Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.arcetri.astro.it/science/k20/papers/gilli.ps Дата изменения: Tue Feb 17 13:50:59 2004 Дата индексирования: Fri Feb 28 06:57:41 2014 Кодировка: Поисковые слова: x-ray background

arXiv:astro­ph/0304177
v1
9
Apr
2003
Tracing the large scale structure in the Chandra Deep Field South
R. Gilli 1;2 , A. Cimatti 1 , E. Daddi 3 , G. Hasinger 4 , P. Rosati 3 , G. Szokoly 4 , P. Tozzi 5 , J.
Bergeron 3 , S. Borgani 6 , R. Giacconi 2;7 , L. Kewley 8 , V. Mainieri 3;9 , M. Mignoli 10 , M.
Nonino 5 , C. Norman 2;11 , J. Wang 2 , G. Zamorani 10 , W. Zheng 2 , and A. Zirm 2
ABSTRACT
We report the discovery of large scale structures of X-ray sources in the 1Msec
observation of the Chandra Deep Field South. Two main structures appear as
narrow (
z . 0:02) spikes in the source redshift distribution at z=0.67 and
z=0.73, respectively. Their angular distribution spans a region at least  17
arcmin wide, corresponding to a physical size of 7:3h 1
70
Mpc at a redshift of
z 
0:7(
m =
0:3,
 = 0:7). These spikes are populated by 19 sources
each, which are mainly identied as Active Galactic Nuclei (AGN). Two sources
in each spike are extended in X-rays, corresponding to galaxy groups/clusters
embedded in larger structures. The X-ray source redshift distribution shows
other spikes, the most remarkable at z=1.04, 1.62 and 2.57. This is one of the
rst evidences for large scale structure traced by X-ray sources and for spatial
clustering of X-ray selected AGN. The X-ray data have been complemented with
the spectroscopic data from the K20 near infrared survey (Cimatti et al. 2002),
1 Istituto Nazionale di Astrosica (INAF) - Osservatorio Astrosico di Arcetri, Largo E. Fermi 5, 50125
Firenze, Italy
2 The Johns Hopkins University, Homewood Campus, Baltimore, MD 21218
3 European Southern Observatory, Karl-Schwarzschild-Strasse 2, Garching, D-85748, Germany
4 Max-Planck-Institut fur extraterrestrische Physik, Postfach 1312, D-85741 Garching, Germany
5 Istituto Nazionale di Astrosica (INAF) - Osservatorio Astronomico, Via G. Tiepolo 11, 34131 Trieste,
Italy
6 INFN, c/o Dip. di Astronomia dell'Universita, Via G. Tiepolo 11, 34131 Trieste, Italy
7 Associated Universities, Inc. 1400 16th Street, NW, Suite 730, Washington, DC 20036
8 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138
9 Dip. di Fisica, Universita degli Studi Roma Tre, Via della Vasca Navale 84, I-00146 Roma, Italy
10 Istituto Nazionale di Astrosica (INAF) - Osservatorio Astronomico di Bologna, Via Ranzani 1, 40127
Bologna, Italy
11 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218

{ 2 {
which covers  1=10 of the X-ray eld. Also in this survey the source redshift
distribution shows several spikes. Two narrow structures at z=0.67 and z=0.73
(again with
z  0:02) are the most signicant features, containing 24 and
47 galaxies, respectively. While the K20 structure at z=0.73 is dominated by
a standard galaxy cluster with a signicant concentration around a central cD
galaxy and morphological segregation, the galaxies at z=0.67 constitute a loose
structure rather uniformly distributed along the K20 eld.
Moreover, we nd a very good correlation (almost one-to-one) also between
less prominent peaks detected in the redshift distributions of X-ray and K20
sources. In particular, at z < 1:3 we nd that 5 out of the 6 more signicant K20
peaks have a corresponding peak in the X-ray selected sources and, similarly, all
the 5 X-ray peaks below that redshift have a corresponding K20 peak. Since the
K20 survey sensitivity drops beyond z  1:3, structures at higher redshift are
traced only by the X-ray sources. This correlation suggests that AGN (from the
X-ray data) and (early-type) galaxies (from the K20 survey) are tracing the same
underlying structures.
We also compared the X-ray and K-band catalogs to search for enhanced X-
ray activity in the sources in the two main redshift spikes. While in the structure
at z=0.73 the fraction of X-ray sources is the same as in the eld, in the structure
at z=0.67 it is higher by a factor of  2, suggesting that X-ray activity may be
triggered preferentially in the structure at z=0.67. Given the limited statistics,
this result is signicant only at  2 level.
1. Introduction
While the large scale structure of the Universe at z < 1 is usually mapped through
galaxy surveys, AGN surveys are a powerful tool to study the clustering of high redshift
objects. AGN clustering has been extensively studied and detected at optical wavelengths
(Shanks et al. 1987; La Franca et al. 1998; Croom et al. 2001), where objects are mainly
selected by means of their strong UV excess and include almost exclusively unobscured-type
1 AGN. An advantage of the X-ray selection, especially in the hard X-rays, resides in the
capability of detecting also obscured AGN, which, based on population synthesis models for
the X-ray background, are believed to be a factor of 4{10 more abundant than unobscured
AGN (Comastri et al. 1995; Gilli et al. 2001).
While angular clustering of X-ray selected AGNwas detected by several authors (Vikhlinin
& Forman 1995; Akylas et al. 2000; Giacconi et al. 2001), yet there are no direct measure-

