1: \documentclass[12pt]{article}
2:
3: \textwidth 180mm
4: \textheight 260mm
5: %\topmargin -20mm
6: \topmargin -20mm
7: \oddsidemargin -7mm
8: \evensidemargin -10mm
9: \pagestyle {plain}
10: \thispagestyle{empty}
11: \let\ni=\noindent
12:
13: %
14: \usepackage{graphicx}
15: \usepackage{natbib}
16: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
17: \usepackage{txfonts}
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19: \newcommand{\cgs}{{\it cgs}}
20: %\newcommand{\fdg}{.\!^\circle}
21: \newcommand{\fdg}{\ensuremath{.\hspace{-3pt}^\circ}}
22: %
23:
24: % added by arXiv admin:
25: \voffset=-1cm
26:
27: \begin{document}
28: %
29: \centerline{\Large{\bf The redshift distribution of the X-ray background}}
30:
31: \vspace{2mm}
32: \centerline {\large Andrzej M. So\l tan}
33:
34:
35: \centerline{Nicolaus Copernicus Astronomical Center}
36:
37: \centerline{{\small Bartycka 18, 00-716 Warsaw, Poland}}
38: \centerline{\small{soltan@camk.edu.pl}}
39:
40: \abstract
41: \ni The X-ray background (XRB) is produced by a large number of faint sources
42: distributed over a wide range of redshifts. The XRB carries information
43: on the spatial distribution and evolution of these sources.
44: {The goals of the paper are: 1. to determine the redshift distribution of the soft X-ray
45: background photons produced by all types of extragalactic sources, in order
46: to relate fluctuations of the background to the large scale structures,
47: 2. to determine the redshift distribution of the soft XRB produced by AGN in
48: order to calculate the evolution of the AGN X-ray luminosity density.}
49: {A set of major X-ray surveys is used to determine the redshift
50: distributions of the X-ray sources selected at various flux
51: levels. Simple analytic fits to the data allow us to determine the
52: smooth relationship between the redshift distribution and the
53: source flux. The redshift distribution of the integral XRB
54: flux is obtained by averaging the fits over the source counts.}
55: {It is shown that the distribution of extragalactic XRB photons in the
56: $0.5-2$\,keV band is adequately represented by the function:
57: $d\,n_{\rm XRB} / d\log z = 5.24\:z^{1.52}\,\exp(-z/0.63)$.
58: The huge voids postulated to explain the cold spots
59: in the CMB maps create dips in the total XRB flux.
60: However, the expected magnitude of the effect is comparable to the
61: fluctuation amplitude of the XRB generated by the individual sources
62: contributing to the background.
63: The cosmic evolution of the AGN X-ray luminosity density up to
64: redshift of $\sim\!5$ is calculated in an elegant and straightforward
65: way. Systematic uncertainties of the present method
66: are assessed and shown to be small.
67: At redshift greater than one the present results could be compared
68: directly with some recent estimates obtained in a standard way and
69: the agreement between both methods is very good.}
70:
71: \vspace{2mm}
72: \ni Keywords{X-rays: diffuse background --
73: intergalactic medium --
74: X-rays: galaxies }
75:
76: \section{\large Introduction \label{intro}}
77:
78: The X-ray background (XRB) is generated mostly by discrete extragalactic sources
79: (e.g. \citealt{lehmann01, kim07}, and references therein), predominantly by
80: various types of Active Galactic Nuclei (AGN) and cluster of galaxies. A
81: question of flux, luminosity and redshift distributions of these sources has
82: been discussed in a great number of papers for the last $30$ years. One of the
83: major outcome of these investigations is the conclusion that X-ray sources
84: associated with the AGN are subject to strong cosmic evolution (e.g.
85: \citealt{miyaji00}, \citealt{silverman07}, and references therein).
86: As a result of the evolution, the redshift distribution of the XRB flux
87: is wide. Thus, the integral XRB comprises the information on the
88: large scale distribution of the X-ray sources over a wide redshift range.
89:
90: In the present paper the redshift distribution of the XRB photons is
91: investigated in detail. The analysis is based on an extensive observational
92: data selected from several published X-ray sky surveys. A convenient analytic
93: approximations are applied to model the observed redshift histograms of the
94: extragalactic X-ray sources selected at several flux levels. These
95: distributions are weighted by the source counts and summed up to obtain the
96: redshift distributions of the integral XRB.
97:
98: Next, the redshift distribution is used to define a relationship between the
99: XRB signal and the large-scale fluctuations of the matter spatial distribution.
100: The investigation has been raised by the recent report on the huge void
101: generating a dip in the surface brightness of the radio background
102: (\citealt{rudnick07}). This void, responsible for the deficit of the radio
103: surface brightness, allegedly generates also a cold spot in the CMB map via the
104: late-time integrated Sachs-Wolfe effect. Although the careful statistical
105: analysis by \cite{smith08} has not confirmed the existence of this
106: particular ``cold spot'' in the radio survey, a relationship between the large
107: scale features of the matter distribution and the integrated sky brightness in
108: various energy bands is a problem deserving some interest.
109:
110: Apart of the question of the XRB fluctuations induced by voids, the XRB
111: redshift distribution is interesting per se, as it allows to assess the
112: evolution of AGN phenomenon. A standard way to estimate a rate and type of this
113: evolution is based on the examination of the X-ray luminosity functions
114: determined at the consecutive redshift bins. Unfortunately, the X-ray surveys
115: produce flux-limited rather than luminosity-limited samples of sources. In
116: effect, luminosity functions at different redshifts cover different luminosity
117: ranges. This in turn severely impedes estimates of the luminosity function over
118: a wide range of luminosities and redshifts. The total level of nuclear
119: activity in galaxies within unit volume is given by the integral of the X-ray
120: luminosity function. The question of the AGN cosmic evolution constitutes one
121: of the central problems of observational cosmology, and has been investigated
122: for the last forty years (this issue was for the first time recognized by
123: \citealt{schmidt68}). Here a question of the AGN evolution is addressed
124: without the calculations of the X-ray luminosity function. The available
125: observational data on X-ray source counts and redshifts are used to evaluate
126: the redshift distribution as a function of source flux. This relationship and
127: the source counts allow to calculate the redshift distribution of the total XRB
128: and the integral luminosity density generated by the AGN as a function of
129: redshift.
130:
131: The organization of the paper is following. First, I present the formulae used
132: in calculations. Next, in Sec.\ref{catalogs}, the basic information on the
133: observational material extracted form the various archives is given. Since the
134: comprehensive characteristics of the data and the source catalogs are described
135: in the original papers, only the basic properties of the material are presented
136: here. The numerical fits to the observed distributions are obtained in
137: Sec.~\ref{fits}. In that section the calculations of the redshift distribution
138: of the XRB photons are described in details. These results are applied in the
139: Sec.~\ref{supervoids} to quantify the relationship between the voids and the
140: XRB variations. In Sec.~\ref{agn} the distribution of the XRB flux produced by
141: AGN is used to calculate the evolution of the AGN activity. Finally, potential
142: sources of errors inherent in the present method are discussed in the
143: Sec.~\ref{discussion}.
