X-ray studies of Active Galactic Nuclei

Transcription

1 X-ray studies of Active Galactic Nuclei Poshak Gandhi European Southern Observatory, Alonso de Cordova 3107, Casilla 19001, Santiago, Chile 1. Introduction Unlike modern optical astronomy that began with the first use of telescopes about 400 years ago, the study of the Universe in X-rays is only about four decades old. The pathlength of typical cosmic X-rays in the Earth s atmosphere is only a few cm this opaqueness being due to high photoelectric absorption cross-section of X-ray photons in air. The very small wavelength of X- rays also necessitates new ways of focusing them in order to form images. These limitations were overcome only with the advancement of technology. X-rays from the Sun were first discovered during short rocket flights in the 1940s; but the flight that marked the birth of cosmic X-ray astronomy was carried out on June 12, Aiming to search for X-rays from the Moon, this mission ended up discovering Scorpius X-1 (the brightest point-source in the X-ray sky later recognized to be a neutron star in a binary system), as well as the X-ray background, a faint isotropic glow of radiation whose origin remained a mystery for almost 40 years [1]. Tremendous progress has been made since then, and the latest all sky surveys can now boast more than 100,000 detected X-ray sources. The X-ray spectral regime covers energies of several hundred kev down to 0.1 kev (equivalent to wavelengths of less than 0.1 Å to above 100 Å). These energies are associated with thermal temperatures above 10 7 K, higher than those of stellar photospheres by factors of several thousands or more (see Fig. 1). Non-thermal processes that generate X-ray photons are: bremsstrahlung; synchrotron; discrete quantum transitions, e.g., fluorescence, leading to line radiation; and Comptonization. Clusters of galaxies are the largest bound structures in the observable Universe. The presence of ionized gas within large, deep potential wells provides the right conditions (density, temperature) for the bremsstrahlung cross-section to be significant. Synchrotron processes are associated with intense magnetic fields in compact objects. In collimated astrophysical jets, synchrotron radio emission can be inverse-compton scattered up to X-ray energies by the very electrons that produce the radio photons, in a process known as Synchrotron Self Compton. The most important process from the perspective of the sources discussed in this paper, however, is Comptonization. One of the most important astronomical discoveries of the past decade is the fact that every large galaxy, including the Milky Way, harbours a nuclear super-massive black hole (SMBH; [2]), with mass ranging from solar masses (M ). About 10 per cent of these appear to be extremely bright active galactic nuclei (AGN) at any given epoch. Typical AGN X-ray luminosities range from erg s 1. This should be compared with the bolometric luminosity of the Sun (L ) erg s 1. Thus, AGN can easily generate between 10 9 and L of power in X-rays alone, enough to rival an entire galaxy. AGN were first discovered in the 1940s as point-like sources of powerful optical emission with spectra showing very broad and strong emission lines having associated doppler velocities of order several thousand km s 1. They were also found to show significant optical variability on time-scales of months, with the emitting source being completely unresolved. Strong X-ray variability is observed over periods of hours to days, implying that the emission region must be very compact, indeed (since net, observed variability is governed by the source light-crossing time). The inferred power and compactness of AGN is most easily explained by accretion of hot gas onto the SMBH and liberation of the gravitational energy associated with this infalling matter through dissipative processes; this 1

2 link between highly variable sources with strong optical emission and accreting SMBHs was first established in the late 1960s [3]. AGN have since been studied at every available wavelength from radio to Gamma rays, and are found to be prodigious sources of emission across the whole electromagnetic spectrum. X-rays, however, are the most direct probe of the accretion processes responsible for this massive energy generation because they can penetrate through large columns of obscuring gas and dust associated with the accreting matter that surround many AGN. Moreover, X-rays are generated in the very hearts of these sources, down to distances within a few gravitational radii of the central black hole (BH), while emission at other wavelengths, e.g., optical radiation, is generated at distances 10 5 times or more farther out. AGN are the dominant sources of cosmic X-rays. In any medium-deep image of the X-ray sky at high galactic latitudes, the majority (about 80 per cent) of detections will be AGN, followed by distant galaxy clusters and X-ray binaries within the Galaxy. This knowledge is relatively new until the early 1990s at least, the X-ray sky was observed to be dominated by a diffuse cosmic X-ray background radiation that could not be resolved into individual sources with the previous generation of X-ray telescopes; this background is now recognized as the integrated emission from distant AGN populations. The role of powerful AGN feedback through winds and ionization of the interstellar media is now seen as an integral part of the process of galaxy formation; AGN can no longer be ignored as rare and compact singularities with no significant impact on their environments. Some of the most recent X-ray surveys are revealing unexpected populations of AGN in the distant Universe, and suggest that there may have been more than one major epoch of black hole mass accretion assembly in the history of the Universe. This paper presents a brief summary of the field of observational AGN X-ray studies. Due to the large scope of the subject, this work is not intended to be comprehensive, but it is an up-to-date analysis of phenomenological studies of AGN in X-rays. In section 2, we outline the knowledge of AGN structure and characteristics that has been built up from multi-wavelength observations over the past few decades. This also includes the standard AGN Unification picture and describes X-ray spectra of AGN in some detail. X-ray detection technology and important satellite missions that studied AGN are mentioned in section 3. Finally, the importance of AGN in the process of galaxy formation and evolution is outlined and key results from the latest generation of X-ray satellites, Chandra and XMM-Newton, are discussed. 2. AGN structure and fiducial characteristics Fig. 2 shows the main physical components believed to exist in most AGN. Most physical sizes are scaled with the radius of the event horizon, which is linearly dependent on mass as R S = 2GM/c 2 for a non-spinning BH. For a Solar mass BH, this radius is approximately 3 km. However, stable structures cannot always extend down to the event horizon. The inner-most stable orbit for matter in the space-time of a non-spinning (Schwarzschild) BH is 3R S, but for a maximally-spinning, Kerr BH, both the event horizon and last stable orbit are found at R S /2. Increasingly, evidence is being found that many black holes may be rapidly spinning [5, 6]: these objects are excellent probes of regions closest to the BHs themselves. Studies of X-ray emitting stars within the Galaxy, primarily X-ray binaries, suggest that accreting matter, as it falls inwards, settles in a planar orbit around the central source (an accretion disk ) where most of the gravitational energy release occurs due to dissipation. The overall efficiency of energy release in this process is higher than most other processes known 1, including nuclear fusion, 1 A fiducial value for the efficiency of conversion of gravitational accretion energy into radiation for bright AGN is 10 per cent. The equivalent efficiency of mass:energy conversion for nuclear fusion of Hydrogen is only 0.7 per cent. 2

