... sample

The HR is a reasonable indicator of the intrinsic column density of an AGN only for $z \la 1.5$. At higher redshifts an absorbed AGN can have a lower HR value due to the higher rest-frame energies probed in the observed X-ray band.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (GOODS)

See: http://www.stsci.edu/science/goods/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... preparation)

See: http://www.eso.org/science/goods/releases/20040430/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 2005

See: http://www.eso.org/science/goods/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... catalogue

http://www.stsci.edu/science/goods/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...2000)

BPZ is available from http://acs.pha.jhu.edu/~txitxo/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...)

$99 \%$ of the X-ray sources in the CDF-S with spectroscopic redshifts and $\rm HR>-0.6$ have $L_{\rm X}[0.5{-}10 ~ {\rm keV}]>{L}^{\rm min}_{\rm X}$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...2004)

This source is not reported in Tables 1 and 3 because, as shown in Szokoly et al. (2004), an optical counterpart (201b) with R=24.34 has been identified inside the Chandra error circle.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... bands

In Koekemoer et al. (2004) only 3/5 of the ACS observation time were considered while here we have used the entire ACS data set.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...2004)

The spectroscopic data of the CDF-S are publicly available at this URL: http://www.eso.org/datagw/prosati/cdfspub/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...$\rm X/O > 10$)

We note that $\sim$$75 \%$ of the X-ray sources in the CDF-S with $\rm X/O > 10$ are OFS.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 243)

We are sampling an X-ray flux regime ( $8 \times 10^{-16} \la {F_{\rm X} [2{-}10 ~ {\rm keV}]} \la 7 \times 10^{-15}$ erg cm^-2 s^-1) much fainter than the one studied by Mignoli et al. (2004), $1 \times 10^{-14} \la {F_{\rm X} [2{-}10 ~ {\rm keV}]} \la 1 \times 10^{-13}$ erg cm^-2 s^-1 (M. Brusa Ph.D. Thesis).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 336

The remaining sources are either without a redshift estimate (four) or are stars (seven).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... [2-10] keV band)

We note that at the depth of the CDF-S survey the whole hard XRB has been resolved according to the HEAO-1 measure (Rosati et al. 2002).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... respectively

http://www.stsci.edu/hst/udf/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 79,515,605

We have also checked the Alexander et al. (2003) catalogue, which includes all of the sixteen sources of Giacconi et al. (2002) plus one extra source with $\sim$12 counts in the full Chandra energy band and with an optically bright counterpart.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.