The VANDELS survey: dust attenuation in star-forming galaxies at z = 3-4
Cimatti, A.; Elbaz, D.; Cullen, F.; McLure, R. J.; Dunlop, J. S.; Castellano, M.; Pentericci, L.; Fynbo, J. P. U.; Cirasuolo, M.; Koekemoer, A.; Zamorani, G.; Garilli, B.; Carnall, A. C.; Talia, M.; Bourne, N.; Marchi, F.; Khochfar, S.; Dalla Vecchia, C.
United Kingdom, Spain, Italy, Germany, France, Denmark, United States
Abstract
We present the results of a new study of dust attenuation at redshifts 3 < z < 4 based on a sample of 236 star-forming galaxies from the VANDELS spectroscopic survey. Motivated by results from the First Billion Years (FiBY) simulation project, we argue that the intrinsic spectral energy distributions (SEDs) of star-forming galaxies at these redshifts have a self-similar shape across the mass range 8.2 ≤ log (M⋆/M⊙) ≤ 10.6 probed by our sample. Using FiBY data, we construct a set of intrinsic SED templates which incorporate both detailed star formation and chemical abundance histories, and a variety of stellar population synthesis (SPS) model assumptions. With this set of intrinsic SEDs, we present a novel approach for directly recovering the shape and normalization of the dust attenuation curve. We find, across all of the intrinsic templates considered, that the average attenuation curve for star-forming galaxies at z ≃ 3.5 is similar in shape to the commonly adopted Calzetti starburst law, with an average total-to-selective attenuation ratio of RV = 4.18 ± 0.29. In contrast, we find that an average attenuation curve as steep as the SMC extinction law is strongly disfavoured. We show that the optical attenuation (AV) versus stellar mass (M⋆) relation predicted using our method is consistent with recent ALMA observations of galaxies at 2 < z < 3 in the Hubble Ultra Deep Field (HUDF), as well as empirical AV - M⋆ relations predicted by a Calzetti-like law. In fact, our results, combined with other literature data, suggest that the AV-M⋆ relation does not evolve over the redshift range 0 < z < 5, at least for galaxies with log(M⋆/M⊙) ≳ 9.5. Finally, we present tentative evidence which suggests that the attenuation curve may become steeper at lower masses log(M⋆/M⊙) ≲ 9.0.