Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Optical Observations of Active Galactic Nuclei ALEXE! V. FTLIPPENKO University of California, Berkeley ABSTRACT I describe the most important observed characteristics of active galac- tic nuclei, concentrating on their optical spectra. Connections with the "standard model" of such objects are emphasized. Some of the new ob- se~vational developments that are leading to modifications of the standard model are outlined; these include spatial variations of the ionization pa- rameter in the broad-line region, the possible existence of a very dense component in the broad-line region, and the presence of an intermediate- density zone between the broad-line and narrow-line regions. Evidence for "hidden" Seyfert 1 nuclei and anisotropic ionizing radiation is also reviewed. The search for, and properties of, intrinsically weak Seyfert 1 nuclei are subsequently summarized, with special attention given to the very low-lummosibr object in the late-type dwarf galaxy NGC 4395. I conclude with a discussion of the possibility that low-level activity in some galactic nuclei might actually be produced by bursts of star foundation and their associated supernovae. BASIC OBSERVATIONS AND THE STANDARD MODEL The optical and ultraviolet (UV) spectra of QSOs and Seyfert 1 nuclei are dominated by strong, broad-permitted emission lines superposed on a featureless continuum (Osterbrock and Mathews 1986, and references therein). These lines typically have full-widths at half maximum (FWHM) of 3,000~,000 km s~i, and full-widths near zero intensity (E;WZI) of 10,00 20,000 km sol. In addition, one often sees narrower cores (FWHM ~ 500 91
92 AMERICAN AND SOVIET PERSPECTIVES km sol) to the permitted lines, as well as forbidden lines of comparable (relatively narrow) width. The forbidden lines include a wide range of ionization states, from [0 I] to [Fe VII] or higher. In Seyfert 2 nuclei, the broad permitted lines seem to be absent, although new data indicate that in at least some cases they are actually present at a very low leveL The featureless continuum of classical AGNs has a roughly power-law shape with index ~ ~ 1 - 2 (f`, or I'm) at red through near-infrared wave- lengths, but it generally becomes flatter (a ~ O-1) at higher frequencies (Neugebauer et at 1979~. In the UV region, the slope is often quite flat (a ~ 0~. There appears, in essence, to be a broad excess, or "big blue bump," over an extrapolation of the near-infrared power laws Starlight is an important contaminant in the spectra of low-luminosity Seyfert 1 nuclei (MaLkan and Filippenko 1983), but their nonstellar spectra are similar to those of QSOs. It has long been recognized that the flux of H,B and other permitted lines is, to first order, directly proportional to the flux density of the optical nonstellar continuum in large samples of AGNs (Tee 1980; Shuder 1981~. Thus, the lines are probably produced by gas photoionized by the UV extension of the featureless continuum emerging from the active nucleus. Photoionization calculations have reinforced our belief in this simple picture, since most of the emission-line intensity ratios can be reproduced (Netzer and Ferland 1984~. The models depend primarily on the ionization parameter, which is simply the ionizing photon number density divided by the nucleon (or electron) number density at the exposed face of a slab of gas. The broad permitted lines are formed by gas having ne ~ 109 - 10~° cm~3. This is deduced from the weakness or absence of broad forbidden [0 IlI] AJ4959, 5007 emission, and by the presence of broad semi-forbidden (intercombination) C III] A1909. Other, similar diagnostics also exist. Photoioni7~tion models imply that Te ~ 10, 000-20, 000 K Comparable temperatures are also found from various intensity ratios among the narrow lines, although the derived densities are much lower (ne ~ 103 - 105 cm~3~. with such low temperatures, the thermal widths of emission lines are only 1~20 lan sol. Bulk motions of gas are undoubtedly responsible for broadening the lines to the observed widths; other proposed mechanisms (e.g., electron scattering) have serious difficulties (Davidson and Netzer 1979~. The gas emitting the line radiation is distributed in clouds or filaments occupying only a small fraction (10-4 - 10-7) of the volume they encompass. This is deduced directly from the observed luminosities of the lines and their average distances from the central source, as determined by photoionization models. In low-luminosity AGNs, these clouds often have a covering factor of nearly unity and absorb soft X-rays traveling along our line of sight
