High energy afterglows and flares from Gamma-Ray Burst by Inverse Compton emission

We perform a detailed study of Inverse Compton (IC) emission for a fireball undergoing External Shock (ES) in either a uniform or a wind-like interstellar medium, and assess the relative importance of IC and synchrotron emission. We determine the primary model parameters driving the IC to synchrotron emission ratio in the case of a short duration central engine. We then investigate the case of ES by a long duration central engine, or Delayed External Shock (DES), a model that can account for some of the flares observed in Gamma-Ray Bursts (GRBs) X-ray light curves at late times. We present model predictions, in particular in terms of GeV vs X-ray behaviour, and compare them with other models proposed to explain the origin of flares. We find that if most of the emission occurs when the fireball is in the fast cooling regime then a substantial GeV emission is expected both for a short (standard ES) and a long (DES) duration central engine activity. In particular, in the context of standard ES we are able to account for the delayed emission observed in GRB940217. In the case of DES we find that IC scattering of X-ray flare photons can produce high energy flares in the GeV band, that can be detected by the Large Area Telescope aboard GLAST. The detectability of high energy flares improves with the burst kinetic energy E: about 30 % of Swift GRBs showing flares in their X-ray light curve have sufficiently large kinetic energy so that the expected high flares can be detected by GLAST. One important prediction of the DES model is the simultaneity between low and high energy flares. To test this simultaneity, the peak energies of both flares need to fall below or within the observational bands. We predict that X-ray flares with peak energy of ∼ 10 eV produce high energy flares with peak energy around hundred of MeV-GeV. Observations by Swift and GLAST then, can test the predicted simultaneity, thus helping to discriminate between different models.