Annu. Rev. Astron. Astrophys. 1997. 35: 445-502 Copyright © 1997 by Annual Reviews. All rights reserved

5.2. Mapping the Velocity Field from the Emission Lines

That the fastest moving gas is the closest to the center implies that radiative acceleration is less important than gravity and rotation. This dynamical information supports the connection of AGN line variability to black hole mass, providing that radial motions are not dominating the velocity field.

THE SEARCH FOR RADIAL MOTIONS FROM THE HIGH-IONIZATION LINES     To first order, the HIL profile variations do not show the systematic change of one wing before the other that would be expected for purely radial flows (spherical winds, spherical accretion). This requires the main motions of the HIL clouds to be circulatory with only minor components of net infall or outflow. The data are consistent with the HIL clouds being in circular orbits in a disk, or having "chaotic" motions along randomly oriented orbits in the gravitational field of the central mass.

The velocity-resolved MEM inversion of Equation 2 applied to the best-sampled data of NGC 4151 and NGC 5548 shows that the time delays vary as function of velocity in a way that is roughly consistent with virialized motions. The central mass so derived is ~ 10⁷ M for NGC 4151 (Ulrich & Horne 1996). In NGC 5548, it is between 2 × 10⁷ M (for two-dimensional random motions) and 8 × 10⁷ M (for three-dimensional random motions; Done & Krolik 1996).

In both NGC 4151 and NGC 5548 (and also in F9, Recondo-González et al 1997), however, the best-sampled data show small differences in the transfer functions of the blue and the red wing of C IV, with the response of the red wing being the stronger at small delays. This small asymmetry could, at first sight, be taken as a subtle indication of radial motions. On the other hand, optical depth effects in a rotating/outflowing wind produce a differential response across the C IV profile, with the red wing leading the blue wing (Bottorff et al 1997, Murray & Chiang 1997).

DIFFERENT EMITTING REGIONS FOR THE HIGH- AND LOW-IONIZATION LINES?     As stated in the introduction, the LIL come from very dense gas with high column density, whereas the HIL come from a more diffuse photoionized medium. Can the variability data establish the existence of these two different emission regions? The answer is positive but provisional: In the few AGN with good variability data on C IV and H;, the inversion of Equation 2 gives a transfer function for H; that is small near zero lag, indicating the near absence of matter along the line-of-sight, whereas in contrast, the transfer function of the C IV wings peaks near zero lag (Ulrich & Horne 1996, Done & Krolik 1996). This is what is expected if the Balmer lines come from a disk at small inclination (there is no matter close to the line of sight) and the HIL come from a broad cone or cylinder, in which case some matter lies along the line of sight producing a nonzero response near zero lag.

According to Wanders & Peterson (1996), however, the lack of response at zero lag for the transfer function of the H; line is spurious and due to the combined effects of noise in the data and lack of resolution. On the other hand, Keith Horne, responding to a friendly challenge, has run his MEMECHO program on sets of simulated noisy data prepared by Dan Maoz and successfully identified which data corresponded to a transfer function peaking at zero and which ones did not (K Horne, private communication; Maoz 1997). This exercise gives weight to the results obtained from inversion of Equation 2, but, clearly, a final answer to this question awaits data of higher quality than presently available.

THE NATURE OF THE MOTIONS OF THE HIL GAS - IS THE DISK OPAQUE?     If indeed a disk is present in AGN (and emits the Balmer lines and the Fe II multiplets), then two important points about the disk need to be specified before interpreting the velocity information on the HIL. First, is the disk opaque or transparent (can we see what is happening below the disk)? Second, could HIL clouds survive if they cross the disk in their chaotic orbits?

Calculations of accretion disk structure (Huré et al 1994) suggest that the disk becomes self-gravitating at 100 r_S (r_S = 2GM_BH / c² is the Schwarzschild radius), much smaller than the radius of the region emitting the Balmer lines, although plausible mechanisms could stabilize the disk at a much larger radius (magnetic fields or internal dissipation; e.g. Sincell & Krolik 1997). It is possible, and we believe likely, that the disk (whether in a continuous structure or broken up in clouds in its outer parts) joins to the molecular torus, and nowhere is it sufficiently transparent that we can see the far side. This has profound consequences for the interpretation of the observed velocity field. In a biconal flow, only the near half would be visible, and if there were truly chaotic motions, only one half of each orbit would be seen. In addition to being opaque, the disk could have a column density such that free-flying clouds would be destroyed when crossing the disk. In this case, only stars can have chaotic motions, and if the HIL BLR clouds are the (modified) atmospheres of stars (Penston 1988, Alexander & Netzer 1994), they could collectively partake in pure gravitational motions, but we would still see only those clouds on the near side of the disk.

A solution to this puzzle is offered by the fact that in radio-quiet AGN, the HIL are blueshifted (by 0 to ~ 1500 km s^-1) with respect to the low HIL, which themselves are at the host galaxy redshift (Gaskell 1982, Wilkes 1984, Corbin 1995, Sulentic et al 1995a, Marziani et al 1996). This indicates that the gas emitting the highly ionized lines is moving towards us, probably emanating from the disk, and still retaining a large part of the angular momentum it had in the disk (thus allowing a derivation of the central mass). The origin of the observed range of the HIL blueshifts is unclear - orientation can explain only part of it. Additional evidence for the presence of outflowing material in AGN includes (a) the blueshifted absorption lines in the emission line profile of a significant number of AGN, (b) the blueshift of the coronal lines (Wagner 1997) and (c) the shift of the reflected broad lines in Seyfert 2.

Magnetically accelerated outflows from accretion disks and radiatively driven winds are promising models for the formation and evolution of the highly ionized gas clouds (Blandford & Payne 1982, Emmering et al 1992, Königl & Kartje 1994, Murray & Chiang 1997, Bottorff et al 1997). Clouds, probably filaments, are pulled from the originally dense low-ionized material of the disk and subjected to the intense ionizing field, forming a more diffuse highly ionized outflowing medium. The densest, coolest inhomogeneities form the BLR clouds emitting the prominent lines (and if one such cloud happens to be crossing our line of sight to the AGN center, it should produce a blueshifted absorption line as, in fact, often observed in AGN). The hottest phase is detected as the fully ionized component of the warm absorber, which could also produce the blueshifted absorption lines in the wings of the HIL. Many features of these promising models remain unspecified and can be adjusted to accommodate the observations.

Two other models are compatible with an opaque accretion disk and the simultaneous response of the blue and red wings of the HIL. Both solve the confinement problem and include elements important for the formation and evolution of the BLR, but they do not specifically include magnetic forces. In the bouncing gas clouds model (Mathews 1993), the clouds congregate at a preferred radius where radiation forces and gravity are balanced. Fluctuations in the gas pressure move the clouds radially in and out, but most of the clouds tend to come back to the preferred radius set by the radiation level. In the bloated stars scenario, the BLR clouds are the modified atmospheres of some of the stars of the central stellar core and thus move along Keplerian orbits around the black hole (Penston 1988, Alexander & Netzer 1994).