Source of Acquisition Goddard Space Flight Center The Hard X-ray 20-40 keV AGN Luminosity Function V. Beckmannl NASA Goddard Space Flight Center, Exploration of the Universe Division, Code 661, Greenbelt, MD 20771, USA beckmannQmilkyway.gsfc.nasa.gov S. S ~ l d i ~C>. R~. ,S hrader4y5,N . Gehrels4, and N. Produit2 ABSTRACT We have compiled a complete, significance limited extragalactic sample based on N 25,000 deg2 to a limiting flux of 3 x lo-'' ergs cm-2 s-l (- 7,000deg2 to a flux limit of IO-'' ergs cmP2s -l) in the 20 - 40 keV band with INTEGRAL. We have constructed a detailed exposure map to compensate for effects of non-uniform exposure. The flux-number relation is best described by a power-law with a slope of Q = 1.66 k 0.11. The integration of the cumu- lative flux per unit area leads to f20-40keV = 2.6 x lo-'' ergs crn-2 s-l sr-', which is about 1% of the known 20 - 40 keV X-ray background. We present the first luminosity function of AGN in the 20-40 keV energy range, based on 68 extragalactic objects detected by the imager IBIS/ISGRI on-board INTEGRAL. The luminosity function shows a smoothly connected two power-law form, with an index of y1 = 0.9 below, and 72 = 2.2 above the turn-over luminos- ity of L, = 4.6 x ergs s-'. The emissivity of all INTEGRAL AGNs per unit volume is W20-40ke"(> 1041 ergss-l) = 2.8 x lo3* ergss-' h;o M ~ c - ~T.h ese results are consistent with those derived in the 2 - 20keV energy band and do not show a significant contribution by Compton-thick objects. Because the sample used in this study is truly local (E = 0.022), only limited conclusions can be drawn for the evolution of AGNs in this energy band. But the objects explaining the peak in the cosmic X-ray background are likely to be either low luminosity .4GN (Lx < 1041e rgss-') or of other type, such as intermediate mass black holes, clusters, and star forming regions. Subject headings; galaxies: active - gamma rays: observations - X-rays: galaxies - surveys -- galaxies: Seyfert 1. Introduction 'also with the Joint Center for Astrophysics, De- partment of Physics, University of Maryland, Baltimore The Galactic X-ray sky is dominated by accret- County, MD 21250, USA ing binary systems, while the extragalactic sky 21NTEGRAL Science Data Centre, Chemin d' Ecogia shows mainly active galactic nuclei (AGN) and 16, 1290 Versoix, Switzerland clusters of galaxies. Studying the population of 3also with Observatoire de Geneve, 51 Ch. des Mail- sources in this energy range has been a challenge lettes, 1290 Sauverny, Switzerland ever since the first observations by rocket borne 4NASA Goddard Space Flight Center, Exploration of the Universe Division, Code 661, Greenbelt, MD 20771, X-ray detectors (Giacconi et al. 1962). At soft USA X-rays (0.1 - 2.4 keV) the ROSAT All-Sky Survey 5also with Universities Space Research Association, (RASS; Voges et al. 1999) has revealed an extra- 10211 Wincopin Circle, Columbia, MD 21044, USA galactic population of mainly broad line AGNs, such as type Seyfert 1 and quasars. Between a 1 c - few keV and lMeV, no all-sky survey using vations of the ROSAT deep X-ray surveys, which imaging instruments has been performed to date. showed that 90% of the 0.5 - 2.0 keV CXB can be In the 2 - 10 keV range surveys based on signif- resolved into AGNs (Miyaji, Hasinger & Schmidt icant fractions of the sky have been carried out 2000). At higher energies (2 - lOkeV), ASCA with ASCA (e.g. Ueda et al. 2001), XMM-Newton and Chandra surveys measured the X-ray lumi- (e.g. Hasinger 2004), and Chandra (e.g. Brandt nosity function (XLF). These studies show that et al. 2001) and have shown that the dominant in this energy range the CXB can be explained extragalactic sources are more strongly absorbed by AGNs, but with a higher fraction of absorbed than those within the RASS energy band, thus (NH> 1022cm-2) objects than in the soft X-rays more Seyfert 2 type objects are detectable here. (e.g. Ueda et al. 2003). A study based on the For a summary on the deep X-ray surveys below RXTEsurvey by Sazonov & Revnivtsev (2004) de- 10 keV see Brandt & Hasinger (2005). At higher rived the hard X-ray luminosity function of AGNs. energies the data become more scarce. The Rossi They showed that the summed output of AGNs in X-ray Timing Explorer (RXTE)s ky survey in the this energy range is too small to explain the CXB, 3 - 20 keV energy band revealed 100 AGNs, show- and suggested that a comparable X-ray flux may ing an ever higher fraction of absorbed sources of be produced together by lower luminosity AGNs, about 60% (Sazonov & Revnivtsev 2004). A sim- non-active galaxies and clusters of galaxies. ilar study at energies above 20 keV had not been With the on-going observations of the sky by possible, because a large field of view in combi- INTEGRAL, a sufficient amount of data is now nation with sufficient sensitivity are required