Searching for dust reddening in SDSS spectra with damped Lyman α systems

Abstract

We searched for evidence of reddening of background SDSS QSO spectra due to dust in intervening DLA systems. We utilise the Data Releases 5 and 7 to arrive at sample sizes of 475 (DR5) and 676 (DR7) absorbers, based on two different published lists of SDSS DLAs. Both samples span roughly the redshift range of 2.2 < z_abs < 5.2, with a mean of z~3.0, and the majority of the DLAs (75%) below z=3.3. We construct geometric mean spectra in the absorber restframes ranging from 1240 to ~2800 A, and composite spectra of samples matching the 'DLA' QSOs in i band magnitude and emission redshift z_em, but without absorption lines. By comparing the slopes of these composite spectra with their matched counterparts, we find no sign of reddening in the ensemble of the absorbers from these samples. Owing to both the unprecedently large sizes of the DLA samples themselves and the non-DLA SDSS QSO sample, from which we can draw our matching spectra, we can place very tight limits for this non-detection (<E(B-V)> =-0.0013+-0.0025 (DR5) and <E(B-V)> =-0.0017+-0.0022 (DR7). Interestingly, when applying our technique to the samples of York et. al. (2006), vandenBerk et al. (2008) (intervening and intrinsic MgII absorbers) and the smaller DLA-subsample and pool of comparison QSOs of Vladilo et al. (2008), we do recover their results, i.e. detect the same amount of reddening as these authors do. Furthermore, we have tested whether subsamples of our large sample in categories involving the absorbers (HI column densities, presence or absence of accompanying metal absorption, absorber redshift) or the background quasars (emission redshift, brightness) do reveal dust extinction, but found no trends. These results are at odds with both detections of dust reddening from previous studies, and also with expectations from observations of high-redshift galaxies. (abridged)