3 Figure 1: View of the sky region covering the Orion constellation compiled from a series of images taken by the ROSAT X-ray observatory (left), compared to the same region as it would be seen in optical light (right). The sky appears quite different in the two bands. The brightest source in the optical image is the Moon, which is conspicuously faint in X-rays. Additionally, the brightest star in visible light is Sirius A, at the left edge of the image. In X-rays, the star seen at this position is actually Sirius B, a known white dwarf companion to Sirius A, that is completely invisible in the optical regime ( years/survey/, [4]). and is enough to explain the incredibly high power-outputs of X-ray binaries and SMBHs. Accretion disk sizes are thought to be of the order of light-days for typical SMBHs of mass 10 6 M. However, even at the distance of the nearest AGN, such sizes are too small to be resolved by the current generation of telescopes, since the resolution required is close to 1 milli-arcsecond, or about 10 7 degrees (though there are some suggestive claims in this regard; [7]). Improved interferometric techniques on the largest ground-based telescopes may soon be able to resolve the disks for the very nearest AGN. There exists one, fortuitous case, NGC 4258, where it has been possible to map motions in the accretion disk due to presence of a strong water maser [8]. Note that any large, stable accretion disk is unlikely to exist in our Galaxy, since the central black hole is only weakly, and sporadically, active [9]. Typical AGN accretion disks are optically-thick and physically-thin [10]. Associated temperatures are in the range of K, making them sources of quasi-thermal optical and ultraviolet radiation that scatters off electrons commonly found in the ambient media of hot and ionized accretion regions; this corona of electrons is heated by energy feeding from magnetic fields (similar to the mechanism of heating of the Solar corona). Cool accretion disk photons thus undergo inverse- Compton scattering off hot electrons, and emerge as higher energy X-rays. Multiple scatterings within the corona increase the energy further, resulting in characteristic power-law non-thermal spectra extending from under 1 kev to several hundred kev. More details of the main features in AGN X-ray spectra are discussed in section

4 Discrete and patchy cloud-like structures much further out from the SMBH produce the bulk of AGN optical emission line radiation, according to which AGN were first classified. Essentially, those with very broad (full width at half maximum FWHM of order 3000 km s 1 or more) permitted recombination lines also typically have blue broad-band optical continua and are designated type 1 AGN. Those with comparatively narrow (FWHM 500 km s 1 ) permitted lines are designated type 2 AGN and usually possess red continua. Intermediate types also exist which display both narrow and broad components to the permitted lines, and all types can show significant forbidden lines, but these are always narrow. Photoionization is now accepted as the principal mechanism for emission line generation, though collisional excitation can have a significant effect, especially for forbidden lines. The fact that forbidden lines are always observed to be narrow suggests the existence of two distinct regions where optical emission principally occurs. A dense ( cm 3 ) broad line region (BLR) is inferred to exist on scales much smaller than a parsec (pc) within the deep BH gravitational potential, leading to the observed large doppler-broadened line-widths. A technique called reverberation mapping that uses the delay time between continuum variation and subsequent emission line intensity variations suggests the size of this region to be 0.1L pc (where L 46 is the photoionizing luminosity in units of erg s 1 ; [11]). The narrow line region (NLR), on the other hand, extends over scales of tens to hundreds of pc, where the emitted lines are all narrow due to weaker gravity. The very existence of strong forbidden lines implies that densities in the NLR cannot exceed the critical density for collisional de-excitation, and are thus inferred to be low ( 10 4 cm 3 ) from analysis of various nebular line species such as the [Oiii]λλ4959, 5007 doublet. Temperatures of the photoionized media are difficult to ascertain due to dependence of radiative cooling on metallicity, but values of 10 4 K are typical for the NLR; the BLR being hotter by a factor of [12]. In addition, AGN have been classified according to their luminosities at radio wavelengths (compared to the optical luminosity). Radio-loud AGN were among the first to be catalogued in large numbers and can have distinctive jet and/or lobe morphologies [13]. Radio-quiet AGN are difficult to identify, especially at high redshift, but probably outnumber their radio-loud counterparts by a factor of 5 or more. Other classes of AGN include BL Lacertae objects, blazars and LINERs. For comprehensive discussion of the different AGN classes, consult the following references: [14-16]. 2.1 The obscuring torus and AGN Unification The variety of seemingly distinct classifications can be united to a large extent under what has now come to be known as the AGN Unification scheme. According to this hypothesis, the common powerhouse in each type of AGN is an accreting SMBH, while most of the differences amongst the classes can be ascribed to anisotropic AGN obscuring environments and their random orientations towards Earth (Fig 2). The idea of unification first gained credence through observations of the type 2 Seyfert 2 NGC 1068 [17], in which a broad-line type 1 AGN was uncovered through polarized light. If optically-thick obscuring matter (mostly dust) blocks the line-of-sight to the nucleus of NGC 1068, the broad permitted lines (generated close to the nucleus) are not seen directly. But if other sight-lines are not completely obscured, a small fraction of radiation could be scattered into our direction. The absorbing matter is thus not sky-covering (as seen by the AGN) and is generically imagined to be in a torus geometry. The polarization of the radiation suggests that the reflector consists of electrons in a plasma, presumably ionized by direct AGN radiation. The narrow-line 2 AGN which are fainter than M B log(h) [where M B is the absolute Vega B-magnitude and h is the Hubble constant in units of 100 km s 1 Mpc 1 ] in the optical are traditionally referred to as Seyferts; QSOs and quasars are sources with higher luminosities. In X-rays, the Seyfert / QSO classification threshold luminosity is at erg s 1. 4

