disks, which can provide a mechanism for self-regulation that stabilizes them against feedback. The intrinsic scatter is compatible with canonical estimates, also at low-$M_\star$, ranging from $0.2-0.3$ dex. This is indicative of rapid variations in star formation being averaged out over $\lesssim100$ Myr.

Figures (9)

• Figure 1: Cartoon depicting different challenges that can arise when trying to fit the MS with no continuous and complete coverage. Current JWST surveys are very deep, but limited in size, selecting mainly low-mass galaxies. Pre-JWST surveys are not deep enough to reach such low $M_\star$. Offsets and/or changes in the slopes can make the connection between the two regimes hard to quantify if we do not consider the ensemble of both pre-JWST and JWST surveys.
• Figure 2: Stellar mass histograms in different redshift intervals showing the distribution of the CANDELS (red), COSMOS2020 (maroon), SHARDS/CANDELS FAINT (green), JWST/CEERS (yellow), JADES (fuchsia), CANUCS (navy), and UNCOVER (blue) surveys. Each histogram is normalized such that the total area under the curve equals 1. The solid gray histograms in the panels underneath show the distribution of the galaxies that entered our final selection, combining the previous surveys (see Sect. \ref{['sec:properties']}). The 16th, 50th, and 84th percentiles for each distribution are displayed as segments, color-coded following the histograms (black for our final sample). Vertical dashed lines depict the 80% mass-completeness limit of our sample (see Sect. \ref{['sec:completeness']}).
• Figure 3: Attenuation histograms in different redshift intervals showing the distribution of the galaxies within the distinct surveys, following the same color code as in Fig. \ref{['fig:mass']}.
• Figure 4: Star formation rate histograms in different redshift intervals showing the distribution of the galaxies within the distinct surveys, following the same color code as in Fig. \ref{['fig:mass']}. Vertical dashed lines highlight our SFR limits (see Sect. \ref{['sec:completeness']}).
• Figure 5: Cartoon depicting the variations induced in the $M_\star$ (left) and SFR (right) completeness limits by changes in the shape of the completeness function (black line). We plot some hypothetical SFR and $M_\star$ limits as dashed lines. The hatched green region encloses the SFR and $M_\star$ values above both completeness limits. The hatched blue region highlights the locus of galaxies that would be detected thanks to their high SFR, but that lie below the $M_\star$ limit. The hatched orange region contains galaxies above the $M_\star$ limit that would be hard to detect because of their low SFR. The white region is occupied by galaxies that are below the $M_\star$ and SFR limits. In yellow (pink), we show how the curve should change in order to increase (decrease) both the $M_\star$ and SFR limits independently. Histograms above each panel exemplify how changes in $M_\star$ or SFR completeness would affect our final sample. Bins are color-coded according to the regions they would occupy in the log SFR$-$log $M_\star$ plane. We display two mirrored histograms, showing galaxies that lie above (blue and green) and below (white and red) the SFR limit. The quiescent or dormant histogram is centered at a lower $M_\star$, as bursty star formation rules at lower $M_\star$ (i.e., quiescence is not permanent and these galaxies should be included in an MS analysis). At higher $M_\star$ galaxies that fall below the MS are more prone to remain quiescent, and thus would be discarded through $UVJ$ and sSFR cuts. Vertical and horizontal lines are color-coded following the bottom panels. The last two histograms exemplify how objects would move from the top to the mirrored histograms, and vice versa, as we change the SFR limit.