X-ray source counts in the COSMOS field
N. Cappelluti, G. Hasinger, M. Brusa, A. Comastri, G. Zamorani, H. Boehringer, H. Brunner, F. Civano, A. Finoguenov, F. Fiore, R. Gilli, R. E. Griffiths, V. Mainieri, I. Matute, T. Miyaji, J. Silverman
aa r X i v : . [ a s t r o - ph ] A p r **FULL TITLE**ASP Conference Series, Vol. **VOLUME**, **YEAR OF PUBLICATION****NAMES OF EDITORS** X-ray source counts in the XMM-COSMOS survey
N. Cappelluti , G. Hasinger , M. Brusa , A. Comastri , G. Zamorani ,H. B¨ohringer , H. Brunner , F. Civano , , A. Finoguenov , F. Fiore ,R. Gilli , R. E. Griffiths , V. Mainieri , I. Matute , T. Miyaji ,J. Silverman Abstract.
We present the analysis of the source counts in the XMM-COSMOSsurvey using data of the first year of XMM-
Newton observations. The surveycovers ∼ within the region of sky bounded by 9 h . m < R.A. < h . m ;1 d . m < DEC < d . m with a total net integration time of 504 ks. Using amaximum likelihood algorithm we detected a total of 1390 sources at least in oneband. Using Monte Carlo simulations to estimate the sky coverage we producedthe logN-logS relations. These relations have been then derived in the 0.5–2 keV,2–10 keV and 5–10 keV energy bands, down to flux limits of 7.2 × − erg cm − s − , 4.0 × − erg cm − s − and 9.7 × − erg cm − s − , respec tively. Theserelations have been compared to previous X-ray survey and to the most recentX-ray background model finding an excellent agreement. The slightly differentnormalizations observed in t he source counts of COSMOS and previous surveyscan be largely explained as a co mbination of low counting statistics and cosmicvariance introduced by the large scale structure.
1. Introduction
A solid knowledge of the X-ray source counts is fundamental to fully under-stand the AGNs population of the X-ray background (XRB). At present thetwo deepest X–ray surveys, the
Chandra
Deep Field North (CDFN; Bauer et al.2004) and
Chandra
Deep Field South (CDFS; Giacconi et al. 2001), have ex-tended the sensitivity by about two orders of magnitude in all bands with re-spect to previous surveys (Hasinger et al. 1993; Ueda et al. 1999; Giommi et al.2000), detecting a large number of faint X–ray sources. However, deep pencilbeam surveys are limited by the area which can be covered to very faint fluxes(typically of the order of 0.1 deg ) and suffer from significant field to field vari-ance. To ovecome this problem, shallower surveys over larger areas have beenundertaken in the last few years with both Chandra (e.g. the 9 deg Bootessurvey (Murray et al. 2005), the Extended Groth strip EGS (Nandra et al.2005), the Extended
Chandra
Deep Field South E-CDFS, (Lehmer et al. 2005;Virani et al. 2006) and th e Champ (Green et al. 2004; Kim et al. 2004)) andXMM–
Newton (e.g. the HELLAS2XMM survey (Fiore et al. 2003), the XMM– Max Planck Institut f¨ur Extraterrestrische Physik, D-85478 Garching, Germany INAF-Osservatorio Astronomico di Bologna, via Ranzani 1, I-40127 Bologna, Italy Department of Physics, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213 Dipartimento di Astronomia, Universit`a di Bologna, via Ranzani 1, I–40127 Bologna, Italy INAF-Osservatorio astronomico di Roma, Via Frascati 33, I-00044 Monteporzio Catone, Italy Nico Cappelluti et al.
Newton
BSS (Della Ceca et al. 2004) and the ELAIS S1 survey (Puccetti et al.2006) ). In this context the XMM–
Newton wide field survey in the COS-MOS field (Scoville et al. 2006), hereinafter XMM–COSMOS (Hasinger et al.2006), has been conceived and designed to maximize the sensitivity and sur-vey area product, and is expected to provide a major step forward toward acomplete characterization of the physical properties of X–ray emitting SuperMassive Black Holes (SMBHs). In this work we concentrate on the first year ofXMM-
Newton observations (AO3) of the COSMOS field (Hasinger et al. 2006).For more details see Cappelluti et al. (2007).