to available to derive the AGN hard X-ray lumi- study the AGN population on a relevant fraction nosity function. In this paper we present anal- of the sky. The International Gamma-Ray Astro- ysis of recent observations performed by the IN- physics Laboratory (INTEGRAL; Winkler et al. TEGRAL satellite, and compare the results with 2003) offers an unprecedented > 20keV collect- previous studies. In Section 2 we describe the ing area and state-of-the-art detector electronics AGN sample and in Section 3 the methods to de- and background rejection capabilities. Notably, rive the number-flux distribution of INTEGRAL the imager IBIS with an operating range from AGNs are presented togethcr with the analysis of 20 - 1OOOkeV and a fully-coded field of view of their distribution. Section 4 shows the local lumi- 10" x 10" enables us now to study a large por- nosity function of AGNs as derived from our data, tion of the sky. A first catalog of AGNs showed a followed by a discussion of the results in Section similar fraction of absorbed objects as the RXTE 5. survey (Beckmann et al. 2006a). Related to the compilation of AGN surveys in 2. The INTEGRAL AGN Sample the hard X-rays is the question of what sources Observations in the X-ray to soft gamma- form the cosmic X-ray background (CXB). While ray domain have been performed by the in- the CXB below 20 keV has been the focus of many struments on-board the INTEGRAL satellite studies, the most reliable measurement in the 10 - 500 keV has beer! provided by the High Energy (Winkler et al. 2003). This mission offers the unique potential to perform simultaneous observa- Astronomical Observatory (HEAO l),l aunched in tions over the 2 - 8000 keV energy region. This is 1977 (Marshall et al. 1980). The most precise achieved by the X-ray monitor (2-30 keV) JEM- measurement provided by the UCSD/MIT Hard X-ray and Gamma-Ray instrument (HEAO 1 A- X (Lund et al. 2003), the soft gamma-ray im- ager (20-1000 keV) ISGRI (Lebrun et al. 2003), 4) shows that the CXB peaks at an energy of about 30 keV (Gruber et al. 1999). The isotropic and the spectrograph SPI (Vedrenne et al. 2003), nature of the X-ray background points to an ex- which operates in the 20 - 8000 keV region. Each tragalactic origin, and as the brightest persistent of these instruments employs the coded-aperture sources are AGNs, it was suggested earlyon that technique (Caroli et al. 1987). In addition to those objects are the main source of the CXB (eg. these data an optical monitor (OMC, Mas-Hesse Zdziarski 1996). In the soft X-rays this concept et al. 2003) provides photometric measurements has been proven to be correct through the obser- in the V band. 2 r 1 Data used for the analysis presented here were and are not used in our analysis. all in the public domain by the end of March 2005. This includes data from orbit revolutions 19 - 137 3. Number-Flux Distribution of INTE- and revolutions 142 - 149. Data before revolution GRAL AGNs 19 have been excluded as the instruments settings In order to compute the AGN number-flux re- changed frequently and therefore the data from lation counts of AGNs it is necessary to have a this period are not suitable to be included into a complete and unbiased sample, and to know the homogeneous survey. characteristics of the survey used to extract the The list of sources was derived from the analysis data. Because of the in-homogeneous character of as described in Beckmann et al. (2006a). Addi- the survey, we had to apply a significance limit tional observations performed later led to further rather than a flux limit to define a complete sam- source detections within the survey area. We ex- ple. The task is to find a significance limit which tracted spectra at those positions from the data ensures that all objects above the given limit have following the same procedure. It is understood been found. To test for completeness, the V,/Va- that most of those objects did not result in a sig- test has been applied. nificant detection in the data set used here, but it ensures completeness of the sample at a moderate The V,/V,-test is a simple method developed by Avni & Bahcall (1980) based on the V/Vma, significance limit. test of Schmidt (1968). V , stands for the vol- The analysis was performed using the Offline va ume, which is enclosed by the object, and is Science Analysis (OSA) software version 5.0 dis- the accessible volume, in which the object could tributed by the ISDC (Courvoisier et al. 2003a). have been found (e.g. due to a flux limit of a We applied the same method for IBIS/ISGRI and survey). Avni & Bahcall showed that different SPI analysis as described in Beckmann et al. survey areas with different flux limits in various (2006a). The analysis of the INTEGRALIIBIS energy bands can be combined by the V,/V,-test. data is based on a cross-correlation procedure be- In the case of