5 region in all AGN lies above the torus and is irradiated by the photoionizing continuum. In the local Universe, obscured AGN are seen to be more common than unobscured AGN by a ratio of at least 3 4 [18]. Whether this ratio evolves out to higher redshifts is an important question currently under study. Thus, the size and orientation angle of the torus is the key variable that decides the appearance of an AGN. Broad permitted lines in type 2 AGN have also been discovered in the near-infrared [19], where the opacity is much lower than at optical wavelengths, while re-processed, warm, thermal radiation from the torus itself has been detected in the far-infrared [20]. It is the X-rays, however, that have the highest penetrating power through obscuring gas that is associated with the enshrouding torus dust. Figure 2: Schematic of an AGN. The central Black Hole accretes infalling matter from the accretion disk whose characteristic temperature of the order of several hundreds of thousands of Kelvin results in thermal ultraviolet emission. This radiation is scattered off a hot electron corona above the accretion disk, gaining energy in the process and emerging as X-rays. This radiation ionizes gas that exists in the form of patchy clouds at distances of several light-days to light-years from the central source, resulting in characteristic optical broad and narrow emission lines. Extended powerful jets and ionization cones are also observed in many AGN; note that they are expected to be symmetric about the SMBH, though in the diagram they are shown on one side only, for clarity. The dusty torus absorbs radiation along certain lines of sight. If the orientation of the AGN is such that light reaches Earth directly (without passing through the torus), we would see an unobscured AGN. Otherwise, an obscured AGN would be observed. If the orientation angle (or, optical depth) is large, this obscuration can absorb most of the direct AGN light, resulting in sources that are optically-thick even to photons upto 10 kev; these sources are called Compton-thick, and are dominated by indirect radiation scattered into the line-of-sight. Diagram is not to scale. 2.2 AGN X-ray spectra There are many prominent features seen in the X-ray spectra of AGN. Several of these are discussed below and shown in Figs. 3 & 4. A few important, but relatively infrequent spectral features, including warm absorbers and soft blackbody excesses, are not discussed here. 5

