[Paper Review] Must Starforming Galaxies Rapidly Get Denser Before They Quench?

Louis E. Abramson, Takahiro Morishita|arXiv (Cornell University)|Aug 26, 2016

Galaxies: Formation, Evolution, Phenomena3 citations

TL;DR

This study challenges the hypothesis that star-forming galaxies must rapidly increase in density (via compaction) to quench star formation. Using deep Hubble data and empirical analyses, it finds no evidence for a density threshold triggering quenching; instead, galaxies evolve at constant surface density, with denser systems aging faster due to intrinsic formation timescales. The key result is that quenching can be explained by gas depletion alone, without requiring compaction-triggered feedback.

ABSTRACT

Using the deepest data yet obtained, we find no evidence preferring compaction-triggered quenching---where rapid increases in galaxy density truncate star formation---over a null hypothesis in which galaxies age at constant surface density ($Σ_e\equiv M_*/2πr_{e}^{2}$). Results from two fully empirical analyses and one quenching-free model calculation support this claim at all $z\leq3$: (1) Qualitatively, galaxies' mean $U-V$ colors at $6.5\lesssim\logΣ_e/{ m M_\odot}\,{ m kpc}^{-2}\lesssim10$ have reddened at rates/times correlated with $Σ_e$, implying that there is no density threshold at which galaxies turn red but that $Σ_e$ sets the pace of maturation; (2) Quantitatively, the abundance of $\log M_*/{ m M_\odot}\geq9.4$ red galaxies never exceeds that of the total population a quenching time earlier at any $Σ_e$, implying that galaxies need not transit from low to high densities before quenching; (3) Applying $d\log r_{e}/dt =1/2\,d\log M_*/dt$ to a suite of lognormal star formation histories reproduces the evolution of the size--mass relation at $\log M_*\geq10$. All results are consistent with evolutionary rates being set ab initio by global densities, with denser objects evolving faster than less-dense ones towards a terminal quiescence induced by gas depletion or other $\sim$Hubble-timescale phenomena. Unless stellar ages demand otherwise, observed $Σ_e$ thresholds need not bear any physical relation to quenching beyond this intrinsic density--formation epoch correlation, adding to Lilly & Carollo's arguments to that effect.

Motivation & Objective

• To test whether star-forming galaxies must undergo rapid compaction (increased surface density) before quenching.
• To determine if observed correlations between high density and quiescence reflect causality (compaction) or correlation (intrinsic evolution speed).
• To assess whether a model assuming constant surface density throughout a galaxy's star-forming phase can reproduce observed size–mass and color–density relations.
• To evaluate whether the observed abundance of massive red galaxies at high redshift requires a pre-quenching density increase or can be explained by formation-time-driven evolution.
• To reconcile discrepancies in galaxy size distributions between models and observations without invoking compaction.

Proposed method

• Analyzes deep Hubble Space Telescope data from the XDF, HLF, HFF, and GLASS surveys to measure galaxy stellar masses, effective radii, and U-V colors at $ z \leq 3 $.
• Empirically tracks the evolution of $ U-V $ color as a function of surface stellar mass density $ \Sigma_e \equiv M_\ast / (2\pi r_e^2) $, using UVJ criteria for quiescent vs. star-forming classification.
• Compares the observed abundance of massive red galaxies ($ \log M_\ast \geq 9.4 $) at different $ \Sigma_e $ and redshifts to the predicted abundance if quenching occurred $ \sim $ 1 Gyr earlier at constant $ \Sigma_e $.
• Applies a lognormal star formation history model with $ d\log r_e/dt = \frac{1}{2} d\log M_\ast/dt $ to simulate size–mass evolution under constant $ \Sigma_e $, matching observed size–mass relations at $ \log M_\ast \geq 10 $.
• Modifies boundary conditions in the G13 model by biasing $ r_{e,0} $ of passive galaxies at $ z=0 $ downward by 0.2 dex to improve agreement with observed sizes, without altering the core $ \Sigma_e $-conservation assumption.
• Performs statistical checks using 20 re-realizations of volume normalization and $ \chi^2 $ tests to assess model–data consistency across redshift bins.

Experimental results

Research questions

• RQ1 Is there a physical threshold in surface stellar mass density $ \Sigma_e $ at which star-forming galaxies are triggered to quench?
• RQ2 Can the observed correlation between high $ \Sigma_e $ and quiescence be explained by galaxies evolving at constant $ \Sigma_e $, with denser systems aging faster due to earlier formation?
• RQ3 Does the abundance of massive red galaxies at high redshift exceed what would be expected if quenching occurred 1 Gyr earlier at constant $ \Sigma_e $, implying a need for pre-quenching compaction?
• RQ4 Can the size–mass relation of massive galaxies at $ \log M_\ast \geq 10 $ be reproduced by a model assuming constant $ \Sigma_e $ growth and lognormal SFR histories?
• RQ5 Are discrepancies in observed vs. predicted sizes of high-redshift passive galaxies due to model boundary conditions or evidence for compaction?

Key findings

• Galaxies at all $ \Sigma_e \gtrsim 10^{6.5} \, M_\odot \, \text{kpc}^{-2} $ show reddening in $ U-V $ color correlated with $ \Sigma_e $, but no evidence of a sharp transition or threshold, indicating no density-triggered quenching.
• The observed abundance of $ \log M_\ast \geq 9.4 $ red galaxies never exceeds the predicted abundance if quenching occurred 1 Gyr earlier at constant $ \Sigma_e $, implying no need for pre-quenching density increases.
• The model assuming $ d\log r_e/dt = \frac{1}{2} d\log M_\ast/dt $ and constant $ \Sigma_e $ reproduces the observed size–mass relation for $ \log M_\ast \geq 10 $ galaxies at $ z \leq 3 $, with reasonable $ \chi^2 $ values.
• At $ z \sim 1.2 $–3, the model underproduces the densest $ \sim $5% of galaxies by a factor of $ \sim $3, but this is not strongly significant and may be resolved by adjusting boundary conditions.
• A simple modification to the G13 model—biasing $ r_{e,0} $ of passive galaxies at $ z=0 $ downward by 0.2 dex—eliminates size discrepancies for $ \log M_\ast \sim 10.7 $ galaxies at $ z < 3 $, without requiring compaction.
• The results are consistent with quenching being driven by Hubble-timescale processes like gas exhaustion, with denser galaxies quenching earlier not due to compaction, but due to earlier formation epochs.