HIGH-ENERGY ASTROPHYSICS 93 pelvis and Lawrence 1985), but in luminous QSOs the clouds only cover 1~10% of the sly. Supporting evidence includes the fact that a sharp drop at 912 &, the hydrogen Lyman limit, is rarely seen in the continua of high-redshift QSOs. T;he calculated radial distances for clouds in the broad-line region (BLR) are usually 0.01-1 pc, and in the narrow-line region (NLR) 1 1,000 pc. It is therefore not surprising that the BLR has never been spatial resolved in optical images or spectra, while the NLR is often resolved in nearby AGNs. This is also consistent with the absence of narrow-line variability in most AGNs (Peterson 1988), whereas vanabili~ of Me broad lines over time scales of weeks, months, or years is common. Photoioni~tion models and obsenations at X-ray energies (Halpern 1982) show that the column density of neutral hydrogen, 1VH, is ~ 1023 ~-2 and ~ 1022 cm~2 in the BER and NLR, respectively, almost independent of the intrinsic luminosity of the AGN. Accretion of matter onto a central, massive black hole has long been thought of as the ultimate source of energy emitted by QSOs and Seyfert 1 nuclei (Zel'dovich and Novikov 1964; Salpeter 1964), largely because of the absence of other viable mechanisms. Although spherical accretion has been considered, it is now generally believed that the gas forms an accretion disk around the black hole. Infalling matter almost certainly has specific angular momentum greater than that of the least stable orbit around a black hole, and accretion disks provide efficient release of energy (up to 0.37 mc2 for a Kerr black hole). Also, the well collimated jets spewing out of some AGNs suggest the presence of disks (Rees 19843. Direct observational evidence for disks is difficult to find, but many (not all!) researchers believe that the "big blue bump" represents thermal emission from an accretion disk (Shields 1978; Sun and MaLkan 1989, and references therein). COMPLICATIONS TO THE STANDARI) MODEL Despite its success, the simple picture described above is far from complete; many problems have been pointed out over the past decade. In particular, there remain serious difficulties in reproducing the observed emission-line spectrum of active galaxies (Ferland and Shields 1985; Netzer 1989~. Here ~ briefly discuss several complications that must be addressed in current and future calculations. Spatially Variable Ionization Parameter A single set of clouds, characterized by one value for the ionization parameter, probably cannot be invoked to explain all the intensity ratios of broad emission lines in AGNs (Collin-Souffrin et at 1988~. Specifically,
94 AMERICAN AND SOVIET PERSPECTIVES a large ionization parameter is needed lo generate highly ionized species such as C IV and O VI, whereas great quantities of X-rays are necessary to penetrate deep into thick clouds to produce the observed Fe II emission. As one possible solution, Netzer (1987) suggests that the ionizing UV radiation is emitted by a thin accretion disk whereas the X-ray continuum comes from a hot, spherical bulge at the center of the dish Clouds near the disk ems would have a much higher ionization parameter than those at large angles from the ems. An additional requirement is that the cloud covering factor must increase with increasing angle away from the disk ens. This is reasonable if the clouds have considerable angular momentum. It may also be the case that extra heating sources, other than the normal photoionizing continuum, are necessary for a good fit to the data (e.g., Joly 198 7~. Ultra-High and Intermediate Densities Another, equally serious problem is that the broad emission lines are observed to vary rapidly in response to changes in the luminosity of the ionizing continuum; see