Spectroscopy of bright quasars with the Hubble Space Telescope and Lyman-alpha absorption lines in the redshift range 0.5<z<1.7

We report ultraviolet spectroscopy of three bright quasars obtained with the Faint Object Spectrograph of the Hubble Space Telescope. The good quality spectra covering the range 1800-3300 Angstrom result from spectropolarimetry acquired for these targets, the interpretation of which has been published elsewhere. Objective algorithms were used to select absorption lines whose strength exceeded 4 times the rms noise in the nearby continuum, resulting in 109 significant lines for PG 1222+228, 91 significant lines for PG 1634+706, and 19 significant lines for PG 2302+029. Most of the spectral range covers the region with a high density of lines due to intervening absorbers, blueward of the Lyman-alpha emission line. In PG 1222+228, we identify about 35% of the lines as being associated with the seven metal line systems already known in this quasar. Three have seven or more metal lines identified. An additional 12% are either galactic lines or Lyman-alpha, Lyman-beta pairs with no associated metals. In PG 1634 + 706, nearly 42% of the absorption lines are identified with metal systems. Some are associated with the two metal line systems previously known in this quasar, others are associated with two newly identified C IV systems at z = 0.6540 and z = 0.9057. Another 19% are galactic lines or Lyman-alpha, Lyman-beta pairs with no associated metals. Six galactic lines are identified in the spectrum of PG 2302 + 029; no lines due to intervening absorbers could be identified. The data for PG 1222+228 and PG 1634+706 can be used to estimate the number density of Lyman-alpha absorbers in the redshift range 0.5 < z < 1.7. Above an effective rest equivalent width of 0.4 Angstrom there are 25 Lyman-alpha lines in PG 1222+228 in the wavelength range 2300-3300 Angstrom and 11 Lyman-alpha lines in PG 1634+706 in the wavelength range 1865-2650 Angstrom. We have been able to demonstrate that the identification procedure and the method of fitting lines in blended regions is unlikely to contribute systematic errors beyond the Poisson error to these numbers. The deduced number density is consistent with the number density of Lyman-alpha absorbers at zero redshift, using published data from the HST Quasar Absorption Line Key Project, indicating little or no evolution over 55% (q(0) = 0) to 70% (g(0) = 0.5) of the age of the universe. It is also consistent with an extrapolation to lower redshift of the rapid increase in number density that is seen before z approximate to 2, implying an inflection at z = 1-1.5. The rapid decline in dN/dz might be caused by an ionization effect caused by the changing comoving space density of quasars, because the decline follows a similar form to the decline in the integrated UV intensity of quasars at the hydrogen Lyman edge. The evolution in dN/dz over the range 1 < z < 3 also mirrors the change in the star formation rate in disk galaxies in the universe. It is not clear whether transition in Lyman-alpha evolution at z = 1-1.5 is due to a change in ionization state, or whether it marks a fundamental change in the intervening absorber population.