{ 3 {
ments of their spatial clustering. Boyle & Mo (1993) studied the AGN at z < 0:2 in the
Einstein Medium Sensitivity Survey (EMSS, Stocke et al. 1991), without nding any posi-
tive clustering signal. Carrera et al. (1998) considered the AGN in the ROSAT International
X-ray Optical Survey (RIXOS, Mason et al. 2000) and in the Deep ROSAT Survey (Boyle
et al. 1994), detecting only a weak ( 2) clustering signal on scales < 60 120 h 1
70
Mpc
for the RIXOS AGN subsample in the redshift range z = 0:5 1:0.
The Chandra Msec surveys in the Deep Field South (CDFS, Rosati et al. 2002) and
North (CDFN, Brandt et al. 2001) will allow a step forward in the detection of X-ray source
structures. A large spectroscopic identication program down to very faint magnitudes
(R < 25:5) is underway in the CDFS, which is expected to provide reliable redshift estimates
for  70% of the X-ray sample. Although 10 20% of the sources in the deep X-ray surveys
are identied as normal/starburst galaxies, still AGN are the dominant population (Giacconi
et al. 2001; Barger et al. 2002). In this paper we will consider the redshift distribution of
those CDFS sources for which a robust redshift determination is already available ( 38%
of the full sample). This distribution shows two prominent spikes at z = 0:67 and z = 0:73
(see also Hasinger 2002) as well as several other spikes at higher redshift. Narrow spikes are
commonly detected in the redshift distribution of optical galaxies observed in deep pencil
beam surveys (Le Fevre et al. 1996; Cohen et al. 1999), and are now being observed also
among X-ray selected objects. Narrow structures at z  0:8 and z  1, similar to those
observed in the CDFS, are indeed observed also in the CDFN (Barger et al. 2002).
In this paper the features in the redshift distribution of X-ray sources in the CDFS
are compared with those observed in the K20 survey (Cimatti et al. 2002), which eфciently
selects massive galaxies in a broad redshift range and covers part of the CDFS eld. In
addition, the X-ray and K20 catalogs are cross correlated to search for any enhanced X-
ray activity in the two main redshift structures with respect to the eld. We note that
a concentration of objects at z=0.67 in the CDFS (mostly early type galaxies) was also
reported by Croom, Warren & Glazebrook (2001), who selected objects with K  19:5 and
J K colors redder than the stellar sequence.
The analysis of the spatial clustering of the full X-ray sample and the comparison with
the spatial clustering of K20 sources will be presented in a future paper.
Throughout this paper we will use a cosmology with H 0 = 70 km s 1 Mpc 1
,
m = 0:3,

 = 0:7.

{ 4 {
2. X-ray sources
The CDFS has been observed for 1Msec with ACIS-I, and represents one of the deepest
X-ray surveys to date (Rosati et al. 2002). Limiting uxes of 5:5 10 17 erg cm 2 s 1 and
4:5 10 16 erg cm 2 s 1 have been reached in the 0.5-2 keV and 2-10 keV band, respectively;
346 sources have been detected in the whole 0.1 deg 2 eld. The full X-ray catalog and the
details of the detection process have been presented by Giacconi et al. (2002). The optical
follow-up was primarily performed using the FORS1 camera at the VLT. The combined R
band data cover a 13:6  13:6 arcmin eld to limiting magnitudes between 26 and 26.7.
In the CDFS area not covered by FORS mosaics, we used shallower data from the ESO
Imaging Survey (EIS, Arnouts et al. 2001). The optical identication process is described
in Tozzi et al. (2001) and Giacconi et al. (2001). Optical spectroscopy for most of the X-ray
counterparts with R< 25:5 has been obtained with FORS1 during several observational runs
at VLT. The analysis of the spectroscopic data is almost completed. The details of the data
reduction, as well as the nal redshift list, will be presented in Szokoly et al. (in preparation).
So far 169 redshifts have been obtained. Quality ags have been assigned to the spectra,
according to their reliability. Here we consider only the 121 sources with a quality ag Q = 2
(excluding a few stars), where two or more emission lines have been observed and the redshift
determination is unambiguous, plus 10 sources whose redshift is determined from the cross-
correlation with K20 sources. The sample includes sources detected either in the soft (0.5-2
keV) or in the hard (2-10 keV) band or both. The 1 errors on the redshift measurements
are typically of the order of
z  0:002. The redshift distribution is dominated by two large
concentrations of sources in the ranges 0:664  z  0:685 and 0:725  z  0:742 containing
19 objects each. In the following we will refer to these two structures as sources at z = 0:67
and sources at z = 0:73, respectively. These striking structures can be seen in the upper
panel of Figure 1, where we plot the redshift distribution of X-ray sources in redshift bins of