144:
145: The `canonical' standard cosmology is assumed throughout, with $H_0 =
146: 70$\,km\,s$^{-1}$Mpc$^{-1}$, $\Omega_m = 0.3$, and $\Omega_\Lambda = 0.70$.
147:
148: \section{\large Basic relationships \label{formulae}}
149:
150: In the present approach the X-ray source catalogs are used to construct
151: redshift distributions of the extragalactic sources as a function of the source
152: flux. At this stage, a question of the source (absolute) luminosities is not
153: addressed. Let $N(S)$ denotes the X-ray source counts, i. e. number of sources
154: brighter than $S$ in a unit solid angle, and $f_S(z) = dn(z\!\mid\!S)/d\log z$
155: is the redshift distribution of sources with flux $S$. Then, the redshift
156: distribution of the XRB surface brightness, $b(z)$, is equal to:
157:
158: \begin{equation}
159: b(z) = \frac{1}{b} \int\! dS\,f_S(z)\:S \left|\frac{dN(S)}{dS}\right|\,,
160: \label{bz}
161: \end{equation}
162: where the integration covers the entire ``interesting'' range of source
163: fluxes $S$ and $b$ denotes the integral background flux:
164: \begin{equation}
165: b = \int\! b(z)\;d\log z = \int\! dS \,S \left|\frac{dN(S)}{dS}\right|\,.
166: \label{b}
167: \end{equation}
168: It is assumed that the $f_S(z)$ distributions are normalized:
169: \begin{equation}
170: \int\! f_S(z)\;d\log z = 1\,.
171: \end{equation}
172: Here the integration limits cover the total range of redshifts occupied
173: by X-ray sources. The actual limits of the ``interesting'' range of
174: fluxes is discussed below.
175:
176: The luminosity density, $\varepsilon(z)$, i. e. a total luminosity $L$
177: generated in a unit comoving volume, $V$:
178:
179: \begin{equation}
180: \varepsilon(z) = \frac{dL}{dV}\,,
181: \end{equation}
182: is related to the flux distribution $b(z)$ and the luminosity distance,
183: $D_L(z)$:
184:
185: \begin{equation}
186: \varepsilon(z)\;\frac{dV}{d\log z} = 4\,\pi\,D_L^2(z)\:b(z)\,.
187: \label{epsilon}
188: \end{equation}
189:
190: The cosmological relationships between the comoving volume and the luminosity
191: distance in a flat space with $\Lambda \neq 0$ is given by \cite{hogg99}:
192:
193: \begin{equation}
194: \frac{dV}{dz} = \frac{c}{H_0}\:\frac{D_L^2}{(1+z)^2
195: \sqrt{\Omega_m (1+z)^3 + \Omega_\Lambda}}\,.
196: \label{dvdz}
197: \end{equation}
198: Combining Eqs.~\ref{epsilon} and \ref{dvdz} we finally get:
199: \begin{equation}
200: \varepsilon(z) = 4\pi\:\frac{H_0}{c}\;\frac{(1+z)^2
201: \sqrt{\Omega_m(1+z)^3 + \Omega_\Lambda}}{\ln\!10\;z}\; b(z)\,.
202: \label{lum_dens}
203: \end{equation}
204:
205: Thus, to calculate the distributions $b(z)$ and $\varepsilon(z)$, the source
206: counts $N(S)$ and the functions $n(z\,|\,S)$ have to be determined using the
207: observational material. In the next section the available X-ray surveys are
208: examined from this point of view.
209:
210: \section{\large Observational material \label{catalogs}}
211:
212: Because the high imaging efficiency of X-ray telescopes in the soft band and
213: numerous extensive identifications programs, we concentrate on the X-ray band
214: of $0.5 - 2.0$\,keV. At these energies a fraction of the background resolved
215: into discrete sources exceeds $90$\,\% and is higher than in the other bands
216: (e.g. \citealt{moretti03}, \citealt{brandt05}). Also a fraction of identified
217: objects with measured spectroscopic or photometric redshifts is relatively
218: high.
219:
220: Equation~\ref{bz} shows that the $b(z)$ distribution is sensitive to sources
221: which perceptibly contribute to the XRB. Consequently, one needs to calculate
222: the $n(z\,|\,S)$ functions over a quite wide range of fluxes. To achieve this
223: objective I have examined numerous X-ray surveys and selected several major
224: source catalogs for further analysis. The overall characteristics of those
225: catalogs are listed in Table~\ref{surveys}. A common name of the survey/catalog
226: is given in column 1. From the each catalog, sources for further processing
227: have been extracted within fixed range of fluxes defined on column 2. The
228: numbers of all sources, extragalactic and AGN with known redshifts are given in
229: columns 3, 4, and 5, respectively.
230:
231: Statistical requirements which have to be satisfied by the source samples to
232: properly determine the $n(z\,|\,S)$ distribution are different than those for
233: the luminosity function calculations. The individual sample should contain
234: sources from possible narrow range of fluxes, but the sample has not to be flux
235: limited. The sample provides unbiased estimate of the redshift distribution as
236: long as the process of identification and redshift measurements does not
237: introduce spurious correlation between flux and redshift.
238:
239:
240: \begin{table}
241: \caption{The X-ray surveys selected for the analysis}
242: \label{surveys}
243: \centering
244: \begin{tabular}{llccc}
245: \hline\hline
246: \noalign{\smallskip}
247: Name &\multicolumn{1}{c}{Flux limits}& \multicolumn{3}{c}{Number of sources} \\
248: &\multicolumn{1}{c}{(erg\,cm$^{-2}$ s$^{-1})$}& All &Extragalactic& AGN \\
249: \hline
250: \noalign{\smallskip}
251: RBS & $1.0\times 10^{-12}-5.0\times 10^{-11}$ & 1764 & 1054 & 681 \\
252: NEP & $5.0\times 10^{-14}-1.0\times 10^{-12}$ & 361 & 248 & 192 \\
253: RIXOS & $2.5\times 10^{-14}-5.0\times 10^{-13}$ & 393 & 318 & 235 \\
254: XMS & $1.0\times 10^{-14}-2.0\times 10^{-13}$ & 275 & 256 & 231 \\
255: CDFS & $5.0\times 10^{-17}-1.0\times 10^{-15}$ & 205 & 201 & 197 \\
256: CDFN & $1.5\times 10^{-17}-5.0\times 10^{-15}$ & 425 & 412 & 268 \\
257: \hline
258: \noalign{\smallskip}
259: \end{tabular}
260: \end{table}
261:
262: \subsection{\normalsize The {\it ROSAT} Bright Survey (RBS)}
263:
264: The identification program of the brightest sources detected in the {\it ROSAT}
265: All-Sky Survey, known as {\it ROSAT} Bright Survey, resulted in a sample of
266: 2072 sources with the total count rate above $0.2$\,s$^{-1}$
267: (\citealt{schwope00}). More than $99.5$\,\% of sources in the final catalogue
268: is identified. The survey covers high galactic latitudes ($|b|> 30\deg$).
