A Theory for the Variation of Dust Attenuation Laws in Galaxies

TL;DR

This paper develops a physical model explaining the variation in dust attenuation laws in galaxies by combining cosmological simulations with radiative transfer, revealing how geometry and star populations influence attenuation features.

Contribution

It introduces a model showing how geometry and star populations cause variations in dust attenuation laws, despite using a constant dust extinction curve.

Findings

Attenuation law slopes depend on star-dust geometry complexities.

UV bump strength varies with unobscured O and B stars.

Dispersion in attenuation laws decreases with increasing redshift.

Abstract

In this paper, we provide a physical model for the origin of variations in the shapes and bump strengths of dust attenuation laws in galaxies by combining a large suite of cosmological "zoom-in" galaxy formation simulations with 3D Monte Carlo dust radiative transfer calculations. We model galaxies over 3 orders of magnitude in stellar mass, ranging from Milky Way like systems through massive galaxies at high-redshift. Critically, for these calculations we employ a constant underlying dust extinction law in all cases, and examine how the role of geometry and radiative transfer effects impact the resultant attenuation curves. Our main results follow. Despite our usage of a constant dust extinction curve, we find dramatic variations in the derived attenuation laws. The slopes of normalized attenuation laws depend primarily on the complexities of star-dust geometry. Increasing fractions of unobscured young stars flatten normalized curves, while increasing fractions of unobscured old stars steepen curves. Similar to the slopes of our model attenuation laws, we find dramatic variation in the 2175 Angstrom ultraviolet (UV) bump strength, including a subset of curves with little to no bump. These bump strengths are primarily influenced by the fraction of unobscured O and B stars in our model, with the impact of scattered light having only a secondary effect. Taken together, these results lead to a natural relationship between the attenuation curve slope and 2175 Angstrom bump strength. Finally, we apply these results to a 25 Mpc/h box cosmological hydrodynamic simulation in order to model the expected dispersion in attenuation laws at integer redshifts from z=0-6. A significant dispersion is expected at low redshifts, and decreases toward z=6. We provide tabulated results for the best fit median attenuation curve at all redshifts.