In the past 10 years, more than 60 quasars at z > 5.7 have been found. Assuming the
current standard cosmology, these redshifts correspond to less than 1 Gyr after the
big-bang. Thus high-redshift quasars provide direct probes of the Universe that was just
1 Gyr old or earlier. At such early times, an abundance ratio of iron to alpha-elements
(such as O, Ne, Mg) [Fe/alpha] is expected to be smaller than the current value; this is
because the Type-Ia supernova, a main factory of iron in the Universe, had never
occurred at that time. Under the assumption that the FeII/MgII emission-line flux ratio
traces the [Fe/Mg] abundance ratio, FeII and MgII emission-lines in quasars have been
measured by many researchers in the past ~20 years; however, the measured FeII/MgII flux
ratio has a large scattering and shows no clear evolution even at z ~ 6. A doubt is then
cast on the assumption that the FeII/MgII flux ratio is a first-order proxy of the
[Fe/Mg] abundance ratio. In this talk, we report our studies on FeII and MgII
emission-lines by investigating ~17,000 SDSS quasars at z~1. We have found that
FeII/MgII flux ratio shows a strong correlation with the Eddington ratio, which seems
not due to difference in abundance. We have performed photoionization simulations with
the code CLOUDY, and found that the gas density in the broad line region (BLR)
presumably varies with the Eddington ratio. Taking these results, we will discuss the
possibility of constraining [Fe/Mg] abundance ratio of the BLR cloud from the measured
FeII and MgII emission lines.