no evolution (Ve/Va) = 0.5 is ex- tween the recorded image on the detector plane pected. This evolutionary test is applicable only and a decoding array derived from the mask pat- to samples with a well-defined significance limit tern (Goldwurm et al. 2003). The ISGRI spectra down to which all objects have been found. It can have been extracted from the count rate and vari- therefore also be used to test the completeness of ance mosaic images at the position of the source, a sample. We performed a series of V,/V,-tests which in all cases corresponds to the brightest to the INTEGRAL .4GN sample, assuming com- pixel in the 20 - 40 keV band. pleteness limits in the range of 1c up to 60 IS- The significances listed in Tab. 1 have been de- GRI 20 - 40 keV significance. For a significance rived by using the OSA software, and refer to the limit below the true completeness limit of the sam- count rate and count rate error for ISGRI in the ple one expects the V,/V,-tests to derive a value 20 .- 40 keV energy band. The luminosities listed (Ve/l’a) < (Ve/Va)true, where (Ve/Va)tTue is the are the observed luminosities in this energy band. true test result for a complete sample. Above the The absorption listed is the intrinsic absorption completeness limit the (Ve/Va) values should be in units of 10’2cm-2 as measured in soft X-rays distributed around (V, /Va)true within the statis- by various missions as referenced. We also include tical uncertainties. An example of this method to the most important reference for the INTEGRAL determine the significance limit can be found in data of the particular source in the last column of Beckmann et al. (2006a). Table 1. The extracted images and source results It appears that the sample becomes complete are available in electronic form’. at a significance cutoff of approximately 50, which In order to provide a complete list of AGNs inclu*de s 38 AGNs. The average value is (Ve/Va) = detected by INTEGRAL, we included also those 0.45 0.05. This is consistent with the expected sources which are not covered by the data used value of 0.5 which reflects no evolution and an even for our study. Those sources are marked in Tab. 1 distribution in the local universe. A pure number flux distribution (i.e. logN,s http://heasarc.gsfc.nasa.gov/docs/integral/inthp-archive.html J versus logs) for the sample presented here would is detectable with a 50 detection significance in the not give meaningful results, because of the differ- 20-40 keV energy band. The resulting correlation ent exposure times across the survey, and there- is shown in Fig. 4. fore the varying sensitivity within the survey. An uncorrected number flux distribution for INTE- 3.1. The Slope of the Number-Flux Dis- GRAL AGNs has been shown in Beckmann et tribution al. (2006b). To correct for the different expo- We applied a maximum-likelihood (ML) algo- sure times it is necessary to count the number rithm to our empirical number-flux distribution of AGNs per unit sky area. Thus the number to obtain a power-law approximation of the form of AGNs above a given flux have to be counted N(> S) = K . S-". Our approach was based on and divided by the sky area in which they are de- the formalism derived by Murdoch, Crawford and tectable throughout the survey. We therefore first Jauncey (1973), also following the implementation determine-d the exposure time in 64,620 sky ele- of Piccinotti et a1 (1982). The htter involved mod- ments of 0.63deg2 size within our survey. In ification of the basic ML incorporated to facilitate each sky bin, the exposure is the sum of each indi- handling of individual source flux-measurement vidual exposure multiplied by the fraction of the errors. The ML method also involves the applica- coded field of view in this particular direction. The tion of a Kolmogorov-Smirnov (K-S) test to eval- dead time and the good time intervals (GTI) are uate the goodness of fit, as detailed in Murdoch, not taken into account but the dead time is fairly Crawford and Jauncey (1973). Once the slope is constant (around 20%) and GTI gaps are very rare determined, a chi-square minimization is used to in IBIS/ISGRI data. Figure 1 shows the exposure deterimine the amplitude K. map in Galactic coordinates for this survey. We For this analysis, we used a sub-sample of 38 excluded those fields with an exposure time less sources for which the statistical significance of our than 2 ks, resulting in 47,868 sky elements with flux determinations was at a level of 50 or greater. a total coverage of 9.89sr. The flux limit for a The dimmest source among this sub-sample was given significance limit should be a function of the fx = 5.6 x ergs cm-' s-', and the brightest square root of the exposure t,ime, if no systemat.ic was fx = 3.2 x 10-'0ergscm-2ss-'. We derived effects apply, but this assumption cannot be made a ML probability distribution, which