269: After the removal of the Virgo clusters and Magellanic Clouds regions, the
270: catalog contains 2012 sources. In the energy band of $0.5-2.0$\,keV, 1773 RBS
271: sources generate flux between $1.0\times 10^{-12}$\,erg\,cm$^{-2}$s$^{-1}$
272: (hereafter \cgs) and $5.0\times 10^{-11}$ \cgs. As one might expect, only for
273: a small fraction of the RBS sources the redshifts are undetermined and
274: relatively large number of sources is associated with galactic sources, mostly
275: late type stars and cataclysmic variables.
276:
277: Since the RBS sample covers rather wide range of fluxes, it is useful for our
278: purposes to divide it into several subsamples with narrow flux limits and to
279: estimate the $n(z\,|\,S)$ distribution for the each set separately. We define
280: the bright source sample, labeled RBS(b), which contains sources with
281: $4.0\times 10^{-12} < S < 1.0\times 10^{-11}$ \cgs. Of $365$ RBS sources in
282: this flux range, $129$ is identified with galactic objects and for the other
283: $15$ sources redshifts have not been measured. Eventually, the sample contains
284: $221$ extragalactic objects with known redshifts. Nearly half of the sample,
285: viz. $97$ sources are identified with clusters of galaxies and normal galaxies.
286: The median flux\footnote{The samples extracted from the RBS span over narrow
287: ranges of fluxes and there is no major difference between the mean and median
288: values. In some other samples investigated in this paper the median flux is
289: distinctly smaller than the average. In those cases, the median flux is adopted
290: as the argument in the $n(z\,|\,S)$ functions.} in the extragalactic subsample
291: $S{\rm_m} = 5.5\times 10^{-12}$ \cgs.
292:
293: \begin{figure*}
294: \begin{center}
295: \includegraphics[width=0.9\textwidth]{z_hist_all_c.eps}
296: \caption{\leftskip 10mm \rightskip 10mm
297: {\small Histograms with labels -- the redshift distributions of extragalactic
298: sources in $9$ samples constructed in the present investigation. The data are
299: binned with $\Delta\log z = 0.125$; The integrals of all the histograms are
300: normalized to unity. The sample designations and the median fluxes in \cgs\
301: are given in the upper left corners; dotted curves -- best 3 parameter
302: fits for the each sample separately; solid curves -- the best model for all
303: the samples. The lower right panel: solid curve -- the model redshift
304: distribution for the whole XRB (the normalization of the distribution is the
305: same as for the histograms); dotted curve -- analytic approximation of the
306: model distribution; points -- the distribution obtained by the linear
307: interpolation between the raw histograms (see text for details).}}
308: \label{z_distr}
309: \end{center}
310: \end{figure*}
311:
312: The flux limits of $2.0\times 10^{-12} < S < 4.0\times 10^{-12}$ \cgs\ have
313: been adopted for the medium flux sample, RBS(m). Within this flux range the RBS
314: comprises of $669$ objects including $272$ galactic stars. Among the $397$
315: extragalactic sources, $233$ are identified with AGN with known redshifts. The
316: median flux in the RBS(m) extragalactic sample, $S_{\rm m} = 2.7\times
317: 10^{-12}$ \cgs. The faint source sample, RBS(f), contains sources with
318: $1.0\times 10^{-12} < S < 2.0\times 10^{-12}$ \cgs. Of $599$ RBS sources in
319: this flux range $268$ is identified with galactic objects and $24$ has unknown
320: redshifts. Thus, the RBS(f) sample contains $307$ sources extragalactic
321: objects with the median flux of $1.6\times 10^{-12}$ \cgs.
322:
323: The redshift distributions of sources in the RBS (b), (m), and (f) samples are
324: shown in three upper left panels in Fig.~\ref{z_distr}. The integrals of the
325: histograms for all the samples in Fig.~\ref{z_distr} are normalized to unity.
326: The distributions are plotted using logarithmic redshift bins with $\Delta \log
327: z = 0.125$. Each histogram is labeled with the survey name and the median flux
328: in \cgs. Analytic fits will be discussed in the next section.
329:
330: \subsection{\normalsize The {\it ROSAT} North Ecliptic Pole Survey (NEP)}
331:
332: The deepest exposure of the {\it ROSAT} All-Sky Survey (RASS) is centered at
333: the north ecliptic pole (\citealt{voges99}). The RASS of this region has been
334: used to construct statistically well defined sample of X-ray sources above a
335: flux limit $\sim\!2\times 10^{-14}$ \cgs\ (\citealt{henry06}) which have been
336: followed-up by the optical observations (\citealt{gioia03}). The identification
337: rate in the final catalog of $443$ sources is very high ($99.6$\,\%).
338:
339: Within the flux limits of $5.0\times 10^{-14}$ and $1.0\times 10^{-12}$ \cgs\
340: the NEP survey provided $361$ sources. After excluding $113$ galactic stars, we
341: are left with $248$ extragalactic sources; for $3$ sources the redshift is
342: unknown. The redshift histogram of $245$ sources (including $53$ clusters) is
343: shown in Fig.~\ref{z_distr}; the median flux in this sample is equal to
344: $1.2\times 10^{-13}$ \cgs.
345:
346: \subsection{\normalsize The {\it ROSAT} International X-ray/Optical Survey (RIXOS)}
347:
348: This {\it ROSAT} medium-sensitivity survey consists of sources found
349: in $82$ PSPC pointing observations at high galactic latitudes ($|b|>28\deg$).
350: A flux limit of $3\times 10^{-14}$ \cgs\ was adopted in $64$ fields and
351: $8\times 10^{-14}$ \cgs\ in the remaining $18$ fields. The source selection
352: procedures, optical identifications and the final catalog are given by
353: \cite{mason00}.
354:
355: For the purpose of the present analysis, $393$ sources with fluxes between
356: $2.5\times 10^{-14}$ and $5\times 10^{-13}$ \cgs\ have been selected. Within
357: these flux limits $75$ sources are associated with galactic stars. Of the
358: remaining $318$ sources, the redshifts of three objects are unknown, and $49$
359: sources are still unidentified. The redshift distribution of $266$ sources
360: (including $33$ clusters) is shown in Fig.~\ref{z_distr}. The median flux in this
361: subsample $S_{\rm m} = 5.2\times 10^{-14}$ \cgs.
362:
363: \subsection{\normalsize The {\it XMM-Newton} serendipitous survey (XMS)}
364:
365: The {\it XMM-Newton} serendipitous survey (XMS) has been constructed in a
366: similar way as the {\it RIXOS}. More than $300$ sources have been isolated in
367: $25$ high galactic latitude ($|b|>22\deg$) pointings covering
368: $\sim\!3$\,deg$^2$ of the sky (\citealt{barcons07}). In the $0.5-2.0$\,keV band
369: the sample is complete above $1.5\times 10^{-14}$ \cgs, and contains many
370: weaker sources.
