Annu. Rev. Astron. Astrophys. 1977. 15: 69-95 Copyright © 1977 by Annual Reviews. All rights reserved

5. BALMER DECREMENT AND LINE PROFILES

The problem of reddening is directly related to the interpretation of the Balmer decrement in Seyfert galaxies. The Balmer emission in most astrophysical plasmas, such as HII regions and planetary nebulae, arises from radiative ionization followed by recombination. Downward cascades within the newly recombined atoms produce emission lines whose intensity ratios can be calculated. For a nebula optically thick only in the Lyman lines, the ratio of Balmer-line intensities (the Balmer decrement) is H: H: H = 2.8:1.0:0.47 for temperatures near 10⁴ °K (Miller 1974). In Seyfert galaxies, the decrement is invariably steeper than this, meaning that the redder Balmer lines are relatively more intense than predicted. There are two categories of explanation. One is that some mechanism selectively populates hydrogen energy levels from the ground up, which increases the relative intensity of the redder lines in a series. Such a mechanism could be collisional excitation, radiative excitation by a continuum, or Balmer-line self absorption. That such effects are important under the high-density conditions found in Seyfert nuclei has been shown by Netzer (1975) and Shields (1974). The alternative circumstance is that the Balmer lines are reddened by dust so that the observed decrement is steepened. If sufficient reddening is present to produce the observed decrement, there must be a lot of absorption by the same dust. In some cases, several magnitudes of absorption would be needed in the visible spectrum. This would mean that Seyfert nuclei are even brighter than they appear. Of course, the absorbed energy would have to come out somewhere in the spectrum, and the dust could be expected to reradiate in the infrared. The fact that Seyfert nuclei are strong infrared sources might mean that there is dust in some.

To resolve the decrement problem, it is therefore necessary to consider the effects that reddening would have both on the lines and the continuum. The most accurate spectrophotometric studies to date of the decrements in Sy 1, through the H line, have been unable to account for the decrements with any assumed amount of normal reddening (Osterbrock et al. 1975, Osterbrock 1976). If dust reddening were responsible for the steep decrements, the continua on average should become redder as the decrement increased. However, there are no empirical correlations between the H: H ratio and the continuum colors, either in the optical or in the optical vs. the infrared (Adams & Weedman 1975, Stein & Weedman 1976). All of these authors therefore concluded that in the Sy 1 nuclei radiative recombination was not the only contributor to excitation of the Balmer lines. Some other of the effects mentioned above are probably important in the region where the broad Balmer lines arise and account for the unexpectedly steep decrement.

This explanation does not necessarily apply to Sy 2, however. Virtually all of the conclusions about Sy 2 depend on NGC 1068, which has emission-line profiles, relative intensities, and luminosities that are representative of other Sy 2. NGC 1068 also has a steep Balmer decrement; the Sy 2 generally have steeper decrements than the Sy 1. In NGC 1068, relative [SII] line intensities show clearly that some dust reddening is present (Wampler 1968). Such lines are measurable in Sy 2 because of the greater relative intensity of the forbidden lines in these Seyferts.

Substantial recent results are also available relative to another problem of spectroscopic interest, which is the cause of the broadening of the line profiles. For the Sy 2, the similar asymmetric profiles in both Balmer and forbidden lines are considered as proof that real mass motions cause Doppler broadening. The classical study of NGC 1068 showing discrete clouds moving at different velocities set the tone of this interpretation (Walker 1968). The gas sometimes escapes the nucleus and can spread through large volumes, having a kinetic energy of 10⁵⁶ ergs. The source of this energy is unknown. The situation was not so well resolved for Sy 1, however, as a difficulty arises if the very broad Balmer-line wings are attributed to Doppler broadening. This is the containment problem, because the emitting gas has to remain in a nucleus on the order of a parsec in diameter while moving at 10⁴ km sec^-1. The travel time across the nucleus would be only 10² years, so the gas must be replaced as it flows out, or it must somehow be bound. To avoid worrying about this, the broad wings were attributed to electron scattering, i.e. scattering between the Balmer-line photons and the thermal electrons (e.g. Mathis 1970, Weymann 1970). This explanation was acceptable for line wings that are smooth and symmetric. However, there are now many unquestionable examples of Sy 1 with asymmetric or structured Balmer-line wings (Table 1) so it seems necessary to attribute these to mass motions of some sort.

