LINERs AS LOW-LUMINOSITY ACTIVE GALACTIC NUCLEI

Many nearby galaxies contain optical signatures of nuclear activity in the form of LINER nuclei. LINERs may be the weakest and most common manifestation of the quasar phenomenon. The physical origin of this class of objects, however, has been ambiguous. I draw upon a number of recent observations to argue that a significant fraction of LINERs are low-luminosity active galactic nuclei. AGN CENSUS IN NEARBY GALAXIES The local space density of active galactic nuclei (AGNs) has bearing on a number of issues in extragalactic astronomy, including the fraction of galaxies hosting massive black holes, the cosmological evolution of quasars, and the contribution of AGNs to the cosmic X-ray background. It is therefore of fundamental importance to establish the extent and nature of nuclear activity in nearby galaxies. This contribution summarizes recent efforts to survey nearby galactic nuclei, discusses complications regarding the interpretation of the results, and presents a variety of fresh observational perspectives that help toward reaching a coherent understanding of nuclear activity in nearby galaxies. Optical surveys find that a large fraction of nearby galaxies have nuclei that emit weak emission lines with a spectrum unexpected for photoionization by normal stars. Heckman (1980) identified low-ionization nuclear emission-line regions (LINERs) as a major constituent of the extragalactic population, particularly among early-type galaxies. The optical spectra of LINERs broadly resemble those of traditional AGNs such as Seyfert nuclei, except that they have characteristically lower ionization levels. These findings were strengthened by a number of subsequent studies, as reviewed by Ho (1996). The latest and most sensitive survey of this kind was completed by Ho et al. (1997a, and references therein) using the Hale 5 m telescope at Palomar Observatory. Long-slit spectra of exceptional quality were taken of the nuclear region of a magnitude-limited (BT ≤ 12.5 mag) sample of 486 northern (δ > 0) galaxies that constitutes an excellent representation of the typical nearby galaxy population. Invited review to appear in The 32 COSPAR Meeting, The AGN-Galaxy Connection (Advances in Space Research). Fig. 1. Detection rate (left) and number distribution (right) of AGNs as a function of Hubble type in the Palomar survey. “Type 1” AGNs (those with broad Hα) are shown separately from the total population (types 1 and 2). The spectra are of moderate resolution (full-width at half maximum [FWHM] ∼ 100–200 km s) and cover two regions of the optical window (4230–5110 Å and 6210–6860 Å) containing important diagnostic emission lines. The main results of the Palomar survey are the following. (1) AGNs are very common in nearby galaxies (Fig. 1). At least 40% of all galaxies brighter than BT = 12.5 mag emit AGN-like spectra. The emission-line nuclei are classified as Seyferts, LINERs, or LINER/H IIregion composites, and most have very low luminosities compared to traditionally studied AGNs. The luminosities of the Hα emission line range from 10 to 10 ergs s, with a median value of ∼10 ergs s. (2) The detectability of AGNs depends strongly on the morphological type of the galaxy, being most common in early-type systems (E–Sbc). The detection rate of AGNs reaches 50%–75% in ellipticals, lenticulars, and bulge-dominated spirals but drops to < ∼ 20% in galaxies classified as Sc or later. (3) LINERs make up the bulk (2/3) of the AGN population and a sizable fraction (1/3) of all galaxies. (4) A significant number of objects show a faint, broad (FWHM ≈ 1000–4000 km s) Hα emission line that qualitatively resembles emission arising from the conventional broad-line region of “classical” Seyfert 1 nuclei and QSOs. Radio observations provide further support for the prevalence of nuclear activity (see the contribution of E. Sadler in these proceedings). Weak radio cores with powers of 10–10 W Hz at 5 GHz are found in ∼50% of nearby elliptical and S0 galaxies (Sadler et al. 1989; Wrobel and Heeschen 1991). Where information is available, the cores have relatively flat spectral indices and nonthermal brightness temperatures (Slee et al. 1994), and the optical spectra of most of these sources are classified as LINERs (Sadler et al. 1989; Ho 1998a). RECENT OBSERVATIONAL RESULTS ON LINERs If LINERs are powered by