2025MNRAS.536.2433P

Estimating physical quantities such as the star formation rate, stellar mass, and active galactic nucleus (AGN) fraction of galaxies is a key step in understanding galaxy formation and evolution. In order to estimate the uncertainties in the predicted values for these quantities, in this paper we explore the impact of adopting four different AGN torus models in fitting the multiwavelength spectral energy distributions (SED) of galaxies. We also explore the impact of adopting two different geometries for the host, a spheroidal geometry, more appropriate for late-stage mergers, and a disc geometry, more appropriate for galaxies forming stars with secular processes. We use optical to submillimetre photometry from the Herschel Extragalactic Legacy Project (HELP) and utilize a Markov chain Monte Carlo SED-fitting code. We use exclusively radiative transfer models for the AGN torus as well as for the starburst and host galaxy. We concentrate on a sample of 200 galaxies at $z\approx 2$, selected in the ELAIS-N1 field. All galaxies have a detection at 250 $\mu \mathrm{ m}$ which ensures the presence of a starburst. We find that the stellar mass and star formation rate of the galaxies can be robustly estimated by the SED fitting but the AGN fraction depends very much on the adopted torus model. We also find that the vast majority of the galaxies in our sample are better fitted by a spheroidal geometry and lie above the main sequence. Our method predicts systematically higher SFR and lower stellar mass than the popular energy balance method CIGALE.