6 AGN for which there is a direct line-of-sight to the central black hole are characterized by power-law continua over the kev regime. Due to the way in which X-ray detectors work, observed spectra are usually parametrized as power-laws in terms of photon density: N E E Γ cts s 1 cm 2 kev 1. In the local Universe, an average value of the photon-index Γ = 1.9 is observed [21] for AGN, with typical variation over Γ = The power-law nonthermal spectrum arises due to multiple up-scatterings of accretion disk photons undergoing inverse-compton scattering off Maxwellian-distributed hot electrons above the accretion disk. A broad hump, or excess of power, is seen in many AGN over kev above the canonical lower-energy power-law. This can be explained as due to reflection of photons that are scattered from the corona back towards the accretion disk [22]. The reflection fraction depends on the ionization state of the disk, but the overall effect is predicted to be a hardening or biasing of the spectrum towards higher energies, due to high albedo at 30 kev combined with photoelectric absorption of lower energy X-ray photons. Above several tens to hundreds of kev, X-ray and gamma-ray telescopes have discovered an exponential cut-off to the AGN power-law extrapolated from lower energies; the characteristic cut-off energy can be as high as a few hundred kev. At these high energies (approaching the electron rest-mass energy of 512 kev), photons lose more energy than they gain in each Compton scattering because of electron recoil, resulting in a roll-over of the spectrum. The generic optical segregation of AGN into type 1 and type 2 classes has an X-ray analogue as well. The lower energy end of the spectrum in AGN (below 10 kev) is modified by obscuring gas through photoelectric absorption if the line-of-sight passes through the obscuring torus. A typical integrated line-of-sight obscuring column density that delineates the boundary between obscured and unobscured AGN is cm 2. The absorbing column-density itself is parametrized in terms of Hydrogen atoms per cm 2 integrated along the line-of-sight, though heavy elements such as O, Mg, Si, S and Fe are responsible for much of the opacity above 1 kev, and discreet photoelectric edges are observed at the ionization energies for these ions [23]. The obscuration associated with an e-folding decrease of transmitted radiation at 10 kev in the source rest-frame is cm 2, and these sources are called Compton-thick. Very little direct radiation is seen from these sources. Additionally, at very high column-densities, multiple scatterings, as modelled by the Klein- Nishina cross-section, lead to a Compton down-scattering over the full X-ray spectral regime, and very little direct flux emerges above column-densities larger than cm 2 [24]. Discreet quantum transitions in the X-ray regime produce emission lines, primarily due to fluorescence. These had been beyond the reach of study by X-ray detectors until a few years ago, due to the relatively-high resolving power required to distinguish them. Iron transitions are the most common lines found in AGN X-ray spectra, due to a combination of high abundance and fluorescence yield, though the recent use of gratings capable of dispersing X-rays into their component energies has revealed an abundance of other elements [25]. Emission lines are very sensitive probes of the inner regions of the accretion disks, due to several relativistic effects that affect their observed profile. The rotation of the accretion disk leads to the doppler-splitting of the line on the approaching and receding sides of the disk relative to the observer; line photons lose energy due to strong gravitational redshift as the photons climb out of the deep potential well associated with the central BH; further distortion occurs due to relativistic boosting of the blue (approaching) part of the spectrum relative to the red part and due to the transverse doppler effect associated with time dilation. The net, 6

7 predicted line profile has a blue peak and a hump with an extended red tail (see Fig. 4), each feature being sensitive to BH characteristics such as mass, spin etc. Such a line profile has been observed in several local [26] as well as very distant [27, 28] AGN, thus also providing a laboratory to test these relativistic effects in action. Figure 3: AGN rest-frame X-ray spectral energy distributions (SEDs) showing key spectral features discussed in section 2. The spectra were simulated using the X-ray spectral analysis code xspec [29]. The y-axis units are such that a horizontal line denotes equal power per unit decade of energy. The intrinsic AGN power-law is modelled with photon-index Γ = 1.9 (the equivalent exponential index in the units of the above plot is Γ + 2). The spectra include a reflection hump from cold material at solar abundance assuming uniform 2π reflection and an inclination angle of 60 degrees, as well as an exponential cut-off at 360 kev. Obscuration by material of solar abundance local to the source is modelled [with log(n H ) = to 25.25, labeled] using the cross-sections of Morrison & McCammon [23], and Compton down-scattering is modelled with the full Klein-Nishina cross-section. Finally, 2 per cent of the primary radiation is added to each SED to model the scattered component into the line-of-sight. See Gandhi & Fabian [30] and Wilman & Fabian [24] for details. The shaded region marks the X-ray spectral regime that is accessible to study to current high spatial resolution X-ray instruments such as Chandra and XMM-Newton. Thus, characteristic AGN spectral features are generated on a wide spatial scale in the nuclear surroundings, with the X-ray view penetrating closest in to the black hole itself. AGN energy output is considerable across the entire spectrum, and is the result of various physical processes occurring on differing scales. The Unification scheme has provided a feasible frame-work for progress in our understanding of AGN structure and environment. Recent observations suggest, however, that the true nature of AGN may be different from that implied by this Unification scheme. Before discussing these, we present a brief review of AGN X-ray detection technology and surveys carried out so far. 7

8 Figure 4: (Left) AGN Fe line profile distortions due to dynamical and relativistic effects discussed in the text and labelled on the plot. The overall line profile is shown in the bottom plot, as a function of the ratio of the observed frequency to the frequency at which the line photons were emitted. (Right) Observed 6.4 kev Fe line in the Seyfert galaxy MCG , showing the best such data available to date. The data are compiled from 300 kiloseconds of XMM-Newton exposure time, and provide spectacular confirmation of the aforementioned predicted effects. Incidentally, the extended red wing of the line suggests that line photons undergo large gravitational redshifts that naturally arise from the deep potential well of a rapidly spinning black hole Credit: A.C. Fabian [26]. 3. X-ray Observatories and AGN Surveys After brief rocket flights that carried proportional counters into space in the 1960s, the first, dedicated X-ray satellite to carry out a uniform all-sky X-ray survey was Uhuru, launched in 1970 [31]. It discovered a total of 339 sources to a limiting sensitivity of 10 3 times the flux of the Crab Nebula, most of these being X-ray binary stars and supernovae remnants, but also some Seyfert Galaxies and galaxy clusters. Uhuru is also credited with discovering the diffuse X-ray emission associated with clusters of galaxies. Towards the end of the 1970s, NASA s High Energy Astronomy Observatory (HEAO) series of missions were a milestone for X-ray astronomy. With scintillation detectors in addition to proportional counters, HEAO provided complete surveys of the X-ray sky, especially high galactic latitude regions [32], and good characterization of the diffuse X-ray background radiation over the energy range 3 50 kev [33]. Incidentally, Aryabhata, India s first indigenous satellite, launched in 1975, 8