Peterson (1988) for a thorough review. Based on the sizes of the BLR calculated from the ionization parameter (inferred from spectra), time delays of a month to a year were expected in many Seyfert 1 gal~es. The measured delays, on the other hand, can be as short as a week to a month. This means that the BLR is typically a factor of three to ten smaller than had previously been thought. In the simplest models, the gas density must be 1~100 times higher to prevent the calculated values of the ionization parameter to deviate significantly from those deduced with spectra. Such high densities, however, produce over problems. The C IIT] A1909 line, for example, should be collisionally suppressed (relative to permitted lines) at the newb derived densities of no ~ TO]}-10~2 cnn~3e Thus, it probably arises from gas of lower densitr (~ 109 cm-3), yet it often has nearly the same profile as the permitted lines. There is also evidence for multiple zones in the gas primarily responsi- ble for the forbidden lines in AGNs. Many studies conducted over the past few years have shown that the concept of distinct, well-separated NLRs and BERs Is often an oversimplification. Rather, it appears that there exists an "intermediate zone," in which ne ~ 106 - 108 cm~3 and velocities of clouds are between those in the NLR and BLR. The evidence is the strong correlation, found in many objects, between line width and critical density for collisional de-excitation (Filippenko 1985; Whittle 1985; De Rober- tis and Osterbrock 19863; see Figure 1. Forbidden lines having ne~cnt) ~ 106 - 107 cm~3 are quite broad in some objects. Moreover, detailed analysis of the HE spectral region in Seyfert 1 galaxies indicates that broad
HIGH-ENERGY ASTROPHYSICS (b) - , 0, E A ~ . I ID o 95 Pictor A Widths of forbidden lines o - Slope ~ 0.103 ~ 0.015 r ~ 0.91 ~ 257. . I I 3 ~s 6 7 8 Log ne(crit) (cm~3) FIGURE 1 The Spyfert 1 galaxy Pictor A shows a strong Correlation between the F\VHM and the critical density of emission lines ~lippenko 19853. A very high correlation coefficient of 0.91 is found. The data were obtained with the 2.5 m du Pant reflector at Las Campanas Obsenratorye [O III1 emission may exist at a low intensity level, and is probably produced by clouds having ne ~ 107 - 108 cm~3 (Crenshaw and Peterson 1986, and references therein). The simple two-zone models are clearly inadequate, except for first-order approximations to the emission-line intensity ratios. Anisotropic Ionizing Radiation There is now considerable evidence that in some AGNs, ionizing radiation might be preferentially beamed along certain directions. The radio galaxy PKS 2152~9, for example, exhibits a high-excitation knot of emission at a projected distance of 8 kpc from its nucleus (headhunter et aL 1987~. without the presence of beamed radiation, it is difficult to reproduce this knot's isolated nature and its very high ionization parameter. The polarized blue continuum detected in this lmot is probably caused by scattering of the beamed radiation (di Serego Alighieri et at 19~. Powerful evidence for a disk-l~e (rather than spherically symmetnc)
96 AMERICAN AND SOVIET PERSPECTIVES geometry is found in the study of NGC 1068 done by Antonucm and Miller (1985~. Optical spectra of NGC 1068 show it to be a bright type 2 Seyfert galaxy, with permitted and forbidden lines having similar widths (FWHM ~ 1100 En sob. A spectrum of the poled flux alone, however, is virtually indistinguishable from normal (total flop spectra of type 1 Seyferts and QSOs! Broad permitted lines of hydrogen and Fe II are visible, and there is a very blue, featureless continuum. Furthermore, Miller (1989) has observed broad Balmer lines scattered from an ofI-nuclear H II region in NGC 1068. The obvious interpretation of these data is that NGC 1068 actually harbors a type 1 Seyfert nucleus that is located inside an optically and geometrically thick torus. Photons from the continuum source and the BLR are scattered into our line of sight, and therefore polarized, by free electrons above and below the disk. The NLR is sufficiently extended Hat it is not obscured by the disk. It is quite possible that the thick torus feeds into an accretion disk