The mechanism that accelerates and contains the gas is unknown. The empirical association between the continuum luminosity source and the broad lines implies that the gas may be accelerated by radiation pressure, such as described by Mathews (1974) and Blumenthal & Mathews (1975). An early suggestion by Woltjer (1959) was that the broad profiles are caused by rotation, which would overcome the containment problem. However, Woltjer assumed that the gas rotated about the nucleus with a radius about 100 pc and inferred large nuclear masses both from assigning the line width to rotational velocity and by assuming the nuclear luminosity came from stars. Rotation curves of Seyfert galaxies deduced later by the Burbidges showed no sign of the large masses needed. The rotation models were therefore not pursued, especially after it was decided that the nuclear luminosities were not from stars and that the nuclei of Sy 1 could be much smaller than 100 pc. Rotation models may nevertheless prove useful if applied to accretion disks around massive, condensed objects (Hills 1975). On a much smaller luminosity scale, such models explain the broad Balmer lines in cataclysmic variable stars (e.g. Warner 1973). (I am indebted to J. S. Gallagher for pointing this out.) Oke & Shields (1976) argue, however, that the FeII emission in some Sy 1 implies that the gas cannot be in ordered rotation around the continuum source. Regardless of whether the high gas velocities are in random filamentary motions or ordered rotation, the association between gas and luminosity source is further emphasized by the correlation between increasing profile width and increasing nuclear luminosity found in some Sy 1 by Osterbrock et al. (1976).

An extensive set of models for explaining line profiles has been produced by Ptak, Stoner and collaborators (e.g. Ptak & Stoner 1973, Stoner et al. 1974, Hubbard 1975, Ptak & Stoner 1975, MacAlpine 1974). These are the most detailed attempts so far to account for the line profiles, especially in trying to fit the asymmetries. They describe their model as follows (Ptak & Stoner 1973): A point source injects high-energy protons into a partially ionized gas cloud. The protons are decelerated by interactions, and "after they have slowed to speeds of a few times the orbital velocities of bound electrons in the atoms of the cloud, the cross-section for charge transfer during collisions with the atoms becomes large enough that the streamers spend a significant amount of time as atoms themselves, and can emit photons characteristic of hydrogen." These models still face the same dilemma as other mass motion models, in that there must be an unknown source for supplying and accelerating the protons. Osterbrock et al. (1976), Osterbrock & Koski (1976), and Katz (1975) contend that the proton-streaming models cannot explain the great width of some profiles, or the relative widths of HeI, HeII, and hydrogen profiles.

It has been accepted for some time that Sy 1 nuclei have a very inhomogeneous distribution of gas, with the broad Balmer-line wings arising in a denser gas than where the Balmer cores and forbidden lines arise (e.g. Oke & Sargent 1968). Because wings have never been seen on any forbidden lines, it is concluded that the Balmer wings arise in gas with N_e 10⁷ cm^-3 so that forbidden line radiation is suppressed by collisional de-excitations. Such two-phase models for Sy 1 nuclei have received substantial confirmation from the observation that permitted FeII emission lines are often found in Sy 1 spectra, but are not accompanied by forbidden FeII (Sargent 1968, Phillips & Osterbrock 1975, Osterbrock 1976, Phillips 1976, Oke 1972, Boksenberg et al. 1975a, Adams 1975, Oke & Shields 1976). This requires N_e 10⁸ cm^-3. Because of the high densities, only a small mass of gas is needed to produce the broad Balmer-line emission. The luminosity of recombination emission depends on the product (mass) × (density), so the value derived for the mass is determined by that assumed for the density. Adopting N_e 10⁷ cm^-3 Adams & Weedman (1975) found that no more than 4 × 10³ M of ionized hydrogen was ever required to account for the broad lines in Sy 1. Osterbrock et al. (1976) increased N_e to 10⁹ and therefore lowered their maximum mass to 26 M. Clearly, only small amounts of gas are needed to produce the Balmer lines if a sufficient ionization or excitation mechanism is present. Proportionally larger gas masses are needed for the emission in the forbidden-line volume, however, because the density is 10⁴ cm^-3. The summary scenario of an Sy 1 nucleus is then a highly localized continuum source, accompanied by high-density clouds or filaments that produce the broad Balmer lines. All of this is immersed in a volume, perhaps 10⁶ times greater, that contains the lower-density gas producing the Balmer-line cores and the forbidden lines.