XMM-Newton observation of the Chandra Deep Field-South: Statistical treatment of faint source spectra

Abstract

We present first results of the X-ray spectral analysis of the 500 ksec deep survey obtained with XMM-Newton on the Chandra Deep Field South (CDFS). Statistical distributions of spectral index and intrinsic absorption are derived for a sample containing 70 sources with a count limit of 100 (flux limit in the [2-10] keV band of 8.9 $\times 10^{-16}$ erg cm$^{-2}$ s$^{-1}$), of which 44 have redshift identification. We observe a separation between the type-1 and the type-2 AGN in diagnostics involving different X-ray parameters. Using the subsample with known $z$, we show that this separation between the AGN populations is a consequence of different absorption column densities. The two populations have the same average spectral index, $<\Gamma> = 2 \pm 0.1$. We present integrated spectrum for the most distant type-2 QSO whith strong X-ray absorption and a clear soft excess; we obtained the best fit for these objects with two difference models: a scattering model and a double power law model. We also confirm a progressive hardening for the combined integrated spectra for faint objects which at first was noted by \cite{toz01a}. Our results shown a clear evolution of decrease of $<\Gamma>$ with decreasing flux in the hard 2-10 keV band. However, we detect not only a regular increase of $<\Gamma>$ for different subsamples of fluxes in comparison with Chandra results, but also an internal discrepancy of the values, if we fitted in the different energy bands.