can be ap- here. The nature of coded-mask imaging leads to proximated by a Gaussian, with our best fit pa- accumulated systematic effects at longer exposure rameters of a = 1.66 f 0.11. A normalization times. In order to achieve a correlation between of K = 0.44 sr-' ( ergs cmP2s -')" was then the exposure time and the flux limit, we therefore obtained by performing a least-squares fit, with used an empirical approach. We correlated the ex- the slope fixed to the ML value. posure time of each object with an equivalent flux correspoding to 5u significance based on the flux 4. The Local Luminosity Function of AGNs and significance of each sample object. By using at 20 - 40 keV only the AGNs of our survey we are additionally assured to consider effects based on the spectral The complete sample of INTEGRAL AGNs r slope (for AGNs 21 2). The correlation was then with a detection significance 2 5a also allows us fit by a smooth polynomial (Fig. 2). This function to derive the density of these objects in the local was then used to estimate the flux limit of each Universe as a function of their luminosity. In or- survey field. It has to be noted that the individ- der to derive the density of objects above a given ual flux limit of each survey field is not important, luminosity, one has to determine for each source but only the correct distribution of those flux lim- in a complete sample the space volume in which its. The total area in the survey for a given flux this source could have been found considering both limit is shown in Figure 3. the flux limit of each survey field and the apparent Based on the flux limits for all survey fields, we brightness of the object. We have again used the are now able to construct the number flux distri- correlation between exposure time and flux limit I bution for the INTEGRAL AGN, determining for as discussed in the previous section in order to as- each source flux the total area in which the source sign a 5a flux limit to each survey field. Then the 4 maximum redshift zmaZa t which an object with 2.23f0.15, and L, = (4.6f2.0) x h;: ergs s-'. flux fx would have been detectable in each sky The errors have been estimated on the basis of the element was used to compute the total accessible uncertainties of the density determination and of volume the flux measurement. These values are consis- tent with values derived from the 2 - 10 keV XLF N - . of AGNs as shown by e.g. Ueda et al. (2003) and La F'ranca et al. (2005). For example the work by Ueda et al. (2003) reveals for a pure with N being the number of sky elements in which density evolution model the same values for A, the object would have been detectable and 0, the 71, and 72, but a higher L, = 1.29 x 1044ergs-1. solid angle covered by sky element i. We applied a This ratio of 2.8 in the turnover luminosity can cosmology with HO = 70 km s-' Mpc-l (h70 = l), be easily explained by the different energy bands k = 0 (flat Universe), Rmatter = 0.3, and A0 = 0.7, applied. A single power law with photon index although a A0 = 0 and qo = 0.5 cosmology does of I' = 1.9 in the range 2 - 40 keV would lead to not change the results significantly because of the L(2-lokev)/L(zo-40kev) = 2.8, assuming no intrin- low redshifts in our sample. Figure 5 shows the sic absorption. Thus it appears that the local lu- cumulative luminosity function for 38 INTEGRAL minosity function of AGNs in the 20 - 40 keV band detected (2 5~7)A GNs in the 20 40 keV energy - can simply be extrapolated from the 2 - 10 keV band. Here the density 4 describes the number range. This has, of course, no implications for of objects per Mpc3 above a given luminosity Lx: the XLF at higher redshifts. The values are also K q5 = VaT: with K being the number of objects consistent with the luminosity function for AGNs i= 1 in the 3 - 20 keV band as derived by Sazonov & with luminosities > Lx. Blazars have been ex- Revnivtsev (2004) from the RXTE all-sky survey. cluded because their emission is not isotropic. The Information about intrinsic absorption is avail- redshifts in the sample range from z = 0.001 to able for 34 of the 38 objects (89%) from soft X-ray z = 0.129 with an average redshift of Z = 0.022. observations. This enables us to derive the lumi- Thus the luminosity function is truly a local one. nosity function for absorbed (NH 2 1022cm-2) Figure 6 shows the luminosity function in differen- and unabsorbed sources, as shown in Figure 7. tial form, indicating tha.t the sample indeed does The absorbed sources have a higher density than not show luminosity bins with large incomplete- the unabsorbed sources at low luminosities, while ness compared to the rest of the sample. The this trend is inverted at high luminosities. The lu- errors are based on the number of objects con- minosity where both -4GY types have similar den- tributing to each value. The