2013A&A...551A.124H

Gamma-ray burst (GRB) spectra globally appear non-thermal, but recent observations of a few bursts with Fermi GBM have confirmed previous indications from BATSE of the presence of an underlying thermal component. Photospheric emission is indeed expected when the relativistic outflow emerging from the central engine becomes transparent to its own radiation, with a quasi-blackbody spectrum in absence of additional sub-photospheric dissipation. However, its intensity strongly depends on the acceleration mechanism - thermal or magnetic - of the flow. We aim to compute the thermal and non-thermal emissions (light curves and spectra) produced by an outflow with a variable Lorentz factor, where the power {dot}(E)_iso injected at the origin is partially thermal (fraction ε_th≤1) and partially magnetic (fraction 1-ε_th). The thermal emission is produced at the photosphere, and the non-thermal emission in the optically thin regime. Apart from the value of ε_th, we want to test how the other model parameters affect the observed ratio of the thermal to non-thermal emission. We followed the adiabatic cooling of the flow from the origin to the photosphere and computed the emitted radiation, which is a sum of modified black bodies at different temperatures (as the temperature strongly depends on the Lorentz factor of each shell at transparency). If the non-thermal emission comes from internal shocks, it is obtained from a multi-shell model where a fraction of the energy dissipated in shell collision is transferred to electrons and radiated via the synchrotron mechanism. If, conversely, the non-thermal emission originates in magnetic reconnection, the lack of any detailed theory for this process forced us to use a very simple parametrisation to estimate the emitted spectrum. If the non-thermal emission is made by internal shocks, we self-consistently obtained the light curves and spectra of the thermal and non-thermal components for any distribution of the Lorentz factor in the flow. If the non-thermal emission results from magnetic reconnection we were unable to produce a light curve and could only compare the respective non-thermal and thermal spectra. In the different considered cases, we varied the model parameters to see when the thermal component in the light curve and/or spectrum is likely to show up or, on the contrary, to be hidden. We finally compared our results to the proposed evidence for the presence of a thermal component in GRB spectra. Focussing on GRB 090902B and GRB10072B, we showed how these observations can be used to constrain the nature and acceleration mechanism of GRBoutflows.