9 also carried an X-ray detector. None of these missions had the capability of focusing X-rays to form sharp images. Scintillation detectors and proportional counters can only measure the energy of an incoming photon, and there is no spatial resolution possible within their field-of-view. The Einstein satellite was the first X-ray telescope with focusing optics necessary to produce images (see section 3.1 below for a thorough discussion of X-ray focusing optics). It was the second satellite in the HEAO series, operational during the 1980s, and imaged thousands of extragalactic X-ray sources, as well as diffuse and extended sources over kev with superb spatial resolution of only a few arcseconds. It also carried a proportional counter sensitive up to 20 kev. Other key missions of the 1980s included the European Space Agency s EXOSAT and the Japanese Ginga satellites. The ROSAT, or Röentgen, satellite was the major German/US/UK collaborative X-ray observatory over the soft (< 2 kev) energy range. It was used to carry out several surveys of varying depth, the largest of which was an all-sky survey containing more than 100,000 detections in all (See Voges et al. [34] and references therein). With its large collecting area and high resolution optics, ROSAT was able to completely resolve the soft X-ray background into discrete sources (mostly AGN) through ultra-deep observations [35]; the origin of this background had remained a mystery for almost 40 years. ASCA [36] and BeppoSAX were important Japanese- and Italian-led missions of the nineties respectively, that were responsible for detailed studies of AGN continua as well as Fe emission lines. BeppoSAX was sensitive over a very large energy range from kev. With a point-spreadfunction of approximately 1 arcminute, it was also able to rapidly localize Gamma Ray Bursts over this spatial scale [37]. As mentioned, X-ray sources are expected to be highly variable due to their compactness. This flux variability has been studied in detail by the dedicated satellite Rossi X-ray Timing Explorer (RXTE), which continues to operate today. RXTE is sensitive over the range kev, and has made several important discoveries, including variable kilohertz Quasi-Periodic Objects, and confirmation of the fact that AGN Fe fluorescence line flux variations need not correlate with continuum variability [38]. 3.1 X-ray imaging technology X-ray detectors, for many years, were relatively simple photon counting devices without much spatial resolution or ability to focus X-rays. Proportional counters are like photon buckets that produce an ionization charge dependent on the energy of the incoming photons. Some spatial resolution may be obtained by the use of coded aperture masks, which are sheets of opaque material placed in the optical path with a well-defined pattern of pinholes to let photons through. Source image construction is achieved by deconvolving the pattern of the pinholes from the observed pattern projected by incoming photons. Though this allows only limited spatial resolution, the total collecting area for even high-energy (hard) X-ray photons can cheaply be made very large. X-ray focusing instruments had to await advances in technology that allowed polishing of surfaces to extremely high precision. A typical X-ray photon has an energy higher than an optical photon by a factor of a 1000 or more. Normal incidence mirrors that are used to focus visible light completely absorb X-rays falling directly on them, making it impossible to use well-known telescope techniques in the optical regime. However, just as a solid piece of rock can bounce off the surface of water if deflected at a grazing angle, X-rays can be focused using grazing incidence reflection off surfaces that are almost parallel to their incoming direction (Fig. 5). There are two basic requirements needed to put such a system into practice. The first is that the angle of reflection be very small with respect to the surface. It can be shown that this angle 9

10 varies as ρ/e, where ρ is the density of the reflecting surface, and E is the energy of the X-ray photon [39]. Thus, the higher the energy of the incoming ray, the smaller has to be the angle of reflection. For photons of a few kev, this angle turns out to be smaller than 1. The second requirement is that the reflecting surface be very highly polished. Given the minute wavelength of X-ray photons, and the small angles under consideration, the reflecting surface of the main mirrors of the Chandra X-ray observatory, for example, are polished to an average surface roughness of only a few angstroms. X-ray mirrors are thus highly polished surfaces placed in the path of, and parallel to, the incoming beam. It turns out that reflection off a paraboloid section followed by a confocal and coaxial hyberboloid surface provides good focus on-axis as well as at small off-axis angles (current largest off-axis angles at which reasonable focus is achieved are about a few arcmin). Mirror reflecting area can be increased by nesting surfaces of different radii within each other. Such a configuration was first suggested by Wolter [40], and is the current standard of X-ray imaging instruments, such as Chandra and XMM-Newton, described in the next section. X-ray detector technology has also progressed steadily. Current solid-state arrays capture individual counts in various channels, depending on the incoming photon energy. These channels are like photon-counting devices, and the mapping of channels to the associated photon energy is a non-trivial (and non-unique) problem. This does, however, allow the simultaneous measurement of a photon s spatial position and energy, without the use of dispersing elements. This is only possible if the flux of X-rays from a source is small enough that only a single photon hits any given detector element between two consecutive read-outs, a condition easily satisfied for most studies of distant AGN with the current generation of observatories. Figure 5: The principal components of a Wolter type 1 nested X-ray focusing mirror configuration. The double reflection off two surfaces with different curvatures (paraboloid and hyperboloid) is done in order to ensure good focus over extended regions in the focal plane, while the nesting of mirrors is done in order to increase the overall effective area. Alignment, polishing and coating of the mirror surfaces are the most critical parts of the assembly. Image courtesy of Chandra and XMM-Newton 10

