Satellites of Milky Way- and M31-like galaxies with TNG50: quenched fractions, gas content, and star formation histories

Abstract

We analyse the quenched fractions, gas content, and star formation histories of ∼1200 satellite galaxies with M* ≥ 5 × 106 M⊙ around 198 Milky Way- (MW) and Andromeda-like (M31) hosts in TNG50, the highest-resolution simulation of IllustrisTNG. Satellite quenched fractions are larger for smaller masses, for smaller distances to their host galaxy, and in the more massive M31-like compared to MW-like hosts. As satellites cross their host’s virial radius, their gas content drops: most satellites within 300 kpc lack detectable gas reservoirs at z = 0, unless they are massive like the Magellanic Clouds and M32. Nevertheless, their stellar assembly exhibits a large degree of diversity. On average, the cumulative star formation histories are more extended for brighter, more massive satellites with a later infall, and for those in less massive hosts. Based on these relationships, we can even infer infall periods for observed MW and M31 dwarfs: e.g. 0–4 Gyr ago for the Magellanic Clouds and Leo I, 4 − 8 and 0–2 Gyr ago for M32 and IC 10, respectively. Ram pressure stripping (in combination with tidal stripping) deprives TNG50 satellites of their gas reservoirs and ultimately quenches their star formation, even though only a few per cent of the present-day satellites around the 198 TNG50 MW/M31-like hosts appear as jellyfish. The typical time since quenching for currently quenched TNG50 satellites is $6.9\substack{+2.5\\-3.3}$ Gyr ago. The TNG50 results are consistent with the quenched fractions and stellar assembly of observed MW and M31 satellites, however, satellites of the SAGA survey with M* ∼ 108 − 9 M⊙ exhibit lower quenched fractions than TNG50 and other, observed analogues.

The Role of Mass and Environment on Satellite Distributions around Milky Way Analogs in the Romulus25 Simulation

“…As the high-mass end of Figure 8(b) is strongly affected by this surface brightness cut, it seems that the preferentially quenched satellites below this threshold are more common in higher-mass hosts, which is consistent with Figure 8(a) implying a larger number of quenched galaxies in this regime. Our results agree with Engler et al (2023) who, using the TNG50 run from the IllustrisTNG simulations, found that massive hosts exhibit systematically larger satellite quenched fractions. Further, Engler et al (2023) found no difference in quenching between isolated and paired analogs when considering satellites within 300 kpc of their host (see Section 5.3 for discussion).…”

“…Our results agree with Engler et al (2023) who, using the TNG50 run from the IllustrisTNG simulations, found that massive hosts exhibit systematically larger satellite quenched fractions. Further, Engler et al (2023) found no difference in quenching between isolated and paired analogs when considering satellites within 300 kpc of their host (see Section 5.3 for discussion).…”

“…This, again, could be alluding to a strong environmental effect on satellite quenching in the highly isolated regime that we do not quite capture in ROMULUS25. This result is at odds with the aforementioned study of TNG50 by Engler et al (2023), as well as the study of FIRE-2 by Samuel et al (2022), both of whom find no difference in quenching between isolated and paired analogs. However, the simulations studied in Samuel et al (2022) are zoom-in simulations, and thus would not capture the large-scale isolation in which we begin to see differences between paired and isolated hosts.…”

“…Fainter surface brightnesses correlate with more quenching at a fixed luminosity in ARTEMIS, and thus bias the SAGA results if true. On the other hand, Engler et al (2023) found that a surface brightness cut could not bring the TNG50 satellite quenched fractions fully into agreement with SAGA, though it did bring the simulation and observational results more into line. Engler et al (2023) were able to use TNG50, a 50 Mpc on-a-side uniform cosmological volume, to study a larger sample of Milky Way analogs and look for statistical trends.…”

“…On the other hand, Engler et al (2023) found that a surface brightness cut could not bring the TNG50 satellite quenched fractions fully into agreement with SAGA, though it did bring the simulation and observational results more into line. Engler et al (2023) were able to use TNG50, a 50 Mpc on-a-side uniform cosmological volume, to study a larger sample of Milky Way analogs and look for statistical trends. In this work, we use ROMULUS25, a 25 Mpc on-a-side uniform cosmological volume with comparable resolution to TNG50, to study similar trends.…”