z = 0:02. Due to the xed bin width, both structures appear to contain 18 rather than 19
sources. Other possible, less prominent structures (see Section 4) are clearly visible in this
plot.
As shown in Figure 2, sources at z=0.67 and z=0.73 are mostly distributed in one half
of the eld. Since the spectroscopic coverage of the CDFS is rather uniform (Szokoly et
al., in preparation), the higher concentration of sources in half of the eld is not likely to
be an artifact. The structures extend to a scale of  17 arcmin, corresponding to a linear
physical size of 7.3 Mpc at z  0:7. This is likely to be a lower limit, since the source angular
distribution appears to be limited only by the nite size of the X-ray image. Two extended
sources (CDFS 560, 645), identied as two galaxy groups, are present in the structure at
z=0.67. Similarly, two extended sources (CDFS 594, 566) belong to the structure at z=0.73:
one is identied as a galaxy group, the other is associated with the galaxy cluster in the

{ 5 {
K20 survey (see next Section and Fig. 2 and 3). The full list of CDFS extended sources
is given in Giacconi et al. (2002). Preliminary results on the CDFS extended sources are
being obtained from XMM observations: while CDFS 566 is not suфciently extended to be
resolved by XMM, CDFS 594 appears to be a factor of  4 more extended than measured
by Chandra (J. Bergeron, priv. comm.).
In addition to the above discussed structures, smaller peaks are also recognized in the
X-ray source redshift distribution. To assess the signicance of these structures we follow
a procedure similar to that adopted by Cohen et al. (1999), who observed features in their
galaxy redshift distribution with typical velocity dispersion of  300 km/s. Sources are
distributed in V = c ln(1 + z) rather than in redshift, since dV corresponds to local velocity
variations granted the Hubble expansion. The observed distribution is then smoothed with
a Gaussian with  S = 300 km/s to obtain the \signal" distribution. Since there is no a
priori knowledge of the \background" eld distribution, we heavily smoothed the observed
distribution with a Gaussian with B = 1:5 10 4 km/s and considered this as the background
distribution.
We then searched for possible redshift peaks in the signal distribution, computing for
each of them a signal-to-noise ratio dened as SNR=(S B)=
p
B, where S is the number
of sources in a velocity interval of xed width
V = 2000 km/s around the center of each
peak candidate and B is the number of background sources in the same interval. The value
of
V was chosen to optimize the SNR values for peaks populated by 3-5 sources. For
more populated peaks the quantity S may underestimate the actual number of sources in
each peak. Adopting the thresholds S  3 and SNR > 3:8 we nd 7 peaks in the signal
distribution. In order to estimate the expected fraction of possibly \spurious" peaks arising
from background uctuations, we simulated 1000 samples of 131 redshifts randomly extracted
from the smoothed background distribution and applied our peak detection method to each
simulated sample. The result is that, with the adopted thresholds, the average number of
spurious peaks due to background uctuations is 0.47. In only 13 of the simulated samples
3 spurious peaks are detected, while in none of simulated samples 4 or more spurious peaks
are detected.
The seven X-ray source peaks detected by our procedure with S  3 and SNR > 3:8
are listed in Table 1, where for each peak we give the average redshift, the number of objects
(N) in the peak and the Poissonian probability of observing N sources given the background
value in the same velocity intervals where the N sources are found.
We checked our detection method with larger values of the smoothing length for the
signal distribution. We veried that, when increasing  S , several peaks are lost; as an
example, when the signal smoothing is as large as 1500 km/s, corresponding to the velocity

