It has been established that in local star-forming galaxies, nebular emission suffers additional dust attenuation than stellar emission (extra attenuation). Several studies at higher redshift have leaded to contradictory results about the existence of extra attenuation in distant star-forming galaxies. Extra attenuation is linked to how and where star formation occurs, which is of primeval importance to understand general galaxy properties, since we mainly observe the light coming from the youngest and more massive stars at high-redshift.
We use a sample of 149 z~2 UV-selected galaxies, spectroscopically confirmed, with large broad-band photometry coverage and with Halpha (or [OIII]5007A) flux measurements to determine if there is a need for extra attenuation. We use a spectral energy distribution fitting code which accounts for nebular emission (continuum+lines) to reproduce all available observed quantities (broad-band photometry, emission line fluxes, and infrared luminosity when available) and derive physical properties. We find that our spectral energy distribution fitting code is able to reproduce all the quantities for galaxies with no or little extinction, while we need to apply an extra attenuation (E(B-V)nebular=1.98xE(B-V)stellar) to reproduce emission lines of galaxies suffering a significant amount of dust attenuation. We also find a tight correlation between star formation rate and the extra attenuation which could explain the contradictory results from previous studies. We finally discuss the implications of extra attenuation at high redshift (z > 3).