371:
372: I have extracted from the original catalog $275$ sources with fluxes between
373: $1.0\times 10^{-14}$ and $2.0\times 10^{-13}$ \cgs. The sample is completely
374: identified; it contains $19$ stars and $2$ clusters of galaxies. For $23$
375: objects the redshifts are unknown. The redshift distribution of $233$ sources
376: is shown in Fig.~\ref{z_distr}. The median flux in this subsample $S_{\rm m} =
377: 1.9\times 10^{-14}$ \cgs.
378:
379: \subsection{\normalsize The Chandra Deep Field--South (CDFS)}
380:
381: The $1$\,Ms {\it Chandra} observations known as the {\it Chandra} Deep Field
382: South are described by \cite{giacconi02}. The catalog of sources detected in
383: this field by two independent algorithms contains $304$ objects, of which $275$
384: have determined fluxes in the $0.5-2.0$\,keV band. For further processing $205$
385: sources with fluxes in the range $5.0\times 10^{-17} - 1.0\times 10^{-15}$
386: \cgs\ have been selected. Four sources are identified with stars. The redshifts
387: either spectroscopic (\citealt{szokoly04}, \citealt{ravikumar07}) or
388: photometric (\citealt{zheng04}) are known for $200$ sources; one source remains
389: unidentified.
390:
391: The redshift distribution is shown in Fig.~\ref{z_distr}. The median flux in
392: this sample $S_{\rm m} = 2.1\times 10^{-16}$ \cgs. The sample contains one
393: galaxy group; two other sources apparently are not associated with the activity
394: in the galactic nuclei (\citealt{lehmer06}). It is likely, however, that more
395: objects in the CDFS survey should be classified as off-nuclear sources. The
396: available data do not allow for unambiguous separation of AGN and off-nuclear
397: sources at the low flux levels in the CDFS. This reservation holds also for the
398: CDFN samples below.
399:
400: \subsection{\normalsize The Chandra Deep Field--North (CDFN)}
401:
402: The ultra deep Chandra field, $2$\,Ms exposure, CDFN, resulted in a catalog
403: of $503$ sources detected over $0.12$ sq.\,deg. (\citealt{alexander03}).
404: Optical follow-up observations by \cite{barger03} have rendered a large
405: number of spectroscopic and photometric redshifts. Several more redshifts
406: are taken from \cite{reddy06}, \cite{donley07} and \cite{georgakakis07}.
407:
408: In the present investigation,
409: the CDFN catalog has been divided into two samples of bright (b) and faint (f)
410: sources. The (b) sample contains $181$ sources between $2.5\times 10^{-16}$ and
411: $5.0\times 10^{-15}$ \cgs; ten sources have been identified with the galactic
412: stars; at least one has been categorized as `starburst' galaxy
413: (\citealt{georgakakis07}), for $49$ objects the redshifts have not been
414: measured. The sample consists of $122$ sources, mostly AGN. The median flux
415: in this subsample $S_{\rm m} = 7.3\times 10^{-16}$ \cgs.
416:
417: The faint CDFN sample has been selected between $1.5\times 10^{-17}$ and
418: $2.5\times 10^{-16}$ \cgs. Among $244$ sources satisfying these flux
419: limits, three sources have been identified with stars, $51$ -- with the
420: starburst galaxies (\citealt{georgakakis07}) and for $43$ objects the redshifts
421: have not been measured. The final sample used in the calculations contains $198$
422: sources with $147$ confirmed AGN. The median flux in the sample
423: $S_{\rm m} = 9.0\times 10^{-17}$ \cgs. The redshift
424: histograms for the (b) and (f) samples are shown in Fig.~\ref{z_distr}.
425:
426: The numbers of objects unidentified or without redshift are in some samples
427: quite large. Hence one could expect that the corresponding redshift histograms
428: are not representative for the whole population of sources at given flux. Below
429: this question is de facto worked out where we construct an analytic function
430: which simultaneously fits all the histograms.
431:
432: \section{\large Approximations and fits \label{fits}}
433:
434: The redshift distribution of sources selected at fixed flux, $n(z|S)$,
435: is a intricate function of a number of parameters, such as the luminosity
436: function, the relationship between the luminosity and observable flux, and the
437: relationship between the volume and redshift. The luminosity function itself
438: depends on redshift and both the latter relationships depend on the cosmological
439: model. However, the existing estimators of the $n(z|S)$ function represented by
440: the nine histograms in Fig.~\ref{z_distr} are strongly affected/degraded by the
441: statistical noise. It implies that a simple analytic function with $2-3$ free
442: parameters will provide a statistically satisfactory fit to the observed
443: distributions.
444:
445: It appears that the histograms in Fig.~\ref{z_distr} are adequately reproduced
446: by:
447: \begin{equation}
448: f_S(z) = f_0\;z^\alpha\:e^{-z/z_{\rm c}}\,,
449: \label{fit_1}
450: \end{equation}
451: where $f_0=f_0(S)$, $\alpha = \alpha (S)$, and $z_{\rm c}=z_{\rm c}(S)$ are
452: three parameters fitted to the the histograms $n(z|S_i)$, $i = 1,... ,9$. In
453: Fig.~\ref{z_distr} the least square fits for all the distributions are shown
454: with the dotted curves. Apart from a few pronounced features visible in the
455: plots which represent the large scale structures reported in the literature
456: (e.g. \citealt{barger02}, \citealt{gilli03}), analytic fits seem to adequately
457: reproduce the observed distributions.
458:
459: It is found that only $z_{\rm c}$ is strongly correlated with $S_{\rm m}$,
460: while the fits do not indicate any statistically significant correlation
461: between $\alpha$ and $S_{\rm m}$. In Fig.~\ref{z_c_vs_s} the best fit values of
462: $\alpha$ are shown with crosses. The labels and scale on the left-hand ordinate
463: refer to $z_{\rm c}$, and on the right-hand -- to $\alpha$. Since the
464: simultaneous fitting of $\alpha$ and $z_{\rm c}$ introduces a spurious
465: correlation between these two parameters, the $\alpha$ parameter has been fixed
466: at the average value found for the $9$ samples, $\bar{\alpha} = 1.934$.
467: Effectively, it means that the shape of the $n(z|S)$ function is fixed and the
468: only dependence on $S$ is limited to the horizontal shift along the $z$ axis.
469: In Fig.~\ref{z_c_vs_s} the best fit parameters $z_{\rm c}$ found for the
470: fixed $\alpha$ are plotted against the median flux $S_{\rm m}$. In agreement
471: with the expectations, the $z_{\rm c}$ increases with diminishing flux $S_{\rm
472: m}$ over a wide range of fluxes. However, a clear flattening of the
473: relationship is observed below $\sim\!10^{-14}$ \cgs. This apparent absence of
474: correlation between $z_{\rm c}$ and $S$ results from the well-known fact that
475: in the X-ray surveys at low flux levels the maximum detected redshift remains
476: stable while significantly increases fraction of intrinsically weak sources.