close to the black hole. Pogge's (1988) discovery of a high-ioniz~tion emission-line "cone" in NGC 1068 supports the hypothesis advanced by Antonucci and Miller (1985~. Moreover, the luminosity of the ionizing continuum from the nucleus of NGC 106S, deduced from the total flux spectrum and the assumption of spherical symmetry, cannot account for the luminosity of forbidden lines in the NLR. No problem exists, on the other hand, if the NLR clouds have a much more direct view of the compact nucleus. Similar conclusions regarding hidden BLRs (Miller and Goodrich 1990), high- ionization emission-line cones (Pogge 1989), and the energetics of the NLR Abelson et at 1988) have been made for many other Seyfert 2 galaxies; thus, NGC 1068 is not an isolated example of this phenomenon. In addition, radio observations (Antonucci 1984) show that the optical polarization position angle of Seyfert 2 galaxies is almost always perpendicular to the radio axis, as expected if they contain an opaque torus which we view i rom the side. LOW-LUMINOSITY ShYFERT 1 NUCLEI Searching for Broad Ha Emission The nonstellar continuum of QSOs Epically has an absolute blue magnitude (MB; ~ ~ 4500 A' between-30 and-23 (i.e., LB ~ 1011- 10~4 Log, while that of Markarian Seyfert 1 nuclei lies in the range-23 to -18 (LB ~ 109 - 1011 Log. The well-known continuity in many observed characteristics of these objects "Ives strong evidence for the idea that their physical properties are also similar. Observations of low-redshift QSOs suggest that all QSOs occur in galaxies; hence, we certainly expect there to be at least a few local Seyfert 1 nuclei with very low luminosity, if
HIGH-ENERGY ASTROPHYSICS 97 some of these objects result from the fading of old QSOs as their fuel suppler steadier dwindles. Continuity in the observed properties of QSOs and Seyfert 1 nuclei further suggests the existence of such objects, but there might be a lower limit to the activity. If so, it could be of fundamental importance, like the lower mass limit (~ 0.08 Me,) of main-sequence stars. The nucleus of the nearby, frequently studied spiral galaxy NGC 4051 was for many years the least luminous known Seyfert 1 (MB ~-16; V6ron 1979~. Owing to contamination by starlight, it is quite difficult to detect the nonstellar blue continuum in very low-luminosity Seyfert 1 nuclei. Broad Ha emission, on the other hand, is easier to discern. For convenience, we will adopt the relation MB ~-2.5 log L(Hc~) ~ 84.7 (where log ~ _ logic z), as empirically derived by Weedman (1985; Ho = 75 lan s~i Mpc-~) for the nuclei of Markarian galaxies having broad Ho emission with luminosity L(Ha) ergs s-l. Low-luminosibr Seyfert 1 nuclei, arbitrarily defined to be those win MB ~ - 18 may, have a corresponding broad He luminosity of L(Hc~) ~ 104~ ergs sol. The standard interpretation for such objects (e.g., Filippenko 1988) is that they have rather small central black holes (M < 106 M(S,Y, or that they are Secreting material at very sub-Eddington rates. Many low-luminosibr Seyfert 1 nuclei have now been found in nearby galaxies (Stauffer 1982; Keel 1983; Filippenko and Sargent 1985~. An excellent example is M81 (Figure 2), whose distance is about 3.3 Mpc; weak, broad Ha emission was discovered by Peimbert and ~rres-Peimbert (1981) and confirmed by Shuder and Osterbrock (1981~. An accurate measurement of the broad Ha luminosity, L(Hcr) ~ 1.2 x 1039 ergs s~t (Filippenko and Sargent 1988), together with Weedman's (1985) relation given above, yields MB ~ -13.0 mag for the blue nonstellar continuum, which has not yet been detected. The nucleus of this galaxy is also a strong, variable X-ray source, as well as a bright, compact, variable, flat- spectrum radio source. There is almost no doubt that its activity is similar to that in QSOs; it could reasonably be called a "microquasar" (Elvis 1984~. Of relevance to the discussion above, its spectrum (Figure 2) shows that forbidden lines associated with high critical density (e.g., [O I] AA6300, 6364; ne~crit) = 1.4 x 106 cm~3) are considerably broader than those with low critical density (e.g., IS II] AN6716, 6731; ne~crit) ~ 2 x 103 cm~3~. In an attempt to quantifier the luminosity function of intrinsically faint Seyfert 1 nuclei W. L. W. Sargent (Caltech) and ~ are searching for broad Ho emission in optical spectra of the nuclei of the 500 brightest galaxies in the northern sky; see Filippenko and Sargent (1985) for initial results. NGC 4639 is an excellent example of a bright, nearby Shapley- Ames galaxy whose Sequent 1 characteristics had previously gone unnoticed. The spectrum at blue wavelengths (Figure 3) is comparable to those of