differential XLF also shows, like the cumulative one, a turnover around sities is about L(20-40 keV) = 3 x erg s-l. This tendency is also evident when comparing the frac- Lx = (5 - IO) x io43e rgss-'. tion of absorbed -4GNs with the luminosity in the Because our study is based solely on low red- three luminosity bins depicted in Figure 8. The shift objects, we are not able to constrain models luminosity bins have been chosen so that an equal involving evolution with redshift. Nevertheless we number of objects are contained in each bin. The can compare the XLF presented here with model position of the data point along the luminosity axis predictions from previous investigations. XLFs indicates the average luminosity in this bin, while are often fit by a smoothly connected two power- the error bars in luminosity indicate the range of law function of the form (Maccacaro et al. 1991) luminosities covered. Based on the luminosity function, the contribu- tion of the AGNs to the total X-ray emissivity W can be estimated (Sazonov & Revnivtsev 2004). This can be done by simply multiplying the XLF by the luminosity in each bin and integrating We fit this function using a least-squares over the range of luminosities (lo4' ergs sP1 < method applying the Levenberg-Marquardt algo- rithm. The best fit values we obtained are A = L20-40keV < 1045.5e rgs s-I). This results in W20-40kev(> 1041e rgss-') = 2.8~10~~ergshs+-ol (2.7f0.5) x hpo MPC-~7,1 = 0.93*0.15,72 = 5 . = 5. Discussion A turnover j the XLF at 4.6 x ergss-' is observed (Fig. 6). Below this luminosity also The number flux distribution (Fig. 4) shows the fraction of absorbed AGNs starts to be larger a slope of cy = 1.66 f 0.11. In the local Uni- than that of the unabsorbed ones, although the verse with no evolutionland isotropic distribution effect is significant only on a la level (Fig. 7). of AGN and assuming Eucledian geometry, the ex- Both effects have been seen also in the 2 - 10 keV pected value is cy = 1.5 Even though the difference (Ueda et al. 2003; La Franca et al. 2005) and in is only significant on a 1.5a level, this might indi- the 3-20 keV band (Sazonov & Revnivtsev 2004). cate that the area density at the low flux end of the This implies that we do detect a similar source distribution has been slightly overcorrected. One population as at lower energies. If a large pop- has to keep in mind that only a few sources derived ulation of absorbed AGN-s is dominating the cos- from a small area of the sky are constraining the mic X-ray background at 30 keV as indicated by low flux end. Krivonos et al. (2005) studied the HEAO 1 A-4 measurements (Gruber et al. 1999), extragalactic source counts as observed by INTE- and the source population is the same through- GRAL in the 20-50 keV energy band in the Coma out the Universe those objects would have to region. Based on 12 source detections they deter- have luminosities L(20--40 keV) < ergs s-l , as mine a surface density of (1.4 6 0.5) x deg-2 might be indicated by the larger fraction of ab- above a threshold of lo-" ergs cm-2 s-l in the sorbed sources toward lower luminosities (Fig. 7). 20 - 50 keV energy band, where we get a consis- Even though it has to be taken into account that tent value of (1.2 f 0.2) x deg2. Comparing the low luminosity end of the XLF is based only the total flux of all the objects in the AGN sam- on a small number of objects, below this lumi- ple (f20-40ke~ = 2.6 x lO-'Oergs cm-2 s-l sr-l) nosity the distribution between active and nor- with the flux of the X-ray background as pre- mal galaxies becomes blurred. Additionally ob- sented by Gruber et al. (1999) shows that the jects like Ultra-luminous X-ray sources (ULX) INTEGRAL AGN account only for about 1% of and star-forming galaxies could provide the nec- the expected value. This is expected when tak- essary emission. One interesting case in this cat- ing into account the high flux limit of our sample: egory is the detection of NGC 4395 with a lu- La Franca et al. (2005) have shown that objects minosity of L(20-40ke~) = 1.4 x 1040ergss-1, with f2-1okev > lo-" ergs cm-2 s-l contribute consistent with measurements by XMM-Newton less than 1%t o the CXB. This flux limit translates which showed L(2-1OkeV) = 1.5 x 1040ergss-1 to the faintest flux in our sample of f20--40kev = (Vaughan et al. 2005). The central engine of this r 5.6 x ergs cmT2s -' for a = 1.9 power law galaxy has been classified as an i'ultra-luminousll spectrum. source, possibly associatied with an intermediate We compared the emissivity per unit volume of mass black hole with MBH = (4'6, x 104Ma our objects W20-40kev(> 104'ergss-') = 2.8 x (McHardy et al. 2005). In addition, NGC 4395 ergss-' h30 M ~ c w- i~th that found in the 3- harbors a ULX with L(&*OkcV) = lo3' ergs s-l at 20 keV band. Assuming an average power law of a distance of 2.9 kpc from the center of the galaxy r = 1.9, the extrapolated value is W3-20keV(> (Colbert & Ptak 2002). 