2.3. Observations of Broad Iron Lines
The X-ray reflection spectrum was first clearly seen by the Japanese X-ray observatory Ginga (Pounds et al 1990; Matsuoka et al 1990). The CCD detectors on board the Advanced Satellite for Cosmology and Astrophysics (ASCA) were the first X-ray spectrometers to provide sufficient spectral resolution and sensitivity for investigating the profile of the iron line in detail. The first clear example of a broad skewed iron line came from a long ASCA observation of the Seyfert 1 galaxy MCG-6-30-15 (Fig. 6; Tanaka et al 1995). The sharp drop seen at about 6.5 keV both demonstrates the good spectral resolution of the CCD detector and, as discussed above, constrains the inclination of the disk to be about 30 degrees. If the inclination were greater then this blue edge to the line moves to higher energies (as seen in the broad iron line of the Seyfert 2 galaxy IRAS 18325-5926, Iwasawa et al 1996a). The redward extent of the line constraints the inner radius of the emission to be 7 rg and the overall shape means that most of the line emission is peaked within 20 rg
Figure 6. The line profile of iron K emission in the ASCA SIS spectrum of the Seyfert 1 galaxy MCG-6-30-15 (Tanaka et al 1995). The emission line is very broad, with full width at zero intensity of ~ 100,000 km s-1. The line shape is skewed toward energies lower than the rest-energy of the emission line (6.35 keV at the source redshift of 0.008). The dotted line shows the best-fit line profile from the model of Fabian et al (1989), an externally-illuminated accretion disk around a Schwarzschild black hole.
Nandra et al. (1997a) and Reynolds (1997) used ASCA data to study the iron line in over 20 Seyfert 1 galaxies and found that most are significantly broader than the instrumental resolution. In a typical ASCA observation of an AGN, the signal-to-noise ratio of the detected iron line is insufficient to study the line beyond simply measuring that it is broad. To combat this problem, Nandra et al. (1997a) have summed together the data from many AGN to produce an average iron line profile. They find that the average line has clear extension to low energies. To the extent that individual sources can be studied, the inferred inclinations of the accretion disks are clustered around 30 deg, indicating some bias to the selected galaxies. Such a bias is expected within the context of the ``unified model'' of Seyfert galaxies (Antonucci 1993). In brief, the unified model states that Seyfert galaxies possess an obscuring torus on scales larger than the accretion disk. When one views the central regions of the AGN along a line of sight that is not blocked by the torus, one seens a type-1 Seyfert galaxy. Otherwise, one would see a type-2 Seyfert galaxy. If the accretion disk and tori are co-aligned (as might be expected on the basis of dynamical models; Krolik & Begelman 1988), and the tori have an average opening angle of 30-40 degrees, then we would naturally expect a bias in the measured disk inclinations in a sample of Seyfert 1 galaxies.
It should be noted that, in some cases, results on the inclination of the disk implied from the broad iron line and orientations of the systems inferred from the other techniques, e.g., ionization cone, radio jet, broad line clouds etc., differ from each other (e.g., Nishiura, Murayama & Taniguchi 1998). The prime example is NGC4151, for which the iron line suggests the inner accretion disk to be almost face-on (Yaqoob et al 1996) whilst the other observations point to an edge-on system. The recent suggestion by Wang et al. (1999) that a significant proportion of the iron line in NGC4151 is scattered into our line-of-sight by an electron scattering disk atmosphere may explain this discrepancy. Also, such a difference in geometry depending on scales of interest may be expected due to warping of the accretion disk or a multiple merger (e.g., NGC1068, Bland-Hawthorn & Begelman 1997). Iron lines from warped accretion disks have been studied theoretically in some detail by Hartnoll & Blackman (2000). In other objects the inclination inferred from the iron line agrees with the classification of the object (e.g. MCG-5-23-16; Weaver, Krolik & Pier 1998).
The observed iron line profiles in active galaxies are not necessarily solely from the accretion disk. Absorption and extra emission may alter the iron line emitted by the accretion disk. Nandra et al (1999) reported detection of an absorption feature at about 5.7 keV imposed on the broad iron line profile obtained from a long observation of NGC3516 (see Fig. 7). They suggest that this feature is due to K resonant line scattering by highly ionized iron (with an intrinsic energy of 6.9 keV). The redshift of the absorption feature has been interpreted as evidence for matter infalling onto a black hole. However, gravitational redshifting of resonant absorption which could occur when the line photons are passing through the hot corona above the disk can also account for the observed feature if it occurs close to the black hole (Ruszkowski & Fabian 2000).
Figure 7. The time-averaged iron line profile observed in the Seyfert galaxy NGC3516, obtained from a long ASCA observation (Nandra et al 1999). It shows a broad red-tail as well as a resonant absorption feature around 5.4 keV.