a nonstellar source, then LINERs clearly would be the most common type of AGNs known in the nearby regions of the universe. However, ever since their discovery, the physical origin of LINERs has been hotly debated. The recognition and definition of LINERs is based on their spectroscopic properties at optical wavelengths. In addition to the AGN scenario, the optical spectra of LINERs unfortunately can be interpreted in several other ways that do not require an exotic energy source (e.g., shocks, hot stars; see Ho et al. 1993 and Filippenko 1996 for reviews). As a consequence, it has often been suggested that LINERs may be a mixed-bag, heterogeneous collection of objects. While the nonstellar nature of some well-studied LINERs is incontrovertible (e.g., M81, M87, M104), the AGN content in the majority of LINERs remains unknown. Determining Fig. 2. Number distribution of morphological types (left) and total absolute blue magnitudes (right) for H II nuclei, all AGNs (LINERs + Seyferts), and LINERs and Seyferts separately. The median of each distribution is marked by an arrow. Adapted from Ho et al. (1997a). the physical origin of LINERs is more than of mere phenomenological interest. Because LINERs are so numerous, they have a tremendous impact on the specification of the faint end of the local AGN luminosity function, which itself bears on a range of issues. A number of recent developments provide considerable new insight into the origin of LINERs. I outline these below, and I use them to advance the proposition that most LINERs are truly AGNs. Host Galaxy Properties LINERs and Seyferts live in virtually identical host galaxies (Fig. 2). The vast majority of both classes occupy bulge-dominated, early-type systems (87% are found in types E–Sbc), which clearly differ from the population of galaxies whose nuclear spectrum indicates photoionization by current star formation (the so-called H II-nuclei), which is dominated by late-type hosts (63% are Sc’s and later). The only noticeable difference in the distribution of morphological types of LINERs and Seyferts is that LINERs occupy a higher proportion of ellipticals. Bars exist with roughly the same frequency within the subsample of disk galaxies in both groups. The similarity in the host galaxy properties of LINERs and Seyferts becomes even more apparent when we examine their absolute magnitude distributions (Fig. 2); they are statistically indistinguishable. Both peak at M BT ≈ –20.5 mag (for H0 = 75 km s −1 Mpc), about 0.4 mag brighter than M BT , the typical absolute magnitude of the field-galaxy luminosity function. The parent galaxies of H II-nuclei, on the other hand, are systematically fainter than the other two groups by ∼0.5 mag in the median. Detection of Massive Black Holes There has been considerable recent progress in the detection of dark masses, plausibly interpreted Note that this paper is concerned only with compact, nuclear LINERs (r < ∼ 200 pc), which are most relevant to the AGN issue. LINER-like spectra are often also observed in extended nebulae such as those associated with cooling flows, nuclear outflows, and circumnuclear disks. Fig. 3 (Left). Decomposition of the Hα + N II region for LINERs and Seyferts. The broad Hα component is shown as a heavy line. Adapted from Ho et al. (1997b). Fig. 4 (Right). Keck spectra of NGC 1052 from Barth (1998). Top panel — Total flux spectrum. Middle panel — Percent polarization. Bottom panel — Stokes flux obtained from Fλ ∗ p. as massive black holes, in nearby galactic nuclei (see Ho 1998c and the contribution by S. Faber). A significant fraction of the known black hole candidates, albeit still a small number, in fact are well known LINERs. These include M81 (MBH ≈ 4×10 6 M⊙), M84 (1.5×10 9 M⊙), M87 (3×10 9 M⊙), the “Sombrero” galaxy (1×10 M⊙), NGC 4261 (5×10 8 M⊙), and Arp 102B (2×10 8 M⊙). Although certainly no statistical conclusions can yet be drawn, these examples nevertheless serve as a powerful proof-of-concept that at least some LINERs are incontrovertibly accretion-powered sources. Detection of Broad-Line Regions Bona fide AGNs such as QSOs and luminous Seyfert 1 nuclei distinguish themselves unambiguously by their characteristic broad (FWHM ∼ few thousand km s) permitted lines which arise from the broad-line region (BLR). The detection of such broad lines in LINERs would constitute strong evidence in favor