1041 ergs s-l) = 6.7 x ergs s-l h;o M ~ c - ~ , Another scenario is that the population of X- which is a factor of 1.3 larger than the value ray emitting sources depends on redshift, i.e. that measured by RXTE (Sazonov & Revnivtsev 2004) there is an evolution of population in time, indi- and consistent within the la error. Extrapolating cating that the fraction of absorbed sources might our result down to 2-10 keV shows an emissiv- be higher at larger redshifts, alt,hough it should ity which is lower by a factor of 1.2 compared be noted that the latter effect, is not clearly de- to the one derived from the HEAO-1 all-sky map tectable in the 2 - 10keV range. The fraction (Miyaji et al. 1994) and also within the statistical of absorbed sources seems to depend on lumi- uncertainties. nosity (Ueda et al. 2003; Treister & Urry 2005), The luminosity function derived from the IN- as seen also in the 20 - 40keV band (Fig. 8). TEGRAL 20 - 40 keV AGN sample appears to be But some studies come to the conclusion that consistent with the XLF in the 2 - 20 keV range. there is no evolution of A r ~(U eda et al. 2004; 6 4 Treister & Urry 2005), while others find the frac- are more likely to belong to the Galaxy: the Sec- tion of absorbed sources increasing with redshift ond IBIS/ISGRI Soft Gamma-Ray Survey Cata- (La Franca et al. 2005). The latter also find that log (Bird et al. 2006) lists 55 new sources detected a combination of effects (the fraction of absorbed by INTEGRAL, of which 93% are located within AGN decreases with the intrinsic X-ray luminos- -10" < b < f10". Among those sources 3 are ity, and increases with the redshift) can be ex- listed as extragalactic sources, 18 are of Galactic plained by a luminosity-dependent density evo- origin, and 29 have not been identified yet. lution model. They further show that the lu- In addition, those objects which have been minosity function o-f AGNs like those presented demonstrated to be AGNs, are no more likely to here peaks at z - 0.7 while high luminosity be Compton-thick objects than the overall AGN AGN peak at z 2. Unified models also pre- population studied here. Only four AGNs (NGC dict, depending on the applied model, a frac- 1068, NGC 4945, MRK 3, Circinus galaxy) de- tion of 0.6 - 0.7 for high flux low redshift AGN tected by INTEGRAL have been proven to be (Treister & Urry 2005). Worsley et al. (2005) Compton thick objects so far, and none of them examined Chandra and XMM-Newton deep fields showed absorbtion of NH > 5 x 1024cm-2. In and come to the conclusion that the missing CXB order to clarify this point, observations at soft X- component i-s formed by highly obscured AGNs rays of those objects without information about at redshifts 0.5 - 1.5 with column densities of intrinsic absorption are required for all INTE- the order of fx = - cm-2. Evidence for GRAL detected AGNs (Tab. 1). Up to now 23 this scenario is also found in a study of Chandra % of the INTEGRAL AGN are missing absorp- and Spitzer data (Polletta et al. 2006). Combin- tion information. A first indication of what the ing multiwavelength data, this work estimates a absorption in these sources might be, can be de- surface density of 25 AGNdegF2 in the infrared rived from comparison of the INTEGRAL fluxes in the 0.6deg2 ChandralSWIRE field, and only with ROSAT All-Sky Survey (RASS) Faint Source 33% of them are detected in the X-rays down Catalogue data (Voges et al. 2000). In order to to f0.3-8 kev = ergs cm-2 s-I. The work do so we assumed a simple power law with photon also indicates a higher abundance of luminous and index r = 2.0 between the ROSAT 0.1 - 2.4keV Compton-thick AGN at higher redshifts (z >> 0.5). band and the INTEGRAL 20 - 40 keV range and This source population would be missed by the fit the absorption. In the six cases where no detec- study presented here, because of the low redshifts tion was achieved in the RASS, an upper limit of (E = 0.022) of the INTEGRAL AGNs. f(0.1--2 4k&) 5 10-~~ergscms--' ~ has been as- Several studies (Ueda et al. 2004; Treister & Urry 2005bumed, resulting in a lower limit for the absorp- propose that the absorbed AGNs needed to ex- tion NH > (5 - 11) x cm-2. In Fig. 9 we show plain the CXB should be Compton thick, and the distribution of intrinsic absorption. It has to therefore would have been missed at 2 - 10 keV. be pointed out that the estimated values can only This argument does not hold for the INTEGRAL give an idea about the distribution of intrinsic ab- observations, where the impact of absorption is sorption and should not be taken literally, as the much less severe than at lower energies. The ef- spectral slope between the measurements is un- fect on the measured flux of a source with pho- known and the observations are not simultaneous. r ton index = 2 for