2015A&A...579A..74J

We use gamma-ray burst (GRB) afterglow spectra observed with the VLT/X-Shooter spectrograph to measure rest-frame extinction in GRB lines-of-sight by modelling the broadband near-infrared (NIR) to X-ray afterglow spectral energy distributions (SEDs). Our sample consists of nine Swift GRBs, of which eight belong to the long-duration and one to the short-duration class. Dust is modelled using the average extinction curves of the Milky Way and the two Magellanic Clouds. We derive the rest-frame extinction of the entire sample, which fall in the range 0≲A_V≲1.2. Moreover, the SMC extinction curve is the preferred extinction curve template for the majority of our sample, a result that is in agreement with those commonly observed in GRB lines of sights. In one analysed case (GRB120119A), the common extinction curve templates fail to reproduce the observed extinction. To illustrate the advantage of using the high-quality, X-Shooter afterglow SEDs over the photometric SEDs, we repeat the modelling using the broadband SEDs with the NIR-to-UV photometric measurements instead of the spectra. The main result is that the spectroscopic data, thanks to a combination of excellent resolution and coverage of the blue part of the SED, are more successful in constraining extinction curves and therefore dust properties in GRB hosts with respect to photometric measurements. In all cases but one the extinction curve of one template is preferred over the others. We show that the modelled values of the extinction A_V and the spectral slope, obtained through spectroscopic and photometric SED analysis, can differ significantly for individual events, though no apparent trend in the differences is observed. Finally we stress that, regardless of the resolution of the optical-to-NIR data, the SED modelling gives reliable results only when the fit is performed on a SED covering a broader spectral region (in our case extending to X-rays).