2. The X-ray logN-logS
The source detection was performed using a maximum likelihood algorithm pro-vided with the XMM–
Newton
Standard Analysis System (SAS) in the 0.5–2keV, 2–10 keV and 5–10 keV bands. A total of A total of 1281, 724 and 186point-like sources were detected in the three bands down to limiting fluxes of7.2 × − erg cm − s − , 4.7 × − erg cm − s − and 9.7 × − erg cm − s − , respectively. Using Monte Carlo simulation to determine the sky cover-age, source number counts can be easily computed using the following equation: N ( > S ) = P N S i =1 1Ω i deg − , where N S is the total number of detected sourcesin the field with fluxes greater than S and Ω i is the sky coverage associatedwith the flux of the i th source. The cumulative number counts, normalized tothe Euclidean slope (multiplied by S . ), are shown in Figure 1, for the 0.5–2keV, 2–10 keV and 5–10 keV energy ranges, respectively. We performed a bro-ken power-law maximum likelihood fit to the unbinned differential counts in the0.5–2 keV and 2–10 keV energy bands. In the 0.5–2 keV energy band the bestfit parameters are α =2.60 +0 . − . , α =1.65 ± . S b =1.55 +0 . − . × − erg cm − s − and A =123. Translating this value of the normalization to that for thecumulative distribution at 2 × − erg cm − s − , which is usually used in theliterature for Chandra surveys, we obtain A ∼
450 which is fully consistent withmost of previous works where a fit result is presented. In the 2–10 keV bandthe best fit parameters are α =2.43 ± . α =1.59 ± . S b =1.02 +0 . − . × − erg cm − s − and A =266. The latest value translates into A ∼ σ . In the5–10 keV energy bands, where the differential counts do not show any evidencefor a break in the sampled flux range, we assumed a single power-law modelfor which the best fit parameters are found to be A =102 and α =2.36 ± . ± . − s − keV − .The corresponding values in the 2–10 keV and 5–10 keV bands are 4.7 ± . − s − keV − and 2.6 ± . − s − keV − . Therefore XMM-COSMOSresolves by itself ∼ ∼
40% and ∼
22% of the XRB into discrete sources inthe 0.5–2 keV, 2–10 keV and 5–10 keV energy bands, respectively. In Figure1 we compared our logN-logS to those predicted by the recent XRB model ofGilli, Comastri & Hasinger (2006).The amplitude of source counts distributions varies significantly among dif-ferent surveys (see e.g. Yang et al. 2003; Cappelluti et al. 2005, and referencestherein). This ”sample variance”, can be explained and predicted as a com- -ray source counts in the COSMOS field Figure 1. The X-ray logN-logS compared to the survey listed in thelabels in the 0.5–2 keV, 2–10 keV and 5–10 keV energy bands, respec-tively
T op Lef t, T op Right and Bottom Lef t P anels . The black − dashed line represents the prediction of Gilli, Comastri & Hasinger (2006). Bottom Right P anel : The counts in cell fluctuations within the XMM-COSMOS field on different angular scales. The dashed lines represent the1 σ expected fluctuation. bination of Poissonian variations and effects due to the clustering of sources(Peebles 1980; Yang et al. 2003). In order to determine whether the differencesobserved in the source counts of different surveys could arise from the clusteringof X-ray sources, we estimated the amplitude of the fluctuations from our data,by producing subsamples of our survey with areas comparable to those of. e.g., Chandra surveys.The XMM-COSMOS field and the Monte Carlo sample fields of Section 4 weredivided in 4,9,16 and 25 square boxes. Making use of the 0.5–2 keV energy banddata, we computed for each subfield, the ratio of the number of real sources tothe number of random source. Both the random and the real sample were cutto a flux limit of 5 × − erg cm − s − . In the lower right panel of Figure 1we plot the ratio of the data to the random sample as a function of the size ofthe cells under investigation. The predicted fractional standard deviations are Nico Cappelluti et al. , 0.19 deg , 0.11 deg and 0.07deg , respectively. The measure fractional standard deviations are 0.09, 0.20,0.21 and 0.24 on the same scales, respectively. The ratio of the clusting to thePoissonian variance is expected to scale as σ cl /σ p ∝ N . θ ( γ − / a (3 − γ ) / . Wetherefore conclude that at this scales and at fluxes sampled here the major con-tribution to source counts fluctuations is due to Poissonian noise. This analysisis at least qualitatively consistent with Figure 2, which shows a significantlylarger dispersion in the data from different surveys in the hard band than inthe soft band. Moreover, the results here discussed are also consistent with theobserved fluctuations in the deep Chandra fields (see, for example, Bauer et al.2004). Large area, moderately deep surveys like XMM-COSMOS are neededto overcome the problem of low counting statistics, typical of deep pencil beamsurveys, and, at the same time, to provide a robust estimate of the effect of largescale structure on observed source counts. Acknowledgments.
This work is based on observations obtained with XMM–
Newton , an ESA science mission with instruments and contributions directlyfunded by ESA Member States and NASA; also based on data collected at theCanada-France-Hawaii Telescope operated by the National Research Council ofCanada, the Centre National de la Recherche Scientifique de France and theUniversity of Hawaii.