98 AMERICAN AND SOVIET PERSPECTIVES , ~ r-l T-~-, I 1 1 1, I I 1 1 ' ' ' ' 1 ' ' 1 ~ ! ' 00 so o 6200 6300 1 M8 1 minus NGC 4339 1986 Feb. 25 UT 1`' x 4, P.\ - 180 (a) ~ A 1 ~ _\, -I ~ 1 ~ 1 ; - \1 [o I] Am Jig (c)=(a)-(b) 1 Ha 1 ~ [N II] [N II] [S II] ~ -~ _ _ = wow ~ ,,,,1,,,,1,,,,1,,,,1,,,,1,,,,1 6400 6s00 6600 Rest wavelength (A) 6700 6800 FIGURE 2 Red spectrum of M81 (Filippenko and Sargent 1988), obtained with the 5 m Hale reflector at Palomar Observatory. A spectrum of the absorption-line template galaxy NGC 4339 is subtracted Tom M81 in (c). M81 has a low-luminosi~ Spyfert 1 nucleus; broad Hcz is clearly visible. Note that the widths of the forbidden lines span a wide range. normal, inactive galaxies devoid of strong emission lines. The red spectrum (Figure 43, by contrast, exhibits prominent, broad He emission with FW~ ~ 3, 000 - 4, 000 km s~: and FWZI ~ 8, 000 - 9, 000 lan sol. A Malone of old redshift surveys of galaxies did not include the red spectral region, so it is possible that quite a few objects of this type have been overlooked. Furthermore, in most galaxies the broad Hc' emission line is very weak, requiring careful deconvolution of the He + IN II] AA6548, 6583 blend for detection and measurement. A Seyfert 1 Nucleus in a Dwarf Sd Galaxy It seems reasonable that most low-luminosity Seyfert 1 nuclei have been found in early-Wpe spiral galaxies, given the drstn~ution of classical Seyferts among different Hubble types. Moreover, early-lope spirals with well-developed bulges are thought to have much deeper potential wells than late-type spirals, perhaps promoting the formation of supermassive black holes and the retention of gas. Indeed, with the possible exception
HIGH-ENERGY ASTROPHYSICS 4 99 1 ' ' ' ' 1 ' ' ' ' 1 ' ' 1 1 1 i,, 1 NGC 4639 1985 June 19 UT [0 II] [O III] o , , , , , , , 1 , , , , 1 , , , , 1 , , 3500 4000 4500 5000 Wavelength (it) FIGURE 3 Blue spectrum of NGC 4639, obtained with the 3 m Shane reflector et Lick Observatory. The galaxy redshift has been removed. An absence of strong emission lines suggests that this is not a Se~rfert galaxy. Of G1200 2038 (Kunth et al 1987), until recently not a single Seyfert 1 nucleus had been found in Sd, irregular, or dwarf galaxies. We were very surprised, therefore, to discover that NGC 4395, a nearby (d ~ 2.6 Mpc) low-luminositr (MB ~-16.4 may) Sd galaxy, harbors a very faint Seyfert 1 nucleus (Filippenko and Sargent 1989~. A superb photograph of this object is shown in panels 10 and 56 of the Atlas of Galaxies compiled by Sandage and Bedke (1988~. The faint star-like nucleus emits a narrow- line spectrum similar to that of a type 2 Seyfert, with emission lines of very high ionization (up to [Ne V1 and [Fe X1) superposed on a featureless condouum. Photoion~tion by a reasonably hard continuum is almost certainly responsible for Heir relative strengths, but tO I] A6300 and IS lI] A)6716, 6731 are unusually intense with respect to IN Il] A)6548, 6583, as illustrated in Figure 5. The narrow lines have FWHM < 60 km s~i smaller than in any other Mown Seyfert nucleus. Weak, broad components are clearly visible in the per~itted-line pro- files, but not in the forbidden lines. The FWZI of Ha is 6,000 7,000 km