11 The Chandra mission is one of the key missions of NASA s Great Observatories Origins Program. It has four polished and iridium-coated mirror surfaces with roughness close to only a few angstroms. It is named after the Nobel-prize winning astronomer, S. Chandrasekhar, who is credited with the first detailed calculations of black hole critical mass limits. The result is excellent spatial resolution of approximately half an arcsecond the best resolution of any X-ray instrument yet built or conceived for up to the next years. Whereas most previous X-ray surveys required detailed multiwavelength follow-up of sources simply to localize and identify them, the sub-arcsecond resolution is good enough for direct counterpart-matching and source identification in a majority of point-source observations. The combination of high sensitivity over the kev range and excellent optics implies that very faint sources can be detected with Chandra. One of the main results from the mission has been the resolution of the hard cosmic X-ray background radiation (CXB) up to approximately 8 kev [41]. It is now recognized that the bulk of the CXB is the summed emission from a distribution of redshifted, and predominantly-obscured AGN. The exact nature and distribution of these distant, faint AGN are now under intense scrutiny [42]. Two sets of X-ray transmission gratings that are present in the optical path of Chandra have been used to observe large numbers of emission lines from X-ray sources, including AGN. For instance, the spectrum obtained for the AGN at the centre of NGC 3783 shows hundreds of absorption features due to various ionic species of Fe, Mg, O, N, Ne, Si etc. [25]. XMM-Newton is ESA s major currently-operational X-ray observatory; it complements Chandra by having a much larger collecting area (a factor of 10 larger at 2 kev) over kev, but a worse spatial resolution by a factor of a few. The mirror surfaces are coated with Gold. Due to its large collecting area and good spectral resolution, XMM-Newton is able to search for spectral features in distant cosmic sources fainter than those detectable by any previous mission: e.g., Fe emission lines in distant AGN [43] and emission from the cores of distant galaxy clusters (cooling flows; [44]). 3.3 Future X-ray missions Astrosat 3 is the latest of India s indigenous space observatory projects, due to be launched in 2007 into a near-earth, equatorial orbit by the Polar Satellite Launch Vehicle [45]. It is designed as a multi-wavelength observatory for spectral, imaging and timing studies of cosmic sources, including AGN. Four payload instruments are planned and will cover the UV( A), soft and hard X-ray regimes (0.3-8 kev; 2-80 kev). On a longer (10 20 year) timescale, ESA s planned XEUS mission will have an effective area larger than that of Chandra by a factor of about a thousand, while NASA s Constellation-X mission will be a collection of small synchronous satellites with mirrors, acting as interferometers to achieve the combination of very high collection area and good resolution. It must be realized that the peak effective area of the Chandra imaging instrument ACIS is equivalent to that of a telescope with a diameter of only 30 cm (assuming normal incidence optics). With the next generation of missions, X-ray astronomy will truly enter the realm of large telescopes such as those currently available at optical or near-infrared wavelengths. An overview of most past, present as well as planned X-ray observatories can be found on web-sites related to the Goddard Space Flight Centre e.g.: 11

12 4. Universal accretion in a broad context 4.1 The cosmic X-ray background radiation The cosmic X-ray background (CXB) was discovered two years before the better-known microwave background [1]. It covers a very wide energy range, extending from 0.1 kev to several hundred kev (Fig. 6). Above 3 kev, the CXB is isotropic to within a few per cent on large scales, after account is taken of a weak Galactic component as well as the dipole radiation field due to the motion of our Galaxy. This is strong evidence that the majority of this background must be cosmological in origin. However, there was no concrete and satisfactory explanation for its existence for almost 30 years. Different processes are now known to dominate the CXB in different energy regimes and have been identified at different times over the past 40 years. A distinction is usually made between the soft CXB below 3 kev and the background above this energy. At the softest energies ( < 0.25 kev), more than 90 per cent of the CXB probably originates as thermal emission in a local bubble of hot, optically-thin ( 10 6 K) gas in the solar neighbourhood. CMB The Cosmic Energy Density Spectrum Flux density on sky [nw/m2/sr] CIB COB CXB 1.0E E E E+25 Frequency [Hz] Figure 6: Observed spectrum of Cosmic Backgrounds from microwaves (CMB), infrared (CIB), optical (COB), X-rays (CXB) out to gamma-rays and cosmic ray particle backgrounds. The units of the plotted flux density (nw m 2 steradian 1 ) are such that a horizontal line represents equal contributions of power per unit decade of frequency. As discussed in the text, the CXB arises from the integrated contribution of AGN with different luminosities and intrinsic obscuring column densities spread over redshift. The CMB is the afterglow of the big bang, redshifted from the surface of last scattering, or recombination, while the optical and infrared backgrounds are due to the summed contributions of stellar and accretion processes. Credit: G. Hasinger [46]. Studies by HEAO-1 found that the spectrum of the CXB over the range 3 50 kev can be fitted to a diffuse 40 kev thermal bremsstrahlung model [33]. Over the range 3 10 kev, a hard spectrum was found with a photon-index Γ = 1.4 [N(E) E Γ ]. The good fit to the free-free emission model 12