477:
478: \begin{figure}
479: \begin{center}
480: \includegraphics[width=0.6\textwidth]{s_med_z_cut_all_c.eps}
481: \caption{\leftskip 10mm \rightskip 10mm
482: {\small Distributions of the best fitted parameters $\alpha$ and $z_{\rm c}$
483: vs. the median flux, $S_{\rm m}$. Points with error bars -- $z_{\rm c}$
484: (left-hand ordinate), crosses -- $\alpha$ (right-hand ordinate). The assumed
485: $z_{\rm c} \sim S$ relationships used to model the XRB redshift distribution is
486: shown with the dotted lines (see text for details).}}
487: \label{z_c_vs_s}
488: \end{center}
489: \end{figure}
490:
491: The points in the $z_{\rm c} - S_{\rm m}$ relationship above $\sim 10^{-14}$
492: \cgs\ seem to follow the power law. In the subsequent calculations it is
493: assumed that this relationship is in fact well approximated by the power law in
494: the whole range of fluxes between $10^{-14}$ and $10^{-11}$ \cgs, although the
495: data coverage is rather sparse. Below $S \approx 10^{-14}$ \cgs\ the data are
496: insufficient to delineate precisely the $z_{\rm c} - S_{\rm m}$ relationship. I
497: have assumed tentatively that $z_{\rm c}$ remains constant and is equal to the
498: average value found for three {\it Chandra} samples. The $z_{\rm c}$ in
499: CDFN(b), CDFS and CDFN(f) are equal to $0.69$, $0.69$ and $0.51$, respectively.
500: The average weighted by the uncertainties $\bar{z}_c = 0.61$. In
501: Fig.~\ref{z_c_vs_s} the model relationship $z_{\rm c} - S$ used in the
502: calculations is shown with the dotted line.
503:
504: The $z_{\rm c} \sim S$ relationship with the fixed $\alpha$ parameter and
505: fixed normalization of the integral:
506: \begin{equation}
507: \int f_S(z)\;d\log z = 1\,,
508: \end{equation}
509: eliminates formally any free parameters in the fitting the analytic function
510: to the nine histograms in Fig.~\ref{z_c_vs_s}:
511: \begin{equation}
512: f_S(z)=\frac{\ln 10}{\Gamma(\alpha)\;z_{\rm c}^\alpha}\;z^\alpha\;e^{-z/z_{\rm c}}\,,
513: \label{fit_2}
514: \end{equation}
515:
516: where $\alpha = 1.934$, $\Gamma(\alpha) = 0.9739$ is the gamma function and
517: $z_{\rm c}$ is specified for each sample by the $z_{\rm c} \sim S$
518: relationship. Analytic distributions defined in Eq.~\ref{fit_2} are shown in
519: Fig.~\ref{z_distr} with the solid curves. In all the histograms the model
520: distribution is astonishingly close to the corresponding best three-parameter
521: fit represented by the dotted curves. Most deviations visible in some plots are
522: easily explained by the statistical nature of the problem and/or the large
523: scale structures present in the catalogs based on the localized sky area
524: (\citealt{barger02}, \citealt{gilli03}). Systematic shifts between the fits are
525: present in three histograms below $S = 10^{-14}$ \cgs. It is a direct result of
526: the assumption of a single $z_{\rm c} = 0.61$ value for all three {\it Chandra}
527: samples. It is noticeable that the constant width (in $\log z$) model fits
528: adequately represent the data over the full range of fluxes. Small differences
529: in the width between the three-parameter fits and the final model which are
530: visible in the NEP, RIXOS and XMS data, apparently do not represent the
531: systematic effects. In the NEP and RIXOS histograms the final model is slightly
532: narrower than the individual fits, while in the XMS sample it is wider.
533:
534: To effectively use the Eq.~\ref{bz} one needs the representation of the source
535: counts $dN(S)/dS$ over the whole range of fluxes $S$. The parametrization by
536: \cite{moretti03} adequately suits the present calculations. The smooth
537: functional form for $N(S)$ proposed by Moretti et al. accurately reproduces the
538: observed counts below $10^{-11}$ \cgs\ down to {\it Chandra} threshold of
539: $\sim\!2\times 10^{-17}$ \cgs. Sources within these flux limits generate more
540: than $90$\,\% of the XRB and smooth extrapolation of the \cite{moretti03}
541: counts down to $\sim\!3\times 10^{-18}$ \cgs\ is consistent with the entire
542: XRB. Substituting all the components into Eq.~\ref{bz} we finally get the
543: redshift distribution of the XRB photons. It is shown with the solid curve in
544: the bottom right panel in Fig.~\ref{z_distr}. The same normalization has been
545: applied to facilitate comparison with the distributions derived for the
546: individual samples. Points in the plot are discussed below in the
547: Sec.~\ref{discussion}.
548:
549: A suitable representation of the $b(z)$ distribution has been found using
550: a smooth function of the same form as for the individual redshift histograms.
551: The function:
552: \begin{equation}
553: b_{\rm fit}(z) = 5.24\;z^{1.52}\;e^{-z/0.63}\,,
554: \end{equation}
555: reproduces the derived distribution of $b(z)$ with the relative error
556: of less than $4$\,\% for $0.06 < z < 6$. It is shown with dotts
557: in the bottom right panel (with normalization rescaled to conform to
558: all the plots in Fig.~\ref{z_distr}).
559:
560: \section{\large XRB and Supervoids \label{supervoids}}
561:
562: The distribution of the XRB photons $b(z)$ peaks at redshift $z\approx 1$ and
563: $50$\,\% of the background originates between the redshifts of $0.4$ and $1.4$
564: (for $80$\,\% the redshift limits are $0.2$ and $2.1$). Thus, very large
565: structures of the matter distribution at redshift within these limits would
566: generate fluctuations of the integral XRB. As an example I discuss below the
567: X-ray signature of the huge void postulated by \cite{rudnick07}. The arguments
568: based on the radio survey in favor of the void with a radius of
569: $\sim\!140$\,Mpc in Eridanus have been questioned (\citealt{smith08}).
570: Nevertheless, the Integrated Sachs-Wolfe effect operating on extremely large
571: structures of matter remains a valid explanation of the strongest CMB
572: fluctuations.
573:
574: The redshift separation, $\Delta z$, corresponding to the far side and the near
575: side of the void with diameter $R_o$ centered at redshift $z$ is equal to
576: (e.g. \citealt{hogg99}):
577: \begin{equation}
578: \Delta z = \frac{2 R_o}{c / H_o} \sqrt{\Omega_m (1+z)^3 + \Omega_\Lambda}\,.
579: \end{equation}
580: Using the $b(z)$ distribution we assess the fractional deficit of the XRB
581: $\delta = \Delta b / b$ created by the completely empty region of size
582: $280$ Mpc. Such void would generate $|\delta| = 5.7$\,\% and $\delta = 4.9$\,\%
583: at redshifts $z=0.5$ and $1$, respectively. Assuming spherical shape of the
584: void, its angular diameter would be $8\fdg 4$ and $4\fdg 7$ at these redshifts.
585: The XRB depression produced by the void should be compared to the XRB
586: intrinsic fluctuations
587: resulting from the discrete nature of sources generating the background.