Compton thick absorption Nevertheless apparently none of the RASS detec- (NH = loz4c m-2) is only a 5% decrease in flux tions and non-detections requires an intrinsic ab- (40% for NH = 1025cm-2). It is therefore un- sorption of NH > 2 x 1023cm-2. Therefore it likely that many Compton-thick objects have been appears unlikely that a significant fraction of IN- missed by the INTEGRAL studies performed to TEGRAL AGNs will show an intrinsic absorption date. One possibility would be, that they are NH > loz4 A similar result in the 3-20 keV among the newly detected sources found by IN- band let Sazonov & Revnivtsev (2004) to the con- TEGRAL. The fraction of unidentified objects clusion that, the missing emission to explain the among those sources is of the order of 50%. It CXB is not produced in 'normal' AGNs, but that should be pointed out though, that most of these a comparable X-ray flux might be produced to- sources are located close to the Galactic plane and gether by low luminosity AGNs, non-active galax- 7 ies and clusters of galaxies. by Worsley et al. (2005), which would imply an evolution towards stronger absorption with red- Most investigations have been focused so far shift, or the missing emission has to be due to low on the X-rays below 20 keV, and INTEGRAL luminosity AGNs, clusters of galaxies, non-active can add substantial information to the nature of galaxies or ultra-luminous X-ray sources. bright AGNs in the local Universe. Considering the expected composition of the hard X-ray back- Over the life time of the INTEGRAL mission ground, it does not currently appear that the pop- we expect to detect of the order of 200 AGNs. ulation detected by INTEGRAL can explain the Combining these data with the studies based on peak at 30 keV, as Compton thick AGNs are ap- SwZfi/BAT, operating in a similar energy band as parently less abundant than expected. But the IBIS/ISGRI, will further constrain the hard X-ray sample presented here might be still too small to luminosity function of AGNs. But we will still be constrain tlhe fraction of obscured sources, and limited to relatively the high flux, low luminosity the missing Compton thick AGN could be de- and low redshift end of the distribution, which will tected when studying sources with f(20-40keV) < be inadequate to explain the cosmic X-ray back- IO-” ergs cmP2 s-’. ground at E > 20keV. Future missions such as EXIST or NuStar will be required to answer the 6. Conclusions question of what dominates the Universe in the hard X-rays. A statistically complete extragalactic sample derived from the INTEGRAL public data archive VB would like to thank Olaf Wucknitz for pro- comprises 58 low redshift Seyfert galaxies ((2) = viding software to handle the A0 > 0 cosmology. 0.022 f 0.003) and 8 blazars in the hard X-ray This research has made use of the NASA/IPAC domain. Two galaxy clusters are also detected, Extragalactic Database (NED) which is operated but no star-burst galaxy has been as yet. This by the Jet Propulsion Laboratory, of data ob- INTEGRAL AGN sample is thus the largest one tained from the High Energy Astrophysics Science presented so far. Archive Research Center (HEASARC), provided The number flux distribution shows a slope by NASA‘s Goddard Space Flight Center, and of of (Y = 1.66 f- 0.11. Because of the high flux the SIMBAD Astronomical Database which is op- limit of our sample the objects account in to- erated by the Centre de Donn6es astronomiques tal for less than 1% of the 20 - 40keV cosmic de Strasbourg. This research has made use of X-ray background. The emissivity of all AGNs the Tartarus (Version 3.1) database, created by io4’ per unit volume ~20--40k~V(> ergss-l) = Paul O’Neill and Kirpal Nandra at Imperial Col- 2.8 x lo3*e rgs s-l h$o M ~ c -ap~p ears to be con- lege London, and Jane Turner at NASA/GSFC. sistent with the background measurements in the Tartarus is supported by funding from PPARC, 2-10 keV energy band. and NPLSA grants N4G5-7385 and NAG5-7067. The luminosity function in the 20 - 40 keV en- ergy mnge is consistent with that measured in the REFERENCES 2 - 20 keV band. Below the turnover luminosity Avni, Y. & Bahcall, J. N. 1980, ApJ, 235, 694 of L, = 4.6 x ergss-’ the absorbed AGNs be- come dominant over the unabsorbed ones. Only Barger, A. J., Cowie, L. L., Mushotzky, R. F., a small fraction of Compton thick -4GNs are de- Yang, Y., Wang, W.-H., Steffen, A. T., & Capa, tectable within the sample with known intrinsic P. 2005, ACT, 129, 578 absorption. For the sources without reliable ab- sorption information we derived an estimate from Bassani, L., et al. 2006, ApJ, 636, L65 the comparison with ROSAT All-Sky Survey data Beckmann, V., Gehrels, N., Favre, P., Walter, R., and find that the data do not require additional Courvoisier, T. J.-L., Petrucci, P.