{ 6 {
dispersion of rich clusters, only the three most signicant peaks at z=0.67, 0.73 and 1.62 are
detected. Therefore we will consider only the results obtained with  S = 300 km/s in the
following.
3. Sources from the K20 survey
Part of the CDFS eld has been covered by the K20 survey (see Cimatti et al. 2002
and references therein), which includes all the sources detected in the EIS K-band images to
Ks < 20. The K20 eld (6.7 by 4.8 arcmin) covers  1=10 of the X-ray eld, and is shown as
a rectangle in Figure 2: 348 sources with Ks < 20 have been found in this area. Spectroscopic
identication of the K20 sources has been performed with FORS1 and FORS2 at the VLT.
We will consider here only the 258 galaxies with highest quality spectra and unambiguous
redshift determination. The typical error on the redshift estimate is
z = 0:002. The
redshift distribution of the K20 sources in bins of
z = 0:02 is shown in the lower panel
of Figure 1. In strict analogy with the X-ray results, the redshift distribution of the K20
galaxies has two prominent peaks at z=0.67 and z=0.73.
While in the X-rays the two structures at z=0.67 and z=0.73 appear equally populated,
in the smaller K20 eld sources at 0.73 (47 objects) are a factor of  2 more numerous
than sources at 0.67 (24 objects). The redshift ranges dening the two spikes are the same
adopted for X-ray sources (0:664  z  0:685 and 0:725  z  0:742). Again, due to the
xed bin size adopted in Fig. 1, only 45 and 22 sources appear in the spikes at z=0.73 and
z=0.67, respectively.
Inspection of the distribution on the sky of the K20 sources together with their spectral
properties reveals that the two structures are indeed very dierent. The structure at z = 0:73
appears to be dominated by a standard galaxy cluster, showing a central cD galaxy and
signicant spectral segregation, with early type objects concentrated around the cD galaxy.
Overall, 25 out of 47 sources at z = 0:73 are classied as early type galaxies, 17 are emission
line galaxies, 4 have intermediate spectra, and 1 object has been recognized as an AGN.
Sixteen out of 19 K20 sources within a circle of radius 1 arcmin ( 440 Kpc) around the
central cD are early type objects (see Fig. 3). On the contrary, sources at z = 0:67 constitute
a loose structure with early and late type galaxies uniformly distributed across the K20 eld.
For completeness, we note that 11 out of 24 sources at z = 0:67 are classied as early type
galaxies, 11 are late type galaxies and 2 have intermediate spectra.
We have searched for other peaks in the K20 redshift distribution using the same pro-
cedure described in the previous Section. Adopting the same parameters as in the analysis

{ 7 {
of the X-ray data for the smoothing lengths ( S = 300 km/s and B = 1:5 10 4 km/s) and
similar values for the thresholds in SNR and S (SNR > 4:0 and S  4) we nd 8 peaks. Also
in this case the average number of spurious peaks, obtained from 1000 simulated samples, is
low (0.11). None of the simulated samples contains 3 or more spurious peaks. The central
redshifts, number of sources, and Poissonian probability related to the K20 peaks are quoted
in Table 1.
Inspection of Table 1 reveals that, besides the two prominent spikes at z=0.67 and
z=0.73, other peaks are in common between the K20 and the X-ray data. The third most
populated K20 peak (15 sources at z=1.036) is detected also in the X-rays, being the third
populated peak of the X-ray redshift distribution (6 sources). Another two K20 peaks (at z 
0.077 and 1.220) have a corresponding peak in the X-ray source redshift distribution. Overall,
5 out of 8 peaks in the K20 source redshift distribution have an X-ray peak counterpart. If
we restrict our analysis to the more signicant K20 peaks with Poissonian probability smaller
than 10 3 and 5 10 4 , the fraction with X-ray counterpart is 5/6 and 4/4, respectively. The
fact that the less signicant K20 peaks do not have a corresponding X-ray peak may be
expected since the total number of X-ray sources is much lower (a factor of  2) than that
of K20 sources and is spread over a wider redshift range. Only about 30% of the X-ray
sources in these peaks is within the area covered by the smaller K20 survey. This means
that, in general, the structures seen in the K20 survey are traced on wider scales by the
X-ray sources. We note that the high redshift peaks detected in the X-rays are not detected
in the K20 observations, whose sensitivity drops dramatically above z  1:3.
4. The X-ray source fraction in the redshift structures
4.1. Total X-ray sample
Since enhanced nuclear or star forming activity (both marked by X-ray emission) are
likely to be produced by galaxy interactions in large scale structures, we searched for any
variation in the X-ray to K20 source number ratio inside and outside the two structures at
z=0.67 and z=0.73.
First, we simply divided the number of X-ray sources by the number of K20 sources
inside and outside the two structures to search for variations with respect to the eld. Ex-
tended X-ray sources (two in each structure) have been excluded from this analysis. Despite
the signicant spectroscopic incompleteness in the X-ray sample ( 60% of the CDFS sources
have no secure redshift determination yet), the results of this analysis should not be biased as
long as the X-ray spectroscopic incompleteness is \random" with respect to the probability