of the AGN interpretation of these sources. Since the strongest permitted line at optical wavelengths is expected to be Hα, one of the primary goals of the Palomar survey was to search for broad Hα emission. Of the sample of objects with broad Hα emission (22% of the AGN candidates), more than half belong to the LINER category (Ho et al. 1997b; see Fig. 3). This is a very important finding, because it implies that LINERs, like Seyferts, evidently come in two flavors — some have a visible BLR, and others do not. By direct analogy with the nomenclature established for Seyferts, we might extend the “type 1” and “type 2” designations to include LINERs. Approximately 15%–25% of the LINER population are LINER 1s, the appropriate fraction depending on whether the so-called transition objects (Ho et al. 1993) are regarded as LINERs. A remaining, outstanding question, however, is what fraction of the LINER 2s are AGNs. Again, by analogy with the Seyfert 2 class, surely some LINER 2s must be genuine AGNs — that is, LINERs that happen to have no BLR or have an obscured BLR. There is no a priori reason why the unification model, which has enjoyed such popular support in the context of Seyfert galaxies, should not equally apply to LINERs. The existence of an obscuring torus does not obviously depend on the value of the ionization level of the line-emitting regions. If we suppose that the ratio of LINER 2s to LINER 1s is the same as the ratio of Seyfert 2s to Seyfert 1s, that ratio being 1.4:1 in the Palomar survey, we might argue that the AGN fraction in LINERs may be as high as ∼40%–60%. What evidence is there, however, that the unified model is applicable to LINERs? The faintness of the sources in question renders application of the classical spectropolarimetric test (e.g., Antonucci and Miller 1985) impractical for moderate-sized telescopes. An important breakthrough was recently achieved by Barth (1998), who successfully used the Keck 10 m telescope to detect a polarized broad Hα line in the prototypical LINER NGC 1052 (Fig. 4). Weak broad Hα wings were previously found in the total-light spectrum after very careful profile decomposition (Ho et al. 1997b), but the broad line is undeniable in scattered light. Some LINER 2s evidently do harbor obscured BLRs. It would be highly desirable to extend these observations to larger samples. Lest one doubts that the existence of BLRs can be established with the detection of a single broad line, it should be remembered that broad lines are seen in other transitions as well, particularly in the ultraviolet (UV) where contamination by old stars poses less of a problem. The two best examples are M81 (Ho et al. 1996) and NGC 4579 (Barth et al. 1996) which were observed with the Hubble Space Telescope (HST). Finally, note that the minority of AGNs that display so-called double-peaked broad emission lines, whose origin is widely thought to lie in a relativistically rotating accretion disk, in fact very often exhibit LINER-like narrow-line spectra (Eracleous 1998 an references therein). Ultraviolet Emission and Constraints on Shock Excitation The nonstellar nature of LINERs might be revealed through the presence of a central compact source responsible for the photoionizing continuum. The UV band is preferred over the optical because it minimizes contamination from old stars, although it is much more adversely affected by dust extinction. Two imaging surveys performed with the HST (Maoz et al. 1995; Barth et al. 1998) find that LINERs in fact do contain compact UV emission, but in only 20%–25% of the cases. By itself, however, this result is ambiguous. Are the central UV sources in most LINERs obscured by dust, are they in the “off” state of a duty cycle most of the time as suggested by Eracleous et al. (1995), or do the majority of LINERs simply lack a pointlike ionizing source because they are not AGNs after all? There is some indication that the sources detected in the UV tend to be in more face-on galaxies than the undetected sources (Barth et al. 1998). Moreover, as discussed below, LINERs seem to be intrinsically weak in the UV, and this may further contribute to the low detection rates. Mere morphological information, of course, cannot specify definitively the physical origin of the UV emission. For example, the point sources could be simply very compact nuclear star clusters. Indeed, follow-up