-O., Malzac, Compton thick objects within the sample pre- J. 2004, ApJ, 614, 641 sented here. The sources to explain the 30 keV peak in the cosmic X-ray background have to be Beckmann, V., et al. 2005, ApJ, 634, 939 either at higher redshift, as proposed for example 8 I Beckmann, V., Gehrels, N., Shrader, C. R., & Levenson, N. A., Weaver, K. A., & Heckman, T. Soldi, S. 2006, ApJ, 638, 642 M. 2001, ApJS, 133, 269 Beckmann, V., Soldi, S., Shrader, C. R., & Lund, N., et al. 2003, A&A, 411, L231 Gehrels, N. 2006, proc. of "The X-ray Universe Lutz, D., Maiolino, R., Spoon, H. W. W., Moor- 2005", San Lorenzo de El Escorial (Madrid, wood, A. F. M. 2004, A&A, 418,465 Spain), 26-30 September 2005, ESA-SP 604, astro-ph/0510833 Maccacaro, T., Della Ceca, R., Gioia, 1. M., Mor- ris, S. L., Stocke, J. T., & Wolter, A. 1991, 374, Bird, A., et al. 2006, ApJ, 636, 765 117 Brandt, W. N., et al. 2001, AJ, 122, 2810 Marshall, F. E., et al. 1980, ApJ, 235, 4 Brandt, W. N., & Hasinger, G. 2005, ARA&A, 43 Mas-Hesse, J. M., et al. 2003, A&A, 411, L261 Caroli, E., Stephen, J. B., Di Cocco, G., Natalucci, Masetti, N., Palazzi, E., Bassani, L., Malizia, A., L., Spizzichino, A. 1987, Space Sci. Rev., 45, Stephen, J. B. 2004, A&A, 426, L41 349 Masetti, N., 2006, A&A accepted, astro- Colbert, E. J. M., & Ptak, A. F. 2002, ApJSS, ph/0512399 143, 25 Matsumoto, C., Nava, A., Maddox, L. A., Leighly, Comastri, A., Fiore, F., Vignali, C., Matt, G., Per- K. M., Grupe, D., Awaki, H., Ueno, S. 2004, ola, G. C., & La Franca, F. 2001, MNRAS, 327, ApJ, 617, 930 78 1 Matt, G., et al. 1997, A&A, 325, L13 Courvoisier, T.J.-L., et al. 2003a, A&A, 411, L53 McHardy, I. A4. M., Gunn, K. F., Uttley, P., & Courvoisier, T.J.-L., et al. 2003b, A&A, 411, L343 Goad, M. R. 2005, MNRAS, 359, 1469 De Rosa, A., et al. 2005, A&A, 438, 121 Miyaji, T., Hasinger, G., & Schmidt, M. 2000, A&A, 353, 25 Donato, D., Sambruna, R. M., Gliozzi, M. 2005, .4&A, 433, 1163 Miyaji, T., Lahav, O., Jahoda, K., & Boldt, E. 1994, ApJ, 434, 424 Giacconi, R., Gursky, H., Paolini, R., & Rossi, B. 1962, Phys. Rev. Lett.,9,439 Murdoch, H.S., Crawford, D.F., & Jauncey, D.L. 1973, ApJ, 183, 1 Gilli, R., Risaliti, G., & Salvati, M. 1999, A&A, 347, 424 Pian, E., et al. 2005, A&A, 429, 427 Goldwurm, A., et al. 2003, A&A, 411, L223 Pian, E., et al. 2006, A&.4 letters accepted, astro- phl0602268 Gruber, D. E., Matteson, J. L., Peterson, L. E., & Jung, G. V. 1999, ApJ, 320, 124 Piccinotti, G., Mushotzky, R. F., Boldt, E. A., Holt, S. S., Marshall, F. E., Serlemitsos, P. J., Hasinger, G. 2004, Nucl. Phys. B (Proc. Suppl.), & Shafer, R. A. 1982, ApJ, 253, 485 132, 86 Polletta, M., et al. 2006, ApJ accepted, astro- Krivonos, R., Vikhlinin, A., Churazov, E., Lutovi- phl0602228 nov, A., Molkov, S., & Sunyaev, R. 2005, ApJ, 625, 89 Revnivtsev, M. G.: et al. 2004, AstL, 30, 328 La Franca, F., et al. 2005, ApJ, 635, 864 Revnivtsev, M. G., Sazonov, S. Yu., Molkov, S. V., Lutovinov, A. A., Churazov, E. M., & Sunyaev, Lebrun, F., et al. 2003, A&A, 411, L141 R. A. 2005, Astronomy Letters accepted, astro- ph/0508155 9 Sazonov, S. Y., Revnivtsev, M. G., Lutovinov, A. A., Sunyaev, R. A., & Grebenev, S. A. 2004, A&A, 421, L21 Sazonov, S. Y. & Revnivtsev, M. G. 2004, A&A, 423, 469 Sazonov, S. Y., Churazov, E., Revnivtsev, M. G., Vikhlinin, A., Sunyaev, R., 2005, ,4&A, 444, L37 Schmidt, M., 1968, ApJ, 151, 393 Fig. 1.- Exposure map representing the data Soldi, S., et al. 2005, A&.4, 444, 431 used in our analysis in Galactic coordinates. Con- tours indicate 2 ks, 5 ks, 10 ks, 20 ks, 100 ks, Treister, E., & Urry, C. M. 2005, ApJ, 630, 115 200 ks, 500 ks, 1 Ms, 2 Ms exposure time. INTE- GRAL spent most of the observing time on and Ueda, Y., Akiyama, M., Ohta, K., & Miyaji, T. near the Galactic plane, with a strong focus on the 2003, ApJ, 598, 886 Galactic center and on areas including bright hard Ueda, Y., Akiyama, M., Ohta, K., & Miyaji, T. X-ray sources like the Cygnus region, Vela, GRS . 2004, PThPS, 155, 209 1915+105, and the Crab. Fields at high Galac- tic latitude include Coma, Virgo and the region Ueda, Y., Ishisaki, Y., Takahashi, T., Makishima, around NGC 4151. K., & Ohashi, T. 2001, .i\pJS, 133, 1 Vaughan, S., Iwasawa, K., Fabian, A. C., & Hayashida, K. 2004, MNRAS, 356, 524 Vedrenne, G., et al. 2003, A&A, 411, L63 -10 Virani, S. N., et al. 2005, AAS, 207, 2007 Voges, W., et al. 1999, A&AS, 349, 389 Voges, W., et al. 2000, IALC, 7432, 3 Winkler, C., et al. 2003, .4&.4, 411, L1 VI Worsley, M. A., et al. 2005; MNRAS, 357, 1281 F Y Young, A. J., Wilson, -4. S., Terashima, Y., Ar- - naud, K. A., Smith, D. A. 2002, ApJ, 564, 176 I -11 M Zdziarski, A. A. 1996, .MNRAS, 281, L9 -11.5 I I I I , I I I ,AI ,I I I , i Fig. 2.- Correlation of exposure time and flux limit (5~72,0 - 40 keV) for the AGNs in this study. The curve shows a smooth polynomial fit for flux limit versus logarithm of the exposure time. ~~ This 2-column preprint was prepared with the AAS MT&i macros v5.2. 10