{ 8 {
of an X-ray source being inside or outside an overdensity. Since the redshift distributions of
the two samples are dierent, in estimating the number of sources in the \eld" we consid-
ered only the sources not belonging to the two structures in a redshift range centered at the
redshift of the two structures. Unfortunately, given the limited statistics, this range can not
be too small. Adopting the redshift range 0.4{1.0, we nd that the X-ray to K-band number
ratio is 0:33  0:07 in the eld (this ratio would be similar, but with a larger uncertainty, if
we adopted the narrower redshift range 0.5{0.9), 0:36  0:10 in the higher redshift spike and
0:71  0:22 in the lower redshift spike. Given the relatively large errors, these ratios are all
consistent with each other, even if the higher value in the lower redshift spike suggests that
X-ray activity may be enhanced in this structure.
We note that most of the X-ray sources are identied as AGN on the basis of their optical
and X-ray properties. Whenever the optical classication is uncertain, we considered as AGN
those sources satisfying at least one the following conditions: L x > 10 42 erg s 1 , where L x is
the observed luminosity in the 0.5-10 keV band; HR > 0, where HR is the X-ray hardness
ratio; f x =f r > 0:1, where f x and f r are the uxes in the 0.5-10 keV and R band, respectively.
Indeed, normal and starburst galaxies typically do not exceed luminosities of 10 42 erg s 1
(e.g. Fabbiano et al. 1992) and have X-ray to optical ux ratios an order of magnitude
smaller than those of AGN (Stocke et al. 1991). Also, starburst spectra cannot be as hard
as those of heavily obscured AGN: we veried that an average starburst template derived
from Dahlem et al. (1998) would always produce HR < 0 in our observations. Overall, 101
out of the 131 X-ray sources in our sample are classied as AGN. Fifteen AGN are found in
the z = 0:67 spike and 16 in the z = 0:73 spike, to be compared with the total number of 17
X-ray point-like sources in each spike. As a consequence, the results of the above analysis
would be essentially the same if only sources identied as AGN were considered.
4.2. X-ray to K20 matched sample
The results presented in the previous Section are based on the comparison of data from
two surveys (X-ray and K20) which cover dierent areas. In order to further check the
possible signicance of the enhanced X-ray activity among the sources at z=0.67, we have
studied in more detail the correlation between the K20 sources and the 49 X-ray sources
detected within the K20 eld.
First we searched for any K20 counterpart of an X-ray source within a radius of 10
arcsec, choosing as the most likely counterpart candidate the K20 source closer to the X-
ray source. Then we computed the RA and DEC oset histograms and tted them with a
Gaussian. The Gaussian width resulted to be  RA = 0:55 arcsec and  DEC = 0:50 arcsec for

{ 9 {
the RA and DEC histogram respectively. We then corrected the K-band coordinates for the
systematic shift with respect to the X-ray coordinates and searched for the K20 counterpart
in a circle with radius = 1.77 arcsec (3.4, where  is the average between  RA and  DEC )
around the X-ray source position. We recall that, given a circular Gaussian of width , a
circle of radius 3:4 encloses 99.7% of the volume and then corresponds to the 3 error box.
The cross correlation produced 30 matches between X-ray and K-band sources (then 19
X-ray sources in the K20 eld have no counterpart down to K s < 20). Given the surface
density of K20 and X-ray sources, we estimate that the expected number of spurious chance
coincidences among these X-ray and K20 matches is of the order of one.
We note that the positional match has been done using the coordinates of the X-ray
source centroids rather than those of the optical counterparts, which are typically separated
by  0:5 arcsec from the X-ray centroids. Using the coordinates of the optical counterparts
rather than those of the X-ray sources would not aect our results.
Although our adopted match radius might not be appropriate in the case of extended
X-ray sources, we note that only two X-ray sources falling in the K20 eld are extended.
Both of them have K20 counterparts and are examined separately. With our adopted search
radius, a unique K20 counterpart is found for the extended source CDFS 566, which coincides
with the cD galaxy at the center of the cluster at z = 0:73. The other extended source,
CDFS 511, can be associated to any of three K20 sources with redshifts between 0.765 and
1.047, and its identication remains uncertain. For all the 28 point-like X-ray sources with
a K20 counterpart, this is unique. Only three out of 30 matches do not have a high quality
redshift determination either from the K-band or the X-ray catalog. We veried that for
the 20 sources with high quality redshifts both in the X-ray and in the K20 catalog the two
measurements are always in excellent agreement, with essentially zero oset and a dispersion
of
z = 0:003. It is interesting to note that a large fraction (8/18) of the sources classied as
AGN in the X-ray catalog were not recognized as such in the K20 spectroscopic data alone.
We then estimated the fraction of K-band selected sources detected in the X-rays inside
and outside the two structures, excluding extended X-ray sources. Again an excess of X-ray
sources in the low redshift spike appears. The fraction of X-ray detected sources in the low
redshift spike is 0:21  0:10 (5/24), while it is 0:08  0:04 both in the high redshift spike
(4/47) and in the neighboring redshift bins (i.e. in the range 0:4 < z < 1:0).
Since we are now dealing with a sample free from selection biases, we tried to estimate
the signicance of the X-ray source overdensity at z = 0:67. Since 25 X-ray to K20 matches
are point-like sources with measured redshift, we simulated 10000 samples of 25 sources
randomly extracted from the K20 spectroscopic catalog, nding that in only 5.7% of the