588: Assuming purely random distribution of sources, the rms fluctuations
589: of the XRB are defined by the source counts $N(S)$:
590: \begin{equation}
591: \sigma_b = \left[ \int_{S_{\rm min}}^{S_{\rm max}} \;dS\:S^2\:\omega\;
592: \left|\frac{dN(S)}{dS}\right|\;\right]^{\;1/2}\,,
593: \end{equation}
594: where $\omega$ is the solid angle subtended by the investigated area. Using the
595: the \cite{moretti03} counts and $S_{\rm max} = 1\cdot 10^{-11}$ \cgs\ (the
596: amplitude of the XRB fluctuations is dominated by the contribution of sources
597: at the bright end of counts), we get $\delta b/b = 0.035$ and $0.024$ for the
598: circular areas of radius $2^\circ$ and $3^\circ$, respectively. Thus, at
599: $z=0.5$ the signal-to-noise ratio for the void detection amounts to
600: $\sim\!2.4$. In the case of $z=1$ the S/N drops to just $1.4$. To reduce the
601: amplitude of the XRB fluctuations one should remove from the XRB the
602: contribution of bright sources. If the $S_{\rm max}$ is decreased to $1\cdot
603: 10^{-12}$ \cgs, the significance of the void signal reaches $3.5\,\sigma$ at
604: $z=0.5$ and $2.0\,\sigma$ at $z=1$. So, only the low redshift voids would
605: produce the XRB deficits significantly stronger than the statistical
606: fluctuations.
607:
608: \section{\large Redshift distribution of the XRB and the AGN evolution \label{agn}}
609:
610: To assess the distribution of the XRB generated just by the AGN, I have
611: repeated all the procedures described in the previous sections using the
612: samples constructed exclusively from the AGN. The AGN sources are easily
613: separated from the clusters and nearby normal galaxies. However, the
614: distinction between the nuclear activity and stellar emitters in the case of
615: distant and weak sources becomes problematic. Such sources are present in both
616: {\it Chandra} surveys. One should keep in mind this limitations in the present
617: investigation. Nevertheless, the AGN are a dominating constituent of all the
618: samples exploited in the paper and even the moderate contamination of the AGN
619: subsamples with the off-nuclear sources would not affect significantly our
620: calculations (see below). The numbers in the AGN samples are smaller than in
621: the full samples and parameter estimates are subject to slightly larger
622: uncertainties. Clusters and normal galaxies populate on the average lower
623: redshift bins and the histograms for the AGN analogous to those in
624: Fig.~\ref{z_distr} are shifted towards the higher redshifts. We notice also a
625: weak correlation of the best fit $\alpha$ parameter with the source flux -- a
626: shape of of the redshift distribution (in $\log z$ bins) varies with $S_{\rm
627: m}$. In Fig.~\ref{z_c_vs_s_agn} the values of $\alpha$ in the nine samples are
628: shown with crosses. The regression line of $\log\alpha$ on $\log S$ is used to
629: fix the value of $\alpha$ for the each sample and to calculate the best fit
630: parameter $z_{\rm c}$. These new $z_{\rm c}$ are shown in
631: Fig.~\ref{z_c_vs_s_agn} with the squares. Finally, the best fit line $\log
632: z_{\rm c} \sim \log S_{\rm m}$ is calculated for six brighter samples to obtain
633: $z_{\rm c}$ for $S > 1.9\cdot 10^{-15}$ \cgs. A constant $z_{\rm c}$ is assumed
634: for lower fluxes, and the complete $z_{\rm c} \sim S_{\rm m}$ relationship
635: adopted for further computations is shown in Fig.~\ref{z_c_vs_s_agn} with the
636: dotted lines.
637:
638: \begin{figure}
639: \begin{center}
640: \includegraphics[width=0.6\textwidth]{s_med_z_cut_agn_c.eps}
641: \caption{\leftskip 10mm \rightskip 10mm
642: {\small Distributions of best fitted parameters $\alpha$ and $z_{\rm c}$ vs.
643: the median flux, $S_{\rm m}$, in the AGN samples. Points with error bars --
644: $z_{\rm c}$ (left-hand ordinate), crosses -- $\alpha$ (right-hand ordinate).
645: The dashed line represents the regression line for the $\alpha \sim S_{\rm m}$
646: relationship; the $z_{\rm c} \sim S$ function used to model the AGN
647: contribution to the XRB is shown with the dotted lines (see text for details).
648: The vertical bars at the bottom mark the effective flux limits of the analyzed
649: samples; numbers indicate a relative contribution to the XRB generated by the
650: AGN within these limits.}}
651: \label{z_c_vs_s_agn}
652: \end{center}
653: \end{figure}
654:
655: The X-ray source counts used in the present case should be limited to the AGN
656: only. Two other major classes of sources contributing to the counts are
657: associated with clusters and normal/starburst galaxies. The analytic formula
658: obtained by \cite{moretti03} quite accurately represents counts of all the
659: types of extragalactic sources, but the relative contribution of the each class
660: in the total counts is not well established. One should notice, however, that
661: most of the XRB is produced by sources in the middle range of fluxes considered
662: here, while the cluster contribution is significant only at the bright end of
663: counts and the normal and starburst galaxies populate mostly the faint end of
664: counts. Vertical bars at the bottom of Fig.~\ref{z_c_vs_s_agn} divide
665: the flux range of $10^{-17} - 10^{-11}$ \cgs\ into $9$ contiguous bands
666: corresponding approximately to fluxes surveyed by the source samples defined in
667: the paper. The numbers between the bars give the relative contribution of each
668: flux band to the total XRB.
669:
670: To extract the cluster and starburst galaxies contributions we corrected the
671: total counts in the following way. In the RBS(b) sample clusters constitute
672: $40$\,\% of all the extragalactic sources. The slope of the cluster counts at
673: the bright end amounts approximately to $1.3$ (\citealt{degrandi99}).
674: Substantially flatter slope than that for the total counts reduces the relative
675: cluster contribution at lower fluxes. Although the cluster counts are not well
676: constrained below $\sim\!10^{-12}$ \cgs, their contribution to the total counts
677: drops at the faint end of counts to a negligible level. The source counts
678: attributed to the AGN are assessed by subtracting the cluster counts from the
679: total counts defined by the \cite{moretti03} formula. The normal and starburst
680: galaxies are relatively abundant in the CDFN(f) sample. Of $241$ extragalactic
681: sources, $147$ have been classified as `AGN', $51$ as `starburst' and for the
682: other $43$ the redshift is unknown. The absolute maximum content of the
683: non-AGN sources in the CDFN(f) sample amounts to $(51+43)/241 \equiv 39$\,\%,
684: assuming that all sources with undetermined redshift are starburst galaxies.
685: The amount of the non-AGN sources at higher flux levels drops quickly. In the
686: XMS sample none extragalactic source with known redshift has been classified as
687: normal or starburst galaxy. The maximum possible contribution of the starburst
688: galaxies at the low flux levels has been accounted for by flattening the slope
689: of the \cite{moretti03} counts below $10^{-14}$ \cgs\ to reproduce the
690: reduction of the AGN in the CDFN(f) sample by $39$\,\%. The counts modified
691: this way have been substituted into Eq.~\ref{bz} to obtain the redshift
692: distribution, $b_{\rm AGN}(z)$.
