IllustrisTNG and S2COSMOS: possible conflicts in the evolution of neutral gas and dust

Abstract

We investigate the evolution in galactic dust mass over cosmic time through i) empirically derived dust masses using stacked submillimetre fluxes at 850μm in the COSMOS field, and ii) dust masses derived using a robust post-processing method on the results from the cosmological hydrodynamical simulation IllustrisTNG. We effectively perform a ‘self-calibration’ of the dust mass absorption coefficient by forcing the model and observations to agree at low redshift and then compare the evolution shown by the observations with that predicted by the model. We create dust mass functions (DMFs) based on the IllustrisTNG simulations from 0 <z < 0.5 and compare these with previously observed DMFs. We find a lack of evolution in the DMFs derived from the simulations, in conflict with the rapid evolution seen in empirically derived estimates of the low redshift DMF. Furthermore, we observe a strong evolution in the observed mean ratio of dust mass to stellar mass of galaxies over the redshift range 0 <z < 5, whereas the corresponding dust masses from IllustrisTNG show relatively little evolution, even after splitting the sample into satellites and centrals. The large discrepancy between the strong observed evolution and the weak evolution predicted by IllustrisTNG plus post-processing may be explained by either strong cosmic evolution in the properties of the dust grains or limitations in the model. In the latter case, the limitation may be connected to previous claims that the neutral gas content of galaxies does not evolve fast enough in IllustrisTNG.

The dust-continuum size of TNG50 galaxies at $z=1-5$: a comparison with the distribution of stellar light, stars, dust and H$_2$

“…Other simulations found that they require a variable DTM in order to reproduce the UV luminosity of low-mass high-redshift galaxies (Lagos et al 2019;Vogelsberger et al 2020). This assumption is different from the classical one of a constant DTM of ∼ 0.3 often adopted in simulation-based works (see for example Cochrane et al 2019;Schulz et al 2020;Liang et al 2020;Millard et al 2021).…”

The dust-continuum size of TNG50 galaxies at z = 1–5: a comparison with the distribution of stellar light, stars, dust, and H2

“…Other simulations found that they require a variable DTM in order to reproduce the UV luminosity of low-mass high-redshift galaxies (Lagos et al 2019 ;Vogelsberger et al 2020 ). This assumption is different from the classical one of a constant DTM of ∼0.3 often adopted in simulation-based works (see for example Cochrane et al 2019 ;Liang et al 2020 ;Schulz et al 2020 ;Millard et al 2021 ).…”

The Evolving Interstellar Medium of Star-forming Galaxies, as Traced by Stardust*

“…Finally, we compliment our analysis by comparing our DMF to the theoretical predictions of the HAGN and Il-lustrisTNG simulations (Millard et al 2020). In order to produce a HAGN DMF, we define a simulated sample following the procedure described above, bin the galaxies in 0.4 dex intervals of M dust , and normalise by the volume of the simulation (4 × (142) 3 Mpc 3 ).…”

“…In order to produce a HAGN DMF, we define a simulated sample following the procedure described above, bin the galaxies in 0.4 dex intervals of M dust , and normalise by the volume of the simulation (4 × (142) 3 Mpc 3 ). For the IllustrisTNG simulation, Millard et al (2020) consider multiple TNG100 snapshots in a box size of 106 Mpc per side, comparable to the HAGN simulated subset presented earlier. The TNG-DMF is constructed through the M dust values of the simulated galaxies, derived in post-processing through a fixed dust-to-metals ratio of 0.5.…”