Evaluating and Improving Semi-analytic modelling of Dust in Galaxies based on Radiative Transfer Calculations

TL;DR

This paper couples semi-analytic galaxy formation models with radiative transfer calculations to develop improved dust attenuation recipes, enhancing the accuracy of galaxy spectral energy distribution predictions across redshifts.

Contribution

It introduces fitting formulae for dust optical depth based on galaxy properties, validated against detailed radiative transfer calculations, improving upon previous analytic prescriptions.

Findings

Fitting formulae predict V-band optical depth with <0.4 dex scatter.

Attenuation curve shapes vary significantly due to geometry, not captured by simple models.

New recipes outperform standard prescriptions, especially at high redshift.

Abstract

The treatment of dust attenuation is crucial in order to compare the predictions of galaxy formation models with multiwavelength observations. Most past studies have either used simple analytic prescriptions or else full radiative transfer (RT) calculations. Here, we couple star formation histories and morphologies predicted by the semi-analytic galaxy formation model MORGANA with RT calculations from the spectrophotometric and dust code GRASIL to create a library of galaxy SEDs from the UV/optical through the far Infrared, and compare the predictions of the RT calculations with analytic prescriptions. We consider a low and high redshift sample, as well as an additional library constructed with empirical, non-cosmological star formation histories and simple (pure bulge or disc) morphologies. Based on these libraries, we derive fitting formulae for the effective dust optical depth as a function of galaxy physical properties such as metallicity, gas mass, and radius. We show that such fitting formulae can predict the V-band optical depth with a scatter smaller than 0.4 dex for both the low and high redshift samples, but that there is a large galaxy-to-galaxy scatter in the shapes of attenuation curves, probably due to geometrical variations, which our simple recipe does not capture well. However, our new recipe provides a better approximation to the GRASIL results at optical wavelength than standard analytic prescriptions from the literature, particularly at high redshift.