{ 10 {
cases these random samples contain 5 or more sources at z=0.67 (i.e. in the redshift interval
0.664{0.685). We can therefore conclude that the X-ray source enhancement in the low
redshift spike is statistically signicant at about 2 level. A similar level of signicance is
found by restricting this analysis to the 18 matches identied with AGN (4 of which are at
z=0.67).
5. Discussion
Our results show that in the CDFS several large scale structures of X-ray sources (mostly
AGN) are being detected, which appear as narrow spikes in the source redshift distribution.
The most prominent structures are detected at z=0.67 and z=0.73, while other peaks appear
at higher redshifts. 12 In the CDFN Barger et al. (2002) found evidence for similar redshift
spikes at z=0.843 and 1.017 with at least ten X-ray sources each. Despite several eorts in
the past years, before the deep X-ray surveys by Chandra there was no convincing evidence
for three-dimensional clustering of X-ray selected AGN. Only a 2 detection was found by
Carrera et al. (1998) in the ROSAT International X-ray Optical Survey (RIXOS, Mason et
al. 2000) on scales < 60 120 h 1
70
Mpc. Interestingly, the 2 signal detected in the RIXOS
refers to the subsample of sources in the redshift range 0.5{1.0, where the biggest structures
in the Chandra deep elds are also detected, although the lack of clustering signal at z < 0:5
and z > 1 might be due to the small volume sampled and to the falling sensitivity of the
RIXOS, respectively.
Signicant clustering signal of X-ray selected AGN was found from two dimensional
analysis based on the angular correlation function (e.g. Akylas et al. 2000; Vikhlinin &
Forman 1995). The clustering length of local X-ray selected AGN was found to be similar
to that of local galaxies, supporting the idea that active nuclei are unbiased with respect
to normal galaxies. This was also reported by Smith et al. (1995) who found that the
amplitude of the low-redshift (z < 0:3) QSO-galaxy angular cross-correlation function is
identical to that of the APM galaxy angular correlation function, implying that QSO inhabit
environments similar to those of normal galaxies. Brown, Boyle & Webster (2001) showed
that in the redshift range 0:2 < z < 0:7 the spatial cross-correlation of AGN with early type
galaxies is much stronger than that between AGN and late type galaxies. We found that
80% of the highly signicant peaks seen in the K20 source redshift distribution, in the range
12 We note in passing that even removing the two structures at z=0.67 and z=0.73, the X-ray source
redshift distribution in the CDFS shows an excess at z < 1 with respect to the predictions of standard
synthesis models of the X-ray background (Comastri et al. 1995; Gilli et al. 2001).

{ 11 {
0:5 < z < 1:3, have a corresponding peak in the X-rays. This supports the idea that at
these redshifts AGN and early type galaxies, whose detection rate is higher in K-band than
in optically selected samples, are tracing the same underlying structures.
Since the large scale structures at z = 0:67 and z = 0:73 appear as narrow spikes
in the source redshift distribution and span the whole Chandra eld of view, we veried
whether they can be actually considered as \sheets", i.e. nearly bi-dimensional structures,
or laments of galaxies as predicted by cosmological models of structure formation. The
structures at z = 0:67 and z = 0:73 have an observed thickness of 18:8 Mpc and 13:4
Mpc respectively (the redshift range covered by sources at z = 0:67 is slightly larger than
that covered by sources at z = 0:73), to be compared with their 7.3 Mpc extension on the
plane of the sky. Given the errors on the redshift measurements, the values obtained for
the physical thickness of the structures are likely to be upper limits. Furthermore, if these
structures do not lay exactly on the plane of the sky, any projection eects would broaden
their apparent thickness. We veried that for the broader structure at z = 0:67 sources at
higher right ascension tend to have smaller redshifts. This projection eect needs however
better statistics to be quantied. We nally note that 7.3 Mpc is likely to be a lower limit
to the angular extension of the two structures, which cover almost entirely the Chandra eld
of view. Given the above uncertainties, an enlarged eld and increased source statistics are
needed to determine the geometry of the two structures.
We searched for enhanced X-ray activity at z = 0:67 and z = 0:73 by cross-correlating
the X-ray and the K20 catalogs and looking at the fraction of K20 sources with X-ray
counterparts at dierent redshifts. Five out of 47 K20 sources at z = 0:73 match an X-ray
source: one is the extended source CDFS 566 and is associated to the cD galaxy of the
rich cluster seen in the K20 survey; one is classied as a normal galaxy and the remaining
three sources are identied as AGN. Five out of 24 K20 sources at z = 0:67 do show X-ray
emission: one is classied as a normal galaxy, while the remaining four are classied as AGN.
We found a weak (2) X-ray source overdensity in the structure at 0:67 with respect to the
eld and to the other structure at z = 0:73. This overdensity is also observed with the
same signicance when considering only those X-ray sources identied as AGN. While in the
X-rays the two structures are very similar (same number of sources, similar angular extent
and redshift thickness), in the K20 survey they appear very dierent. The K20 sources at
z = 0:73 are dominated by a virialized galaxy cluster, with a central cD galaxy and spectral
segregation, with early type galaxies clustered in the inner regions. On the contrary the K20
sources at z = 0:67 do not appear to constitute a galaxy cluster, being uniformly distributed
across the K20 eld. We note incidentally that > 50% of the K20 sources classied as early
type galaxies lay in the two structures at z = 0:67 and z = 0:73 and that a concentration of
early type galaxies at 0:67 in the CDFS was also reported by Croom, Warren & Glazebrook

