Probing re-ionization with quasar spectra: the impact of the intrinsic Lyman α emission line shape uncertainty

Abstract

Arguably the best hope of understanding the tail end of the re-ionization of the intergalactic medium (IGM) at redshift z > 6 is through the detection and characterization of the Gunn-Peterson damping wing absorption of the IGM in bright quasar spectra. However, the use of quasar spectra to measure the IGM damping wing requires a model of the quasar's intrinsic Lyman α emission line. Here we quantify the uncertainties in the intrinsic line shapes, and how those uncertainties affect the determination of the IGM neutral fraction. We have assembled a catalogue of high-resolution Hubble Space Telescope spectra of the emission lines of unobscured low-redshift quasars, and have characterized the variance in the shapes of their lines. We then add simulated absorption from the high-redshift IGM to these quasar spectra in order to determine the corresponding uncertainties in re-ionization constraints using current and future samples of z > 6 quasar spectra. We find that, if the redshift of the Lyman α emission line is presumed to coincide with the systemic redshift determined from metal lines, the inferred IGM neutral fraction is systematically biased to low values due to a systematic blueshift of the Lyman α line relative to the metal lines. If a similar blueshift persists in quasars at z > 6, this bias strengthens previous claims of a significant neutral hydrogen fraction at z ~ 6. The bias can be reduced by including a Lyman α blueshift in the modelling procedure, or by excising wavelengths near the Lyman α line centre from the modelling. Intrinsic Lyman α line shape variations still induce significant scatter in the inferred x_IGM values. Nevertheless, this scatter still allows a robust distinction between a highly ionized (x_IGM ~ 10^-3) and a neutral (x_IGM = 1) IGM with even a few bright quasars. We conclude that if the variations of the intrinsic Lyman α emission line shapes in high-z quasars are similar to those at low-z, this variation will not limit the usefulness of quasar spectra in probing re-ionization.