1 Introduction

The deep ROSAT survey of the Lockman Hole showed that about 80$\%$ of the soft (0.5-2 keV) X-ray background (XRB) is resolved into discrete sources (Hasinger et al. 1998). These findings have recently been confirmed and strengthened using the two deep Chandra surveys of 1 Msec each (Brandt et al. 2001; Rosati et al. 2002). An important population of X-ray sources with hard spectra, most probably obscured active galactic nuclei (AGN), is present in the Chandra (Barger et al. 2001; Hornschemeier et al. 2001; Rosati et al. 2002) and XMM-Newton (Hasinger et al. 2001, hereafter Paper I) deep surveys; a few objects of this class had already been detected in ROSAT deep and shallower surveys (Lehmann et al. 2001a; Mittaz et al. 1999). In the hard band (2-10 keV), the X-ray source density derived from the number counts in the two Chandra deep surveys is about 4000 deg^-2 (Brandt et al. 2001; Rosati et al. 2002) resolving $\sim$85-90% of the 2-10 keV XRB. This population of X-ray sources show a progressive hardening of the average X-ray spectrum towards fainter fluxes (Tozzi et al. 2001; Mittaz et al. 1999).

The XMM-Newton deep survey ($\simeq$100 ksec of good quality data) of the Lockman Hole was obtained during Performance Verification. The X-ray data reduction and analysis (restricted to sources within a 10 arcmin radius) was reported in Paper I where it was demonstrated that the different populations of X-ray sources are well separated in X-ray spectral diagnostics based on hardness ratios. The extensive optical follow-up programs of this field (Lehmann et al. 2001a, and references therein) provide an understanding of the physical nature of the X-ray sources. The point sources detected in the soft band by ROSAT are predominantly unobscured (in both optical and X-ray bands) AGN spanning a wide redshift range. In the XMM-Newton sample, there is a significant fraction of sources with hard spectra. This new population is most probably dominated by intrinsically absorbed AGN. This assumption can be tested using the available optical spectra and, more efficently, by X-ray spectral study.

To this aim, we have performed an X-ray spectral analysis of the sources in the Lockman Hole to understand their physical nature combining the X-ray data with the optical/near IR information. We also use the subsample with redshift identification to check the validity of our conclusions concerning the specific properties of the obscured AGN population. Preliminary results of this work were reported by Mainieri et al. (2002).

In the following we will refer to Type-1 (broad and narrow emission lines) and Type-2 (high ionization narrow emission lines) AGN using the optical spectroscopic classification.

The observations are presented in Sect. 2. The results of the spectral analysis are described in Sect. 3, in particular the range of the X-ray spectral index, the observed $N_{\rm H}$ distribution and colour-colour diagnostic diagrams. The optical/near IR properties are discussed in Sect. 4 together with a comparison with QSO and galaxy evolutionary tracks. The search for relations between X-ray and optical/near IR fluxes is presented in Sect. 5. The effect of the absorbing column density on the X-ray luminosity and the Type-2 QSO candidates are discussed in Sect. 6. Representative spectra of the different classes of X-ray sources are given in Sect. 7. Finally, our conclusions are outlined in Sect. 8.