693:
694: \begin{figure}
695: \begin{center}
696: \includegraphics[width=0.7\textwidth]{lg_ve_z.eps}
697: \caption{\leftskip 10mm \rightskip 10mm
698: {\small The cosmic evolution of the X-ray luminosity density generated by AGN
699: as a function of: (a) - redshift $z$, (b) - $\log(1+z)$ and (c) - the cosmic time.
700: The plots are based on the soft X-ray emission only and do not include
701: radiation absorbed and re-emitted in different energies. Points with the
702: error bars in the panel (a) are taken from \cite{hasinger05} and represents
703: only the type-1 AGN.}}
704: \label{evolution}
705: \end{center}
706: \end{figure}
707:
708: One can use the observed distribution of the background flux produced by the
709: AGN to calculate the cosmic history of the luminosity density generated by
710: these objects. The $b_{\rm AGN}(z)$ distribution is inserted into
711: Eq.~\ref{lum_dens} which relates the cosmological evolution of the X-ray
712: luminosity density, $\varepsilon(z)$, to the redshift distribution of the
713: background, $b(z)$. Variations of the luminosity density obtained this way are
714: shown in Fig.~\ref{evolution} with the solid curves. The data are displayed in
715: three panels as a function of redshift, $z$, logarithm of $(1+z)$ and the
716: cosmic time, assuming $t_0 = 13.47\cdot 10^9$ years for the present age of the
717: Universe\footnote{For the cosmological model defined in Sec.~\ref{intro} and
718: using the formulae given by \cite{hogg99}.}.
719:
720: The accuracy of the present $\varepsilon(z)$ estimate depends strongly
721: on a quality of our $b(z)$ fits. Relatively small numbers of sources
722: at redshifts below $\sim\!0.03$ and above $\sim\!3$ generate large
723: statistical fluctuations and weakly constrains the analytic fits $b(z)$
724: in these redshift ranges. Hence, the present estimates of $\varepsilon(z)$
725: are also subject to large uncertainties at low and high redshifts.
726:
727: In order to assess the importance of the $N(S)$ uncertainties on the present
728: estimates of $\varepsilon(z)$, I have plotted in Fig.~\ref{lum_dens} with the
729: dotted curves the $\varepsilon(z)$ function using the original \cite{moretti03}
730: formula, i.e. assuming no corrections for clusters and starburst galaxies. The
731: discrepancies between both solutions do not exceed $20$\,\% for redshifts below
732: $\sim\!3$. It implies that our procedure to isolate the contribution of AGN
733: from the total counts, albeit crude, does not contribute significantly to the
734: final errors of $\varepsilon(z)$.
735:
736: \section{\large Discussion \label{discussion}}
737:
738: The main objectives of the present investigation, viz. estimates of the redshift
739: distribution of the XRB photons, $b(z)$, and the evolution of the AGN
740: luminosity density, $\varepsilon(z)$, have been achieved using the smooth,
741: analytic fits to the observed source redshift histograms. The present method
742: is conceptionally simple and computationally straightforward. Unfortunately,
743: it does not provide error estimates. The major sources of uncertainties
744: have been indicated in the previous section. Here a quantitative estimate
745: of the errors is discussed.
746:
747: The errors of the present measurement of $b(z)$ are generated by the
748: statistical nature of the investigated material and a chain of approximations
749: applied to substitute the observed redshift distributions centered on a
750: selected fluxes by an analytic function $f_S(z)$ continuous in both parameters,
751: $z$ and $S$. In fact, the visual inspection of the analytic fits displayed
752: with the solid curves in Fig.~\ref{z_distr} reveals some deviations from the
753: redshift histograms. To estimate the significance of these differences, the
754: calculations have been performed using the actual histograms shown in
755: Fig.~\ref{z_distr} with broken solid lines instead of $f_S(z)$. For the each
756: value of flux $S$ in the range $10^{-17} - 10^{-11}$ \cgs, the corresponding
757: redshift distribution has been obtained by the linear interpolation between two
758: histograms from the samples centered on the median fluxes nearest to $S$. The
759: results of this procedure are shown in the lower right panel of
760: Fig.~\ref{z_distr} with dots. Generally good agreement between the
761: distribution of points and the solid curve proves that the analytic
762: approximations do not introduce perceptible systematic errors in the present
763: investigation. It appears that the relatively large deviations for three data
764: points (centered at redshifts: $0.087$, $0.65$, and $1.16$) result purely
765: from the large scale structures. This is particularly likely for the first
766: bin ($0.075 < z < 0.1$), where the discrepancy between the fits is
767: produced entirely by the excess of sources in the localized NEP survey.
768:
769: One should also notice, that the uncertainties of our main results are only
770: weakly affected by the limited statistics of the individual samples and
771: histograms. This is because the final distributions are obtained by averaging
772: the individual distributions and this procedure effectively reduces statistical
773: fluctuations.
774:
775: In the upper panel of Fig.~\ref{evolution} the AGN emissivity calculated by
776: \cite{hasinger05} is shown. The points with the error bars are redrawn here
777: from their original paper. \cite{hasinger05} apply more stringent criteria to
778: select sources and use only well defined samples of type-1 AGN. Optically these
779: objects are identified by the broad Balmer emission lines, while using the
780: X-ray criteria, they have unabsorbed spectra indicating low intrinsic column
781: densities. In the present analysis I have included all objects in which the
782: X-ray emission originates in the active nuclei. Thus, our results cannot be
783: compared directly to those by \cite{hasinger05}. Nevertheless, despite entirely
784: different method applied in the present paper, the distributions show good
785: agreement at redshifts above $\sim\!1$. Although most of the apparent
786: discrepancies, which reach a factor of $3$ at $z\approx 0.5$, are probably due
787: to the distinct selection criteria of both investigations, one cannot exclude
788: that some differences are caused by unrecognized systematic effects inherent in
789: one or both methods.
790:
791: The present method of the luminosity density calculations has also some
792: disadvantages. In our approach the absolute luminosities of the individual
793: objects are not determined. Consequently, only the integral luminosity density
794: is obtained, and the cosmic evolution of any selected AGN luminosity class
795: has to be studied by means of the standard methods.
796:
797: \vspace{2mm}
798: \ni ACKNOWLEDGEMENTS\\
799: This work has been partially supported by the Polish MNiSW grant N~N203~395934.