spectroscopy indicates that the bulk of the UV emission in some sources comes from young massive stars (Maoz et al. 1998). Others, on the other hand, exhibit featureless, power-law continua as expected for an energetically significant AGN component (M81: Ho et al. 1996; NGC 4579: Barth et al. 1996; M87: Tsvetanov et al. 1998). Collisional ionization by shocks has been considered a plausible energy source for LINERs since the discovery of these objects (Fosbury et al. 1978; Heckman 1980). Dopita and Sutherland (1995) recently showed that the diffuse radiation field generated by fast (v ≈ 150–500 km s) shocks can reproduce the optical narrow emission lines seen in both LINERs and Seyferts. In their models, LINER-like spectra are realized under conditions in which the precursor H II region of the shock is absent, as might be the case in gas-poor environments. The postshock cooling zone attains a much higher equilibrium electron temperature than a photoionized plasma; consequently, a robust prediction of the shock model is that it should produce a higher excitation spectrum, most readily discernible in the UV, than photoionization models. In all the cases studied so far, the UV spectra are inconsistent with the fast-shock scenario because the observed intensities of the high-excitation lines such as C IV λ1549 and He II λ1640 are much weaker than predicted (Barth et al. 1996, 1997; Nicholson et al. 1998; Maoz et al. 1998). [The case of M87 presented by Dopita et al. (1997) is irrelevant to the present discussion because those observations explicitly avoided the nucleus of the galaxy.] The data, however, cannot rule out contributions from slower shocks (v < ∼ 150 km s), although the viability of shock ionization in luminous AGNs has been criticized on energetic grounds by Laor (1998). Clues from the X-rays Compact soft X-ray emission on the scale of the ROSAT HRI camera (∼5) has now been detected in a handful of LINERs (e.g., Worral and Birkinshaw 1994; Koratkar et al. 1995; Fabbiano and Juda 1997), although no statistical conclusions can yet be drawn based on the scant data available. Most of the core sources have luminosities clustering near L(0.5–2 keV) ≈ 10–10 ergs s because of selection effects. The pointlike morphology of the ROSAT images certainly agrees with our expectation for an AGN source, but we must remember that the 5 point-spread function of the HRI subtends an uncomfortably large region (several hundred parsecs) at the typical distances of these objects. Images taken at much higher angular resolution and ideally at harder energies, such as would be possible with the ACIS camera on AXAF (see the concluding remarks at the end of the paper), are needed to put more stringent constraints on the nature of the X-ray emission. In the meantime, progress can be made by examining the ASCA hard X-ray spectra of LINERs whose X-ray structure is found to be compact on HRI images. These data, again, are scarce, and current constraints by necessity bias the sample in favor of the brightest targets. Nonetheless, when the observations are considered collectively (Serlemitsos et al. 1996; Terashima et al. 1997; Ptak 1998; Ptak et al. 1998; Iyomoto et al. 1998; Nicholson et al. 1998; see contributions by H. Awaki and Y. Terashima), the following trends appear. (1) The 2–10 keV continuum can generally be modeled as a single power-law function modified by cold absorption; the best-fitting photon index, Γ ≈ 1.7–1.8, agrees well with values normally measured in luminous AGNs. In some cases the fits require an additional soft thermal component with a temperature of kT ≈ 1 keV. (2) There is no evidence for significant amounts of cold material along the line of sight. Any measurable absorbing column in excess of the Galactic contribution usually does not exceed NH ≈ 10 21 cm. (3) Broad Fe Kα emission at 6.4 keV, a feature common to many luminous Seyfert 1 nuclei, is usually either absent or unusually weak. The composite LINER spectrum of Terashima et al. (1997) shows no detectable Fe Kα line to an equivalent-width limit of 140 eV. In the few cases where an iron line has been detected, the rest energy is ∼6.7 keV, consistent with ionized instead of neutral iron. And (4), these sources do not undergo rapid X-ray variability at the level expected from extrapolation of the variability behavior established for more luminous sources.