2025MNRAS.537.1518C

Over the last decade, the Atacama large millimeter array has revealed massive, dusty star-forming galaxies at $z\gtrsim 5$, and the JWST is primed to uncover even more information about them. These observations need dust evolution theory to provide context and are excellent benchmarks to test this theory. Here, we investigate the evolution of galactic dust budget at cosmic dawn using a suite of cosmological zoom-in simulations of moderately massive, high-redshift ($M_{*}\gtrsim 10^9\, \, {\rm M}_{\odot }$; $z\gtrsim 5$) galaxies from the Feedback in Realistic Environments project, the highest resolution ($m_{\rm b} \approx 7100\, \, {\rm M}_{\odot }$) of such simulations to date. Our simulations incorporate a dust evolution model that accounts for the dominant sources of dust production, growth and destruction and follows the evolution of specific dust species, allowing it to replicate a wide range of present-day observations. We find, similar to other theoretical works, that dust growth via gas-dust accretion is the dominant producer of dust mass for these massive, $z\gtrsim 5$ galaxies. However, our fiducial model produces $M_{\rm dust}$ that fall ${\gtrsim }1$ dex below observations at any given $M_{*}$ (typical uncertainties are ${\sim }1$ dex), which we attribute to reduced accretion efficiencies caused by a combination of low galactic metallicities and extremely bursty star formation. Modest enhancements (i.e. within observational/theoretical uncertainties) to accretion and SNe II dust creation raise $M_{\rm dust}$ by ${\lesssim }1$ dex, but this still falls below observations which assume $T_{\rm dust}\sim 25$ K. One possibility is that inferred dust masses for $z\gtrsim 4$ galaxies are overestimated, and recent observational/analytical works that find $T_{\rm dust}\sim 50$ K along with metallicity constraints tentatively support this.