800:
801: \begin{thebibliography}{}
802: \vspace{-10pt}
803: \bibitem[Alexander et al.(2003)]{alexander03}
804: Alexander, D. M., Bauer, F. E., Brandt, W. N., Schneider, D. P.,
805: Hornschemeier, A. E., et al. 2003, AJ, 126, 539
806: \vspace{-10pt}
807: \bibitem[Barcons et al.(2007)]{barcons07}
808: Barcons, X., Carrera, F. J., Ceballos, M. T., Page, M. J., Bussons-Gordo, J.,
809: et al. 2007, AA, 476, 1191
810: \vspace{-10pt}
811: \bibitem[Barger et al.(2002)]{barger02} Barger, A. J., Cowie, L. L., Brandt, W. N.,
812: Capak, P., Garmire, G. P., et al. 2003, AJ, 124, 1839
813: \vspace{-10pt}
814: \bibitem[Barger et al.(2003)]{barger03} Barger, A. J., Cowie, L. L., Capak, P.,
815: Alexander, D. M., Bauer, F. E., et al. 2003, AJ, 126, 632
816: \vspace{-10pt}
817: \bibitem[Brandt \& Hasinger (2005)]{brandt05}
818: Brandt, W. N. \& Hasinger, G. 2005, ARA\&A, 43, 827
819: \vspace{-10pt}
820: \bibitem[De Grandi et al.(1999)]{degrandi99}
821: De Grandi, S., B\"ohringer, H., Guzzo, L., Molendi, S., Chincarini, G.,
822: et al. 1999, ApJ, 514, 148
823: \vspace{-10pt}
824: \bibitem[Donley et al.(2007)]{donley07} Donley, J. L., Rieke, G. H.,
825: P\'erez-Gonz\'alez, P. G., Rigby, J. R., \& Alonso-Herrero, A. 2007,
826: ApJ, 660, 167
827: \vspace{-10pt}
828: \bibitem[Georgakakis et al.(2007)]{georgakakis07} Georgakakis, A.,
829: Rowan-Robinson, M., Babbedge, T. S. R., \& Georgantopoulos, I. 2007,
830: MNRAS, 377, 203
831: \vspace{-10pt}
832: \bibitem[Giacconi et al.(2002)]{giacconi02}
833: Giacconi, R., Zirm, A., Wang, J-X., Rosati, P., Nonino, M., et al.
834: 2002, ApJS, 139, 369
835: \vspace{-10pt}
836: \bibitem[Gilli et al.(2003)]{gilli03} Gilli, R., Cimatti, A., Daddi, E.,
837: Hasinger, G., Rosati, P., et al. 2003, ApJ, 592, 721
838: \vspace{-10pt}
839: \bibitem[Gioia et al.(2003)]{gioia03}
840: Gioia, I., Henry, J., Mullis, C., B\"ohringer, H., Briel, U., et al.
841: 2003, ApJS, 149, 29
842: \vspace{-10pt}
843: \bibitem[Hasinger et al.(2005)]{hasinger05}
844: Hasinger, G., Miyaji, T, \& Schmidt, M., 2005, AA, 441, 417
845: \vspace{-10pt}
846: \bibitem[Henry et al.(2006)]{henry06}
847: Henry, J., Mullis, C., Voges, W., B\"ohringer, H., Briel, U., et al.
848: 2006, ApJS, 162, 304
849: \vspace{-10pt}
850: \bibitem[Hogg (1999)]{hogg99} Hogg, D. W. 1999, {\it astro-ph/9905116}
851: \vspace{-10pt}
852: \bibitem[Kim et al.(2007)]{kim07}
853: Kim, M., Wilkes, B. J., Kim, D.-W., et al. 2007, ApJ, 659, 29
854: \vspace{-10pt}
855: \bibitem[Lehmann et al.(2001)]{lehmann01}
856: Lehmann, I., Hasinger, G., Schmidt, M., et al. 2001, A\&A, 371, 833
857: \vspace{-10pt}
858: \bibitem[Lehmer et al.(2006)]{lehmer06}
859: Lehmer, B. D., Brandt, W. N., Hornschemeier, A. E., Alexander, D. M.,
860: Bauer, F. E., et al. 2006, AJ, 131, 2394
861: \vspace{-10pt}
862: \bibitem[Mason et al.(2000)] {mason00}
863: Mason, K. O., Carrera, F. J., Hasinger, G., Andernach, H, Aragon-Salamanca, A.,
864: et al. MNRAS, 311, 456
865: \vspace{-10pt}
866: \bibitem[Miyaji et al.(2000)]{miyaji00}
867: Miyaji, T., Hasinger, G., \& Schmidt, M. 2000, ApJ, 353, 25
868: \vspace{-10pt}
869: \bibitem[Moretti et al.(2003)]{moretti03}
870: Moretti, A., Campana, S., Lazzati, D., \& Tagliaferri, G. 2003, ApJ, 588, 696
871: \vspace{-10pt}
872: \bibitem[Ravikumar et al.(2007)]{ravikumar07}
873: Ravikumar, C. D., Puech, M., Flores, H., Proust, D., Hammer, H., et al.
874: 2007, AA, 465, 1099
875: \vspace{-10pt}
876: \bibitem[Reddy et al.(2006)]{reddy06} Reddy, N. A., Steidel, C. C., Erb,
877: D. K., Shapley, A. E., \& Pettini, M. 2006, ApJ, 653, 1004
878: \vspace{-10pt}
879: \bibitem[Rudnick et al.(2007)]{rudnick07}
880: Rudnick, L., Brown, S., \& Williams L. R., 2007, ApJ, 671, 40
881: [Erratum: ApJ, 678, 1531 (2008)]
882: \vspace{-10pt}
883: \bibitem[Schmidt (1968))]{schmidt68}
884: Schmidt, M. 1968, ApJ, 151, 398
885: \vspace{-10pt}
886: \bibitem[Schwope et al.(2000)]{schwope00}
887: Schwope, A. D., Hasinger, G., Lehmann, I., Schwarz, R., Brunner, H.,
888: et al. 2000, AN, 321, 1
889: \vspace{-10pt}
890: \bibitem[Silverman et al.(2007)]{silverman07}
891: Silverman, J. D., Green, P. J., Barkhouse, W. A., et al. 2007,
892: {\it astro-ph/0710.2461}
893: \vspace{-10pt}
894: \bibitem[Smith \& Huterer(2008)]{smith08}
895: Smith, K, M. \& Huterer, D., 2008, {\it arXiv:0805.2751}
896: \vspace{-10pt}
897: \bibitem[Stocke et al.(1991)]{stocke91}
898: Stocke, J. T., Morris, S. L., Gioia, I. M., Maccacaro, T., Schild, R.,
899: et al. 1991, ApJS, 76, 813
900: \vspace{-10pt}
901: \bibitem[Szokoli et al.(2004)]{szokoly04}
902: Szokoly, G. p., Bergeron, J., Hasinger, G., Lehmann, I., Kewley, L., et al.
903: 2004, ApJS, 155, 271
904: \vspace{-10pt}
905: \bibitem[Voges et al.(1999)]{voges99}
906: Voges, W., Aschenbach, B., Boller, Th., Br\"auninger, H., Briel, U., et al.
907: 1999, AA, 349, 389
908: \vspace{-10pt}
909: \bibitem[Zheng et al.(2004)]{zheng04}
910: Zheng, W., Mikles, V. J., Mainieri, V., Hasinger, G., Rosati, P., et al.
911: 2004, ApJS, 155, 73
912: \end{thebibliography}
913:
914: \end{document}
915:
916: