Mergers, Starbursts, and Quenching in the Simba Simulation

Abstract

We use the Simba cosmological galaxy formation simulation to investigate the relationship between major mergers ( < ∼ 4:1), starbursts, and galaxy quenching. Mergers are identified via sudden jumps in stellar mass M * well above that expected from in situ star formation, while quenching is defined as going from specific star formation rate sSFR> t −1 H to sSFR< 0.2t −1 H , where t H is the Hubble time. At z ≈ 0 − 3, mergers show ∼ ×2 − 3 higher SFR than a mass-matched sample of star-forming galaxies, but globally represent < ∼ 1% of the cosmic SF budget. At low masses, the increase in SFR in mergers is mostly attributed to an increase in the H 2 content, but for M * > ∼ 10 10.5 M mergers also show an elevated star formation efficiency suggesting denser gas within merging galaxies. The merger rate for star-forming galaxies shows a rapid increase with redshift ∝ (1 + z) 3.5 , but the quenching rate evolves much more slowly, ∝ (1 + z) 0.9 ; there are insufficient mergers to explain the quenching rate at z < ∼ 1.5. Simba first quenches galaxies at z > ∼ 3, with a number density in good agreement with observations. The quenching timescales τ q are strongly bimodal, with 'slow' quenchings (τ q ∼ 0.1t H ) dominating overall, but 'fast' quenchings (τ q ∼ 0.01t H ) dominating in M * ∼ 10 10 − 10 10.5 M galaxies, likely induced by Simba's jet-mode black hole feedback. The delay time distribution between mergers and quenching events suggests no physical connection to either fast or slow quenching. Hence, Simba predicts that major mergers induce starbursts, but are unrelated to quenching in either fast or slow mode.

CANDELS Meets GSWLC: Evolution of the Relationship between Morphology and Star Formation Since z = 2

“…We find that Mergers contribute 5% to the SFR budget for all masses and redshifts (see Figure 9). Similar results have been obtained for major mergers (typi- cally with a mass ratio < 4:1) using a variety of selection methods for a wide range of redshifts (Robaina et al 2009;De Propris et al 2014;Lofthouse et al 2017a), as well as simulations (Kaviraj et al 2015;Rodríguez Montero et al 2019). Inferring the mass ratio of merger progenitors from the visible tidal features is very challenging and has not been explored extensively in the literature so far.…”

Reproducing sub-millimetre galaxy number counts with cosmological hydrodynamic simulations

“…At z ∼ 2 − 3, some massive galaxies satisfy this and fall off the main sequence, while others do not and end up vigorously forming stars, appearing at the top end of the main sequence. We note that Simba agrees well with the number density of galaxies that lie > ∼ 1 dex below the main sequence at these epochs (Rodríguez Montero et al 2019), though it fails to sufficiently quench those galaxies since it does not match the counts lying > ∼ 2 dex below the main sequence (Merloni et al, submitted; Finkelstein et al, submitted). So it appears that Simba's AGN feedback is approximately striking the correct balance between quenching sufficient galaxies at z ∼ 2, while not quenching too many massive galaxies which would eliminate the SMG population entirely.…”

“…Simba was tuned primarily to match the evolution of the overall stellar mass function and the stellar mass-black hole mass relation (Davé et al 2019). The model reproduces a number of key observables at both low and high redshift that do not rely on this tuning, and are bona fide predictions of the model, including SFR functions, the cosmic SFR density, passive galaxy number densities (Rodríguez Montero et al 2019), galaxy sizes and star formation rate profiles Appleby et al (2020), central supermassive black hole properties (Thomas et al 2019), damped Lyman-α abundances (Hassan et al 2020), star formation histories (Mamon et al 2020), reionisation-epoch UV luminosity function (Wu et al 2019), and the low-redshift Lyα absorption (Christiansen et al 2019). Importantly for this study, Simba reproduces the bright-end CO luminosity function at z = 2 (Davé et al 2020), which has been difficult to match in other recent models (see Riechers et al 2019;Popping et al 2019).…”