{ 12 {
(2001). The dierence between the two structures when observed by the K20 survey can be
ascribed to the small region covered by the K20 survey, which has sampled a rich cluster at
z=0.73 without sampling any cluster at z = 0:67. Since the two structures are very similar in
the X-rays, we might speculate that, in a region free from clusters, we would have observed
an X-ray source overdensity even at z = 0:73. In principle, this suggests that X-ray activity
in the large scale structures is triggered preferentially away from the higher density peaks
corresponding to galaxy clusters. However, further work is necessary to increase the source
statistics and test this idea.
6. Conclusions and future work
Our main results can be summarized as follows:
1) Several large scale structures of X-ray sources (mostly AGN) are being detected in the
Chandra Deep Field South, which appear as narrow spikes in the source redshift distribution.
The most prominent structures are detected at z=0.67 and z=0.73. In addition, high redshift
structures are signicantly detected at z=1.04, 1.62 and 2.57. This represents one of the rst
evidences for spatial clustering of X-ray selected AGN.
2) Similar redshift structures, the most prominent at z = 0:67 and z = 0:73, are observed
among the K-band selected galaxies of the K20 survey. About 80% of the most signicant
peaks in the redshift distribution of K20 sources have a corresponding peak in the X-rays.
Since only a fraction of the X-ray sources in these peaks is within the smaller K20 eld,
we can conclude that, in general, the structures seen in the K20 survey are traced on wider
scales by the X-ray sources. We also notice that, since the K20 survey sensitivity drops
above z  1:3, high redshift structures are detected only in the X-rays.
3) By cross-correlating the K20 and the X-ray catalog we nd a weak evidence ( 2) of
enhanced nuclear activity among the K20 sources at z = 0:67 with respect to the eld,
suggesting a preferential trigger in this structure. Given its low signicance, this result has
to be conrmed by further observations.
Signicant improvements in the understanding of the large scale structures in the CDFS
is expected from the on going and planned multiwavelength observations of the eld. The
XMM deep pointing (500 ksec in a 30 arcmin diameter region; Hasinger et al. 2002) is
actually expanding the area covered by X-ray data by a factor of  2. Accurate photometric
redshifts for the optically faint X-ray sources are being obtained from FORS optical images
and ISAAC near-IR data and will be rened further using data from the Advanced Camera
for Survey (ACS) on HST.

{ 13 {
We wish to thank the referee for useful comments.
REFERENCES
Akylas, A., Georgantopoulos, I., Plionis, M. 2000, MNRAS, 318, 1036
Arnouts, S., Vandame, B., Benoist, C., et al. 2001, A&A, 379, 740
Boyle, B.J., & Mo, H.J. 1993, MNRAS, 260, 925
Boyle, B. J., Shanks, T., Georgantopoulos, I., Stewart, G. C., Griфths, R. E. 1994, MNRAS,
271, 639
Barger, A.J., Cowie, L.L., Brandt, W.N., Capak, P., Garmire, G.P., Hornschemeier, A.E.,
Steen, A.T., Wehner, E.H. 2002, AJ, in press [astro{ph/0206370]
Brandt, W.N., Alexander, D.M., Hornschemeier, A.E., et al. 2001, AJ, 122, 2810
Brown, M.J.I., Boyle, B.J. & Webster, R.L. 2001, AJ, 122, 26
Carrera F.J., Barcons, X., Fabian, A.C., Hasinger, G., Mason, K.O., McMahon, R.G., Mit-
taz, J.P.D., Page, M.J. 1998, MNRAS, 299
Cimatti, A., Mignoli, M., Daddi, E., et al. 2002, A&A, 392, 395
Cohen, J.G., Blandford, R., Hogg, D.W., Pahre, M.A., Shopbell, P.L. 1999, ApJ, 512, 30
Comastri, A., Setti, G., Zamorani, G., Hasinger, G. 1995, A&A, 296, 1
Croom, S.M., Warren, S.J., Glazebrook, K. 2001, MNRAS, 328, 150
Croom, S.M., Shanks, T., Boyle, B.J., Smith, R.J., Miller, L., Loaring, N.S., Hoyle, F. 2001,
MNRAS, 325, 483
Dahlem, M., Weaver, K., & Heckman, T. 1998, ApJS, 118, 40
Fabbiano, G., Kim, D.-W., & Trinchieri, G. 1992, ApJS, 80, 531
Giacconi, R., Rosati, P., Tozzi, P., Nonino, M., Hasinger, G., Norman, C., Bergeron, J.,
Borgani, S., Gilli, R., Gilmozzi, R. & Zheng, W. 2001, ApJ, 551, 624
Giacconi, R., Zirm, A., Wang, J., Rosati, P., Nonino, M., Tozzi, P., Gilli, R., Mainieri,
V., Hasinger, G., Kewley, L., Bergeron, J., Borgani, S., Gilmozzi, R., Grogin, N.,
Koekemoer, A., Schreier, E., Zheng, W., Norman, C. 2002, ApJS, 139, 369

