2.3.3 X-ray Reflection and Absorption

Figure 3 shows a single power law fit to the composite spectrum of 12 Ginga observations of 8 Seyferts, whose detailed X-ray properties have been described in Refs. 58-64. The data-minus-model residuals in the lower figure comprise an Fe emission line at 6.4 keV, a dip in the 7-10 keV range and another excess above 12 keV, all of which presumably arise from reprocessing of incident radiation from the central hard X-ray source. The line, which typically has an equivalent width (EW) ~ 150 eV and a breadth (FWHM) ~ a few keV, has been found in 90% of all Seyferts. ⁶⁵ Figure 4 shows how the EW depends on the incident angle , assuming that hard X-rays impinge on a semi-infinite slab of optically thick matter. The solid angle assumed equals 4, so the expected EW becomes / 4 times smaller, where is the ``real'' solid angle. Obviously, observed widths can only be reproduced by a very large -value, i.e., the hard X-rays should hit the surface at almost grazing incidence, as in the case of an accretion disk. A spherical geometry cannot produce an EW in excess of ~ 100 eV, and is therefore ruled out. ⁵⁷ The accretion disk must also be optically thick, in order to produce enough fluorescence flux. ⁶⁴ The observed correlation between the EW and the relative strength of the soft excess also supports the presence of an accretion disk. ⁶⁶ Nandra ⁶⁵ et al. argued that the accretion disk picture seems consistent with available data, in spite of some remaining problems (see below).

Figure 3. Single power law fit to the composite spectrum of 8 Seyferts, observed by Ginga. A spectral index ~ 0.7 seems to be indicated, but note the residuals in the lower part. An improved model which takes these into account is shown in Fig. 5. From Ref. 57.

Figure 4. Equivalent widths as a function of incident angle , as hard X-rays impinge on a face-on semi-infinite slab of optically thick matter. Observed values can only be reproduced if the hard X-rays are ``reflected'' at almost grazing incidence, in support of an accretion disk geometry. From Ref. 57.

The observed FWHM values are consistent 48 with gravitational and Doppler broadening at a typical radius ~ 10r_g. The 6.4 keV line profile may thus yield important information about physical conditions in the inner disk. A general relativistic formula for the ratio between emitted and observed photon energies, using the Schwarzschild metric, is ⁶⁷

(6)

where = 2r_em (1 + tan² ) / r_g, and being two angles from geometrical optics. Integrating Eq. (6) over the line emitting part of the accretion disk enables a calculation of the line profile. ⁶⁸ The first term contains the dependence on radius, i.e., the gravitational redshift, whereas the second one yields the kinematic Doppler-shift, which can be either ``blue'' or ``red''.

Typical profiles obtained using this procedure show some similarities with the double peaked ones in standard theory, but in this case the blue horn appears much brighter than the red one, due to time dilation, aberration and blueshift. A high inclination angle has a similar effect. For low inclination angles i the line profile peaks close to its rest energy, but a gravitationally redshifted wing becomes apparent for small radii. The line width increases with i, as does the centroid energy. However, Matt et al. ^{69, 70} have showed that additional features emerge between the two horns for high inclination angles, due to a general relativistic effect. It follows that both the line width and the centroid energy have maxima at about i ~ 80^°.

In principle, X-ray line and continuum variability correlations may be used to map out the core region in AGN, since line-emitting regions close to the central source should be the first to respond to a continuum change. Fabian ⁴⁸ and Stella ⁷¹ argued that the space-time metric and the central mass may then be directly measurable.

Recent observations have cast some doubts on this procedure. ⁷² The accretion disk should be highly inclined in about 15% of a sample of randomly oriented objects, causing an unobserved centroid redshift 0.2 keV. It is unclear whether this is due to a selection effect, caused by absorption. Furthermore, the large EW (~ 160-300 eV) observed in some sources (such as NGC 6814) seems difficult to explain using the simple models adopted so far. ⁶⁵

As regards the other residuals in Fig. 2, the absorption feature at ~ 7-10 keV has been identified with edge absorption, which has also been found in some SBHCs. ⁵⁷ The excess above 10 keV has been attributed to Comptonized flux which grows with energy, until saturation sets in at a few tens of keV, partly due to incoherence of the scatterings, ³¹ and partly to the increase of penetrative power, which increases the number of scatterings before the radiation can escape. ⁷³ Figure 5 shows a model which includes the above reflection components (iron K-line and edge), as well as a power law. The spectral index for the ``new'' power law fit is ~ 0.9, somewhat steeper than the standard value.

Figure 5. (a) Multi-component fit to the same data as in Fig. 3, composed of a power law, the Fe line and edge and a ``warm absorber''. (b) Reflection component only. From Ref. 49.

To summarize, it is now established that most Seyfert galaxies indeed show the above X-ray reflection features. The same applies to quasars (QSOs) nearby enough (or bright enough) to yield good-quality X-ray spectra, such as 3C 273 and E1821 + 643. ^{66, 74, 75} Models which have the ability to include these features will (if successful) have a significant predictive power, but the information content as to the geometry of the reprocessing region, its size and the mass of the central object seems limited at present. However, the fact that X-ray reflection features and a hard X-ray power law with about the same spectral index have been found in both AGN and X-ray binaries may imply that detailed knowledge of the central engine may be gained by comparing different types of objects, using forthcoming X-ray satellites.