1 Introduction

Deuterium (D or $^2{\rm H}$) is the only element entirely produced by nuclear reactions in the first minutes after Big Bang (Wagoner et al. 1967). The D yields are the most sensitive to the nuclear density at the nucleosynthesis epoch among the primordial light elements ³He, ⁴He and ⁷Li, thus making the D abundance the most sensitive measurement of the baryon density in the universe (Wagoner 1973; Schramm & Turner 1998).

Deuterium is currently measured in the local interstellar medium (ISM), (D/H) $_{\rm ISM} = (1.6\pm 0.1)~ 10^{-5}$ (Linsky et al. 1993), but since whenever it is cycled through stars it is completely burned away, extrapolation to the primordial D/H value requires a modeling of the Galactic chemical evolution. Direct D measurements of primordial material are thus of high interest. Adams (1976) suggested that almost primordial D could be measured in low metallicity absorption line systems in the spectra of distant quasars (QSOs). This was recently achieved for a few systems, but with conflicting results differing by almost an order of magnitude.

A few systems provide high D/H values (see e.g. Webb et al. 1997), who measure $\approx 2~ 10^{-4}$), while two other systems give a low abundance at $(3.39\pm 0.25)~ 10^{-5}$ (Burles & Tytler 1998a, 1998b). An even lower D/H estimation was obtained by Molaro et al. (1999), further discussed by Levshakov et al. (2000). Kirkman et al. (2000) measured an upper limit of 6.76  10^-5. The handful of D detections found so far does not allow a firm conclusion. Different arguments favour a low primordial D/H ratio: the possible H I contamination of the D I absorption lines and, on the modeling side, the results by Tosi et al. (1998) which predict for a variety of chemical evolution scenarios and to be consistent with the Galactic data a maximum decrease of the primordial D abundance by a factor of 3.

The paucity of suitable absorption systems for accurate D/H measurements is due to the fact that only absorption line systems with simple velocity structures and with intermediate H I column densities allow the detection of the D I lines. At too low column densities the D I lines are too weak for detection, whereas at high column densities the lines are normally washed out by the saturation of the H I line. We show here for the first time that in the latter case the deuterium signature can be successfully detected through the higher members of the Lyman series, when the target is a damped Ly$\alpha$ system ($\log N$(H I) $\geq 20.35$) at high redshift. This approach was first suggested by Khersonsky et al. (1995).

Figure 1: Absorption metal line profiles plotted against velocity for the DLA system at $z_{\rm abs} = 3.025$. The vertical scale goes from 0 to 1 for each plotted transition. The zero velocity is fixed at z = 3.024856. The vertical lines mark the positions of 7 components. The solid thin curve represents the best fit solution. The transitions O I$\l$1302 and Si II$\l$1304 are from the HIRES-Keck spectra published by Prochaska & Wolfe (1999)