The iron line profiles observed in some Seyfert galaxies, especially Compton-thin Seyfert 2s (or those classified as Seyfert 1.8 or 1.9 in optical), have significant contribution of a narrow line component originating from matter far away from a the central black hole, e.g., a molecular torus (Weaver & Reynolds 1998). Such a narrow component would become clear if a primary X-ray source had faded away as has been seen in NGC2992 (Weaver et al 1996) and NGC4051 (Guainazzi et al 1998), since light travel times puts a fundamental limit on how rapidly line emission from the torus can respond to the central source. The iron line variability observed in NGC7314 (Yaqoob et al 1996 is exactly what is expected from a line consisting of a broad component originating from the accretion disk and a torus line. In this source, the broad iron line responds to changes in the primary power-law X-ray flux, while a narrow line component is found to be constant. It is interesting to note that this composite nature is not always found in Seyfert 1 galaxies. Any contribution of a stable, narrow line component in MCG-6-30-15 has been found to be very small (Iwasawa et al 1996; 1999). This might suggest that, in comparison to Seyfert 2 galaxies, Seyfert 1 galaxies have tori that have either smaller optical depths or smaller geometric covering factors.
Not all bright Seyfert 1 galaxies have iron lines so broad that the disk is required to extend down to the marginally stable orbit (i.e. < 6 rg). IC4329a is found to have a relatively narrow line (Done et al 2000) which can be modelled with a disk of inner radius of about 50 rg. The disk within this radius may either be missing or highly ionized.
Rapid X-ray variability in active galaxies predicts the iron line also to vary in response to the continuum with a small time lag. The light crossing time for 10 rg in a black hole with a mass of 107 M7 M is about 500 r1 M7 s, which is much shorter than an integration time required for ASCA to collect enough line photons to perform a meaningful measurement in X-ray bright AGN (> 10-11 erg cm-2 s-1). It means that no reverberation effects in the line can be detected with ASCA. We will return to reverberation effects in Section 5.
Despite the inability of current detectors to measure reverberation, significant and complex variations of the iron line in profile as well as intensity have been observed in the Seyfert 1 galaxy MCG-6-30-15 (see Fig. 8; Iwasawa et al 1996b). During the first long observation of this galaxy in 1994, a line profile with unusually strong blue peak was found in a time interval of a bright flare whilst the line showed a very-broad, red-wing dominated profile during a deep minimum period. In this very-broad state, line emission from within 6 rg is required to explain the line profile and width. Possible theoretical interpretations of this are discussed in Section 4. A succession of large flares on the approaching side of the disk could produce the blue-peak dominated line profile, although it can also be explained if the line is produced predominantly at large radii (~ 100 rg). It is worth noting that this bright flare showed a continuum spectral evolution similar to that seen in a shot in a Galactic black hole candidate, e.g., Cyg X-1 (Negoro et. al 1995).
Figure 8. Time-averaged (upper panels) and peculiar line profiles (lower panels) of the iron K emission from MCG-6-30-15 seen in the two long ASCA observations in 1994 (left) and 1997 (right). In the 1994 observation, a very broad profile with a pronounced red-wing is seen during a period of Deep Minimum of the light curve (lower left), compared to the time-averaged line profile shown in the upper panel. In contrast, during a sharp flare in the 1997 observation, whole line emission is shifted to energies below 6 keV and there is no significant emission at the rest line-energy of 6.4 keV (lower right). Both peculiar line shapes can be explained by large gravitational redshift in small radii on the accretion disk.
Another peculiar line shape seen in a brief period (~ 1 hr) of a flare during the 1997 long observation of MCG-6-30-15 also requires a large redshift within 6 rg. The blue wing of this line is, this time, shifted well below 6 keV and no significant line was detected around 6.4 keV. A possible explanation is that the line production occurred either in a thin annulus at 4 rg or a small patch at ~ 2.5 rg on the approaching side of the disk (Iwasawa et al 1999).
Detailed studies of the RXTE data on the iron line variability in MCG-6-30-15 have presented a puzzling problem; most of the line flux appears to be constant in spite of strong continuum changes (Lee et al 1999; Reynolds 2000: see also studies of NGC5548 by Chiang et al 2000 and of the Galactic Black Hole Candidate Cygnus X-1 by Revnivtsev, Gilfanov & Churazov 1999). There should therefore be some self-regulating mechanism to produce a constant line flux, which has yet to be understood. However, a separate investigation of the broad red-wing and narrow core of the iron line for the 1994 observation (Iwasawa et al 1996) has revealed interesting behaviours of each component. The narrow core remains constant on time scales shorter than 103 s but follows the continuum variations on longer time scales (> 104 s). In contrast, the broad red-wing appears to follow the continuum on the short time scales. This is consistent with a line produced from a relativistic disk (see also Blackman 1999). Also puzzling is an anti-correlation between the reflected fraction and the equivalent width of the iron line measured in MCG-6-30-15 (Lee et al 2000) and NGC5548 (Chiang et al 2000).