{ 14 {
Gilli, R., Salvati, M., & Hasinger, G. 2001, A&A, 366, 407
Hasinger, G., et al. 2002, ESO Msngr, 108, 11
La Franca, F., Andreani, P., & Cristiani, S., 1998, ApJ, 497, 529
Le Fevre, O., Hudon, D., Lilly, S. J., Crampton, David, Hammer, F., Tresse, L. 1996, ApJ,
461, 534
Mason, K. O., Carrera, F. J., Hasinger, G., et al. 2000, MNRAS, 311, 456
Hornschemeier, A., et al. 2000, ApJ, 541, 49
Shanks, T., Fong, R., Boyle, B.J., Peterson, B.A. 1987, MNRAS, 277, 739
Smith, R.J., Boyle, B.J., Maddox, S.J. 1995, MNRAS, 277, 270
Smith, R.J., Boyle, B.J., Maddox, S.J. 2000, MNRAS, 313, 252
Stocke, John T., Morris, Simon L., Gioia, I. M., Maccacaro, T., Schild, R., Wolter, A.,
Fleming, Thomas A., Henry, J. P. 1991, ApJS, 76, 813
Rosati, P., Tozzi, P., Giacconi, R., Gilli, R., Hasinger, G., Kewley, L., Mainieri, V., Nonino,
M., Norman, C., Szokoly, G., Wang, J. X., Zirm, A., Bergeron, J., Borgani, S.,
Gilmozzi, R., Grogin, N., Koekemoer, A., Schreier, E., Zheng, W. 2002, ApJ, 566,
667
Tozzi, P., Rosati, P., Nonino, M., Bergeron, J., Borgani, S., Gilli, R., Gilmozzi, G., Hasinger,
G., Grogin, N., Kellerman, K., Kewley, L., Koekemoer, A., Norman, C., Schreier, E.,
Shaver, P., Wang, J.X., Zheng, W., Zirm, A. & Giacconi, R. 2001, ApJ, 562, 42
Vikhlinin, A., Forman, W. 1995, ApJ, 455, L109
This preprint was prepared with the AAS L A T E X macros v5.0.

{ 15 {
K20 X-ray
z N Prob. N Prob.
0.077 5 4:9 10 4 3 1:0 10 2
0.218 5 3:3 10 2
0.367 7 3:7 10 2
0.524 11 6:0 10 4
0.670 24 1:9 10 4 19 9:1 10 5
0.738 47 6:0 10 8 19 1:7 10 6
1.036 15 6:4 10 4 6 4:2 10 3
1.220 10 3:2 10 4 4 2:2 10 2
1.618 5 3:8 10 3
2.572 4 9:7 10 3
Table 1: Peaks detected in the X-ray and K20 source redshift distributions, sorted by in-
creasing redshift. The signal and background distributions are smoothed with  S = 300
km/s and B = 1:5 10 4 km/s, respectively. Together with the central redshift of each peak,
the number of sources N in each peak and the Poissonian probability of observing N or more
sources given the background value are also shown.

{ 16 {
Fig. 1.| Redshift distribution for X-ray sources (Upper Panel) and K20 sources (Lower
Panel) with high quality optical spectra. The binning shown in this Figure is
z = 0:02.
The insets show a zoom on the two main redshift spikes at z=0.67 and z=0.73 (binning is

z = 0:005). The black histograms represent the matches between the X-ray and the K20
catalogs. The \background" eld distributions, derived by smoothing the observed ones with
B = 1:5 10 4 km/s (see text), are shown as continuous lines.

{ 17 {
Fig. 2.| Chandra ACIS-I image of the CDFS with the sources in the two redshift spikes
marked with dierent symbols: circles for sources at z = 0:67 and squares for sources at
z = 0:73. Extended sources in the spikes are represented as big symbols. The ACIS-I
detector is  17 arcmin on a side. The dashed box indicates the 6:7  4:8 arcmin region
covered by the K20 survey.

{ 18 {
Fig. 3.| Spatial distribution of the K20 sources at z = 0:67 (open circles) and at z = 0:73
(lled squares). Their classication from K20 spectroscopy is also indicated by the numbers
inside the symbols. Crosses are overplotted on sources with X-ray counterparts. Sources
at = 0:73 appear to constitute a standard galaxy cluster with a cD galaxy marking its
center (indicated by the big cross) and showing spectral segregation, with early type galaxies
concentrated in the inner regions. On the contrary, sources at z = 0:67 do not cluster around
any cD galaxy and are uniformly distributed across the eld. One galaxy group at z=0.67
(CDFS 560) and one at z=0.73 (CDFS 594) detected in the X-rays just outside the K20 eld
are also indicated.