|Annu. Rev. Astron. Astrophys. 1997. 35:
Copyright © 1997 by . All rights reserved
2.3. Simultaneity of the Flux Variations at Various Energies
SIMULTANEITY OF THE UV AND OPTICAL FLUX VARIATIONS: CONSEQUENCES FOR THE NATURE OF THE OPTICAL-UV CONTINUUM The long viscous time scale for a standard optically thick, geometrically thin accretion disk (Shakura & Sunyaev 1973, Pringle 1981) is incompatible with models in which rapid flux variability is caused by variable fueling (Clarke 1987). That variable fueling does not cause rapid optical-UV-EUV variability is confirmed by the simultaneity of the variations in those wavelength bands; the current tightest limit on the time delay between the UV and optical flux variations is 0.2 days in NGC 4151 (Crenshaw et al 1996) and 0.25 days between the UV and EUV in NGC 5548 (Marshall et al 1997).
Consider the example of a disk around a black hole of mass 4 × 107 M. Because information cannot travel through the disk faster than the sound speed (3 × 106 cm s-1 for T = 105 °K), the delay between the rings in the disk emitting at 5400 Å and 1350 Å should be > 3 years, in clear contradiction to the simultaneous flux variations in the UV and optical ranges (Krolik et al 1991, Collin-Souffrin 1991, Courvoisier & Clavel 1991).
SIMULTANEITY IN THE UV, SOFT-, AND HARD-X-RAY RANGES: REPROCESSING IS CURRENTLY THE BEST MODEL Given that viscous heating cannot be the emission process of the rapidly variable optical-UV continuum, the best alternative, suggested by the X-ray variations, is reprocessing. Simultaneity and proportionality on time scales of weeks to months between the medium energy X-ray flux and optical-UV flux have been observed in several AGN (NGC 4151: Perola et al 1986, Warwick et al 1996; NGC 5548: Clavel et al 1992, Tagliaferri et al 1996). This is readily explained by a model in which the optical-UV flux is emitted by an optically thick dense medium irradiated by a nearby variable central X-ray source (Collin-Souffrin 1991, Krolik et al 1991, Nandra et al 1991, Molendi et al 1992, Haardt & Maraschi 1993, Rokaki et al 1993, Petrucci & Henri 1997); this dense medium could be the accretion disk.
In one of the most actively investigated models, fueling occurs through an accretion disk, with angular momentum removed magnetically by field lines threading the disk surface. Explosive reconnections dissipate significant power via magnetic flares within the disk corona (Galeev et al 1979, Blandford & Payne 1982), and X-ray emission is produced via inverse Compton emission in the hot corona surrounding the cooler accretion disk (Haardt & Maraschi 1991).
These flares and variations in the optical depth of the corona (Haardt et al 1997) could be the primary cause of variability on time scales of days or less and could cause variations of the X-ray emission even with a constant accretion rate. This Comptonization model explains the average power-law slope at medium energy, the high-energy cutoff, the reflection hump, and the iron line emission. Reprocessing readily explains the correlation among power-law X rays, soft X rays, and UV emission on short time scales. In fact, it seems at present to be the only viable explanation.
If the corona is not uniform but is patchy (Haardt et al 1994, 1997), the reprocessed radiation is only a fraction of the UV emission and the rest is presumably accretion energy dissipated within the optically thick disk. The fraction of the UV emission due to reprocessed radiation can change "secularly." This is quite clear in NGC 4151, as can be seen in Figures 1b and 3. A major test of the reprocessing scenario is to verify the energy budget among the variable components, i.e. to check that the energy of the medium-hard X-ray component exceeds that of the the UV (and perhaps also the soft X-ray) component. This has to be checked with care and is at present very uncertain (cf Perola et al 1986, Ulrich 1994, Edelson et al 1996).
Figure 3. X-ray flux (absorption corrected, 2-10 keV) versus UV flux (1440 Å) for NGC 4151. Right: ASCA observations in November 30-December 13, 1993, with best-fitting linear correlation. Left: EXOSAT observations in December 16, 1984, to January 28, 1985 (large crosses), and November 7-19, 1983 (small crosses), with best-fitting linear correlation. The good correlation of the UV and X-ray flux on a time scale of weeks and months breaks down at long (years) and short (days) time scales. [Adapted from Warwick et al (1996), Perola et al (1986).]
The origin of the soft X-ray emission in the reprocessing model is not definite. An attractive possibility is that the soft excess is due to reprocessing of the Comptonized emission into thermal radiation. This clearly has strong implications on the expected correlation of variability in different bands.
On time scales of years, the proportionality of the UV and medium X-ray fluxes breaks down, as is evident in NGC 4151 (Figure 3), NGC 5548, and F9, with the UV varying more than the X-ray flux (Morini et al 1986b, Perola et al 1986, Clavel et al 1992, Warwick et al 1996). In NGC 4151, a large slow-varying UV component has disappeared and reappeared during the lifetime of IUE, between 1978 and 1996 (Figures 1b and 3).
On time scales of days, the UV and X-ray variations are correlated, though not in a detailed way. In NGC 5548 and NGC 4151, the short time-scale variability amplitude is smaller in the UV than in the X-rays (Türler et al 1996 for NGC 5548; see Figure 2 for NGC 4151). The discrepancy from proportionality could be caused by variations of the soft X-ray emission due to viscous effects.
High-luminosity AGN tend to have larger UV/X-ray flux ratios when compared to low-luminosity AGN, which could pose an energy budget problem for the reprocessing scenario. This could be solved if the X-ray source is anisotropic and the disk receives more X-ray flux than can be inferred from the observed flux. Anisotropy of the Comptonized emission could account for a deficiency of only a factor of 2-3, but not more. In any case, at present, for high-luminosity AGN, the data are insufficient to establish whether or not there is simultaneity between the optical-UV and X-ray variations, and thus the relevance of the irradiation model for high-luminosity AGN is still uncertain.
VARIABILITY AND THE STARBURST MODEL A contrasting model views AGN as giant young stellar clusters (Terlevich et al 1992). In this case, variability results from the random superposition of "events" - supernova explosions generating rapidly evolving compact supernova remnants (cSNRs) due to the interaction of their ejecta with the high density circumstellar environment. This model is supported by the striking similarity between the optical spectra of AGN and of cSNRs (e.g. Filippenko 1989). The characteristics of an event (i.e. its light curve, amplitude, and time scale) result from the combination of complicated processes (Terlevich et al 1992). Still, the light curves of AGN of various absolute luminosities and redshifts can be predicted from this model and are found to be consistent with the observed dependence of the structure function (the curve of growth of variability with time) on luminosity and redshift (Cid Fernandes et al 1996, Cristiani et al 1996).
The light curves of cSNRs are still poorly known. At present, they appear to be consistent with the optical light curves of the low-luminosity AGN NGC 4151 and NGC 5548 (Aretxaga & Terlevich 1994). Much more detailed light curves of cSNRs are required to make a definitive check. The production of strong and rapidly variable X-rays is still difficult in this model.