Quiescent Galaxies - Why Do Galaxies Stop Producing Stars?

Quenching can refer within the astronomy community to either of two separate processes: star formation activity failing to continue or a galaxy being maintained as quiescent throughout a long period of time even though star formation materials are readily available (e.g. Man & Belli 2018). A quiescent galaxy is a galaxy that remains quenched over a period of time. Cosmological simulations require some form of galaxy quenching in order to obtain accurate models of galaxy populations. However, the physical mechanisms behind quenching cannot be too tightly constrained within the simulation because we do not know enough about what causes a galaxy to stop producing stars (Hahn et al. 2019). Whether quenching arises from a myriad of mechanisms or a singular phenomena is a key question in galaxy research today (Faucher-Giguère 2018) With this study we trace the evolution of large populations of quenched galaxies to achieve a tighter constraint on the theories behind quenching mechanisms. A quiescent galaxy is a galaxy that remains quenched over a period of time.

Broadly, the current theories for central galaxy quenching include gas not accreting, gas not cooling, cold gas being quickly used, gas being removed, and inefficiency in cold gas being used for star formation. Many of these quenching mechanisms are driven by feedback. For example, active galactic nuclei (AGN) feedback plays a crucial role in gas being removed, in rapid consumption of cold gas, in the inability of cold gas to be used efficiently, and in gas not cooling (e.g. Man & Belli 2018). Past studies have shown that different quenching mechanisms in cosmological simulations have provided similar results, which proves troubling when attempting to constrain quenching to fewer mechanisms (Weinberger et al. 2017). Quenching does not have to be caused by only a single physical mechanism - the phenomena can be due to multiple quenching processes all at once. Cosmological simulations require some form of galaxy quenching in order to obtain accurate models of galaxy populations.

Implementing large-scale dynamics of the universe with a built-in subgrid of small-scale physics is a common method of obtaining a robust galaxy formation model while compensating for the unavoidable low resolution of large volume cosmological simulations. If a simulation is large enough to encompass large scale universal structures (i.e. galaxy clusters, dark matter structure) there will not be enough computing power available to resolve the interstellar medium (ISM), supernovae, and star formation (SFR). However, these small-scale baryonic processes constrain the physics seen on the scale of dark matter halos, and thus must be included in order to create a simulation accurate to observations (Faucher-Giguèere 2018). If not included, simulations produce a population of galaxies much more massive and more dense than seen observationally (e.g. Pillepich et al. 2018a). The IllustrisTNG simulation, which we use in this study, includes these processes using a subgrid physics model.

SELECTING QUIESCENT GALAXIES

To select quiescent galaxies, we first found a shallow power law relationship between stellar mass and SFR in log space. We define quiescent galaxies as falling one dex below the line of best fit to the SFS.

Figure 1: All points that fall below the red line in the figure above are quenched galaxies.

The relation between the stellar mass and star formation rate (SFR) of a galaxy is called the galazy star-forming sequence (SFS). This study utilizes the IllustrisTNG-100 simulation (Marinacci et al. 2018; Naiman et al. 2018; Nelson et al. 2018; Pillepich et al. 2018b; Springel et al. 2018; Nelson et al. 2019 ) data to look at the star formation sequence (SFS) of galaxy populations at various redshifts and to select interesting quenched galaxies to study. The SFS is the relationship between the total stellar mass of a galaxy and its star formation rate normalized by mass (sSFR). IllustrisTNG contains data in 100 snapshots with a box of 100 kpc on a side large at redshifts ranging from 0 to 12. For ease of comparison to previous work, we chose to initially study the SFS in redshifts of 0.1, 0.5, 0.76, 1.21, 1.60, 2.0, 2.44, and 2.73.

Quenched galaxies are selected for study by first looking at the overall galaxy population trends throughout cosmological time. Selected galaxies must be central galaxies (the largest galaxy in its local field) and have a stellar mass greater than 3x10^10.8 M_☉ . We binned the data by stellar mass and took the mean, median, and standard deviation of each bin. We also fit a line to the star formation sequences in log space for the galaxies at each studied redshift (Figure 1). From this line of best fit, we selected our quiescent population by defining a quenched galaxy selected as more than one dex below the mean at its stellar mass.

RESULTS

Figure 2: The mean and median overlap, suggesting a robust power law relationship exists for the star formation sequence.

When looking at the star formation sequence in a singular redshift, it can be noted that the mean and median SFR of this selected galaxy population in each studied redshift overlaps until the noted 10^10.8 M_☉ cutoff point (Figure 2). It is seen that the standard deviation jumps in value at this cutoff mass and that the mean and median separate in value. The overlap in mean and median indicates we robustly found the relationship between the SFR and the stellar mass for galaxies with a nonzero star formation rate. The lower stellar mass end of this figure is not included, because it is below the resolution limit of the simulation. The relation between SFR and stellar mass exists throughout cosmic time.

Figure 3: Mean SFR by stellar mass for varying redshifts. Mean is consistent until the M_*= 10^10.8 M_☉ cutoff.

However, zooming out to look at the star formation sequence in multiple redshifts and doing direct comparisons of the approximated means of galaxies binned by stellar mass and at different redshifts, it is seen that the mean star formation rate of galaxies increases systematically with increasing redshift (Figure 3). This is to be expected of the total galaxy population - at early times in the universe, there was more star formation activity occurring than at present day. The relationship begins to break down at approximately 10^10.8 M_☉. This mass limit is where the galaxy becomes too massive for the star formation rate to maintain its size and thus quenching occurs.

Figure 4: Quescient fraction of galaxy population binned by stellar mass in varying redshift.

Quenched galaxies in our analysis are defined as having sSFR values less than one dex below the line of best fit for the SFS. The quiescent fraction (the number of quiescent galaxies out of the total number of galaxies in each bin) increases with decreasing redshift (Figure 4). At 10^10.8 M_☉, the quiescent population spikes and continues to rise as stellar mass increases. This figure only includes bins with at least 100 galaxies, because with any less galaxies the bin statistics are not statistically significant due to small sample size. Generally most quiescent galaxies occur at late redshifts and at high stellar masses.

Figure 5: Fraction of quiescent galaxy population that remains non-star-forming at different redshift selections.

We traced the entire quenched galaxy population from z = 0.76 and z = 2.76 forward in time to record the percentages of galaxies that reach and maintain an effectively zero SFR (Figure 5). Out of the total quenched population, 70% reach and stay at an effectively zero star formation rate (SFR) until the present day when selected at z = 0.76, and 30% reach and stay at an effectively zero SFR when selected at z = 2.73. The fates of quiescent galaxies appear to be widely different depending on when they are observed to be quiescent.