13 prompted the development of a number of theories in which a diffuse hot inter-galactic medium (IGM) was the principal emitter, e.g., [47]. However, the total IGM matter and energy density predicted by such models was found to over-predict the observed density significantly. The nearperfect black body spectrum of the CMB found by the COBE satellite [48] conclusively showed that there were no Compton up-scattering features of microwave photons due to such an IGM, implying that discrete sources, and not diffuse matter, must be responsible for the bulk of the background. The difficulty faced by early CXB synthesis models was that no class of sources was known to have a spectral slope that either matched the CXB slope above 3 kev, or that could be made to match the slope by a summation of objects widely spread in redshift. The known sources of powerful extragalactic X-ray emission included thermal bremsstrahlung cooling in galaxy clusters, massive X-ray binaries in starburst galaxies and active galactic nuclei (AGN). While AGN were subsequently shown to constitute the majority of the CXB flux at 1 2 kev, most objects known twenty years ago were bright, unobscured AGN with spectra that were too steep (i.e., soft) and inconsistent with the CXB photon-index of 1.4 above 3 kev. Either an entirely new class of AGN (such as AGN with radiatively-inefficient flows 5 ) had to exist, or a larger population of AGN (with intrinsic, steep spectra similar to those observed, but flattened by the effects of photoelectric absorption local to the source) had been missed from previous surveys [50]. This latter, absorbed AGN hypothesis now has been shown to be correct, but it had to await for more than a decade for observational verification. Operations by the Chandra and XMM-Newton satellites over the past five years have proven beyond all doubt that the integrated emission from a wide distribution of AGN spread out in redshift and with different obscuring column-densities can account for the bulk of the X-ray background. While uncertainty up to 30 per cent remains on the exact normalization of the X-ray background intensity due to cross-calibration difficulties between various satellite missions, at least 75 per cent of the hard X-ray background between 2 10 kev is now resolved [41], but a significant percentage still remains unidentified if one considers only the high end of this energy range [51]; this could arise from truly diffuse emission or be due to distant and unresolved obscured / Compton-thick AGN. Detailed multi-wavelength follow-up of the X-ray sources detected and resolved so-far has revealed several interesting properties, including: the discovery of large numbers of so-called type-2 QSOs, or AGN with intrinsic luminosities similar to those of local quasars, but hidden behind obscuring columns of log(n H ) = 22 or higher [27, 52]. These are natural analogues of the lower-luminosity Seyfert-2s that have been known for a long time. QSO2s are being found in large numbers only now because of the fact that X-rays are more efficient at locating them than optical surveys, and an empirical observation that the fraction of sources that are obscured may actually decrease as a function of intrinsic luminosity, making QSO2s less abundant than Seyfert-2s relative to their unobscured counterparts [53]. the discovery of large numbers of objects whose powerful X-ray emission suggests the presence of actively accreting supermassive black holes (i.e, AGN) within, yet which show few or no signs of AGN activity in visible light [54, 55]. The optical as well as near-ir spectra of these X-ray bright, optically-normal galaxies (XBONGs; also referred to as optically-weak AGN, elusive AGN etc.) possess absorption features characteristic of early-type host galaxies and a common feature is the distinct lack of strong, redshifted emission lines associated with powerful 5 Examples of radiatively-inefficient flows include Advection-Dominated Accretion Flows, or ADAFs; in these sources, almost all of the accreted energy is lost into the black hole instead of being radiated away, due to ineffective coupling amongst viscous dissipative phases in the accreting matter [49]. These are now known not to be major contributors to the CXB. 13

14 (unobscured, as well as obscured) nuclear activity. XBONGs are found in luminous, as well as faint, X-ray sources, and are thought to constitute 10 per cent of all X-ray selected AGN [56], though the true fraction is likely to be much higher, given the difficulty of classification and incompleteness of current spectroscopic follow-up [43]. This incompleteness fraction is often as high as 40 per cent, simply because the targets are too faint to study even with the largest, current 10 m class telescopes; the discovery of a relatively low-redshift peak in the distribution of sources that contribute to the CXB. Relative to optical quasars and their radio analogues that have been mapped extensively and are seen to peak in number density at z 2, the newly discovered X-ray AGN peak at z 0.7, even after account is taken of the surveys incompleteness. This fact suggests a resurgence in the phase of Universal accretion activity, possibly associated with anti-hierarchical growth in which lower-luminosity (X-ray) Seyferts were assembled after their more powerful (optical) quasar counterparts [57]. Concerning the discovery of XBONGs (second point in the above list), Moran et al. [58] have pointed out that ground-based follow-up of distant AGN with finite-width spectroscopic slits is significantly biased by the host galaxy continuum, thus making nuclear emission lines difficult to distinguish. Diluting AGN light becomes increasingly difficult with increasing nuclear luminosity, however, unless these objects are completely obscured (4π covering fraction of obscuring dust with little or no scattering into our line-of-sight) on scales of less than a kiloparsec [27]. The mechanism which generates such large sky covering factors of obscuration and prevents its dissipation into a planar disk is unknown. The power of embedded starbursts as well as nuclear radiation pressure could provide the necessary randomized motions required [59, 60]. XBONGs thus provide an important tool to study the obscuring environment of AGN; their very existence has important consequences for orientation-based Unification models, since XBONGs should, instead, appear spherically-symmetric. Other models considered for XBONGs include sources with extremely large dust:gas ratios (atypical for local Seyferts), low-luminosity AGN and BL-Lacs [56]. Models with centrally-truncated accretion disks undergoing advective accretion have also been considered [61]. The general consensus is that the more powerful sources (possibly including some obscured quasars; [27]) are likely to be completely obscured AGN. 4.2 AGN evolutionary schemes The study of ever-fainter populations of active galaxies seems to be revealing more variety in AGN activity and making the overall picture more complex. On the other hand, the fact that AGN come with disparate properties is a natural consequence of the different stages of their life-cycle at which they are caught, and can be understood as part of an evolutionary scheme. This is an active field of research at the moment, with results being drawn together from right across the electromagnetic spectrum. A significant unknown in this whole puzzle is the mechanism by which this life-cycle begins. While stellar BHs have a theoretically sound basis of formation from the collapse of an evolved, massive star, it is not known whether super-massive black holes (that can easily be a factor of 10 9 times larger than stars), form via monolithic, run-away collapse of material onto a much smaller (stellar?) seed BH, or by the hierarchical process of merging of smaller-component, intermediatemass BHs. There are recent claims of discovery of intermediate mass BHs weighing M at the cores of globular clusters, but their true nature is still under debate [62]. The fact that bright AGN have been observed at z 6 [63], less than 1 Gyr after the big bang, implies that BH formation must be a very efficient process, at least in the early Universe. Strong 14

15 correlations between properties of SMBHs in the local Universe and the galaxies that host them (such as the black hole mass vs. brightness or velocity dispersion of the stellar bulge) suggests a process of symbiotic growth between the two components. The galaxy spheroid feeds the SMBH with gas and dust from stars; at the same time, the galaxy itself is forming around (and being shaped by) the central gravitational potential well due to the SMBH. Super-massive black hole growth can continue until all fuel in the environment is consumed. On the other hand, this growth may be a self-regulated process: accretion onto the central nucleus is likely to result in some feedback in the form of outward radiation pressure or a strong wind, that could expel the fueling supply of gas from the environment, automatically halting AGN activity as well as stellar formation near the nucleus 6. Phenomenological models incorporating such processes have been elaborated upon by several authors [64-66], and are able to reproduce the observed correlations between black hole mass and bulge luminosity [67] or bulge velocity dispersion [68, 69]. If the power of the wind is substantial, there could be a large component of the energy output of AGN that is invisible across the radiated spectrum, since that energy is instead removed mechanically. Clues about the formation environment and evolution of AGN can also be drawn from their spatial clustering statistics [70]. From their observed angular (sky-projected) clustering, evidence has emerged that while X-ray and sub-mm detected sources trace the same large scale structure, they do not necessarily coincide [71]. This is consistent with proposed evolutionary scenarios in which peak AGN (X-ray) activity is delayed from the initial peak of stellar (sub-mm) activity while the black hole is still growing [72], thus providing a natural explanation for the failure to detect many sub-mm selected sources with Chandra, and vice versa [73]. On the other hand, evidence for a strong correlation between X-ray detections and far-infrared sources has recently been emerging, and telling us in more detail about type 2 AGN populations. Recall that obscured AGN are likely to outnumber their unobscured counterparts by a large factor, and it is likely that the number of Compton-thick AGN has been severely underestimated, simply because of the difficulty of identifying them. They are expected to be bright in the IR because of the primary radiation component being absorbed and thermalized by the torus. From deep observations of two well-studied regions of the sky (referred to as the Lockman Hole and the Hubble Deep Field North), Fadda et al. [74] found that (while the fraction of mid-infrared sources with X-ray counterparts is more than 10 per cent) at least 30 per cent of X-ray sources have mid-infrared (MIR) counterparts discovered by ISO (the Infrared Space Observatory), and this fraction increases to 63 per cent for the sources detected in the higher energy 5 10 kev band, where the optical depth through obscuring gas is smaller. Moreover, the luminosity function (LF) of infrared galaxies observed by missions such as the InfraRed Astronomy Satellite (IRAS) and ISO has been recently inferred to evolve steeply to z 0.8 and flatten thereafter [75]. While this IR wavelength regime is dominated by obscured star-formation, reprocessed AGN accretion activity may also have a non-negligible contribution. Radiative transfer modelling of the broad-band spectral energy distribution (SED) of a number of ISO sources has suggested that absorbed AGN emission can indeed account for observed far-infrared SED peaks [20, 76]. This evidence suggests that X-ray type 2 Seyferts and quasars may be ubiquitously detected in the m regime. Indeed, Gandhi & Fabian [30] were able to construct a synthesis model for the CXB and X-ray number counts, based on the assumption that type 2 obscured AGN are traced by a LF which follows 6 An important scaling quantity as regards radiation feedback is the Eddington accretion rate, that defines the maximum rate at which steady, spherically-symmetric matter can be accreted onto a compact source without being blown away. This can be computed from the balance between radiation pressure and gravity, and scales directly with the BH mass. The associated radiative luminosity is the Eddington Luminosity and is erg s 1 for M BH = 10 6 M. 15