100 . =- - - ~ - - ~ - ~ - oD c -· ~ - ~> - ~ c, +. - NGC 4639 1985 February 24 UT [0 1] AMERICAN AND SOY7ET PERSPECTIVES H`x [N II] J fit [N lI] 1 [S 11] Ah . . . . 6300 6400 6500 6600 6700 Wavelength (A) FIGURE 4 Red spectrum of NGC 4639 (Filippenko and Sargent 1986), obtained with the 5 m Hale redector at Palomar Observatory. The galaxy redshift has been removed. Atmospheric O2 produces the absorption line near 6260 ~ The obvious presence of broad Hat emission indicates that the nucleus is a type 1 Se~rfert. s-1 in the very high-quality spectrum (Figure 5), although the FWHM is only about 60~700 km sol. The luminosity of the broad Ha line is ~ 1.2 x 1038 ergs s~i, about one tenth that in Met, and MB ~-9.8 mag is measured from the spectrum. Hence, NGC 4395 has the intrinsically weakest known Seyfert 1 nucleus, with a blue luminosity no greater than that of the most luminous supergiant stars! If NGC 4395 were much farther away, so that the spectroscopic entrance aperture included a large amount of extranuclear light, its spectrum would have been dominated by emission lines from H II regions and continuum from OB associations, making the Seyfert activity difficult to detect. At optical wavelengths the featureless continuum of NGC 4395 can be described by a power law with spectral index 1.5, comparable to that of other Seyfert 1 nuclei. Moreover, the equivalent width of the broad Ho emission, 270 A, is also Apical of type 1 Seyferts. If we assume that the bolometnc luminosity of the nucleus is roughly 1000 times that of the broad H,0 emission, as for other Seyfert 1 nuclei (Weedman 1976), we find that LAO! ~ 1.5 x 104° ergs sol. This could be emitted by a black hole having
HIGH-ENERGY ASTROPHYSICS 50 40 30 20 83 10 lo_ 2.0 1.0 [o I] (b) [o I] [/m] [Fe X] _ ~ Hi/ [ArV] 0.0 I I I I I I I I t I I 6200 6300 101 J I I I I I I I I I I I I I I I I l I I I I I I i I I I I I I Ill - (a) NGC 4395: Nucleus Ha [N II] 1 1 ~ 1,,,, 1, ., . [N lo] at_ [S B] I ~ ~ I 6400 6500 6600 6700 6800 Observed Wavelength (~) FIGURE 5 Red spectrum of the nucleus of NGC 4395 ~lippenko and Sargent 19&9), obtained with the 5 m Hale reflector at Palomar Observatory. The gala~s heliocentric radial velocity is 320 An s-1. The luminosity of the broad Ha line is the lowest measured in any Spyfert 1 nucleus a mass of only ~ 100 Me if it were accreting at the Eddington limit! Such a black hole could be produced in only 200 million years if it started out with one solar mass and always Secreted at the Eddington limit Adopting the formalism of Wandel and Yahil (1985), on the other hand, we find that the current dynamical mass is about 104 M ,, if gravity accelerates the emission-line clouds to typical observed speeds of ~ 500 lan sol. Thus, it is possible that L ~ 0.01 LE&~. Given its low luminosity, it is interesting to consider whether the nucleus of NGC 4395 is actually a single very massive star (M ~ 200 Mom. Its spectrum, of course, is unlike that of known objects such as main- sequence O-type stars or even Wolf-Rayet stars, but this is nonetheless an interesting idea; we might never have observed such a star, because they form rarely and have very short lives. If the broad emission lines are caused by an outflowing wind, then the mass loss rate can be calculated under the assumption of steady-state conditions. Photoioni~ntion models give a distance of ~ 10~5 an between the star and the emission-line clouds, and the absence of broad forbidden lines suggests that ne ~ 109 am~3. Adopting an outflow velocity of SOO km s~i, the equation of cont~nui~r
102 AMERICAN AND SOVIET PERSPECTIVES yields M ~ 0.02 M<3 yr~l. At this rate, the 200 Me star would only last about 104 years! It Is very important to obtain additional observations, at all wave- lengths, of low-luminosi~ Seyfert 1 nuclei such as that in NGC 4395. They offer counterexamples to the conventional wisdom that activity is restricted to early-type galaxies with large bulges. If a massive black hole is ultimately responsible for the observed phenomena in the nucleus of NGC 4395, we must explain how it was able to form in this particular object, but not in a majority of other extended, low-mass galaxies. On the other hand, perhaps most dwarf and very late-type galaxies can, in principle, be active, but are usually quiescent because the fueling mechanism does not operate efficiently. Basically, we do not yet know what conditions are necessary for a galaxy to have an active nucleus. Warmers: A Different Point of View The narrow emission lines in classical AGNs, even those without broad emission lines, are almost certainly the result of photoionization of gas by a relatively flat UV and X-ray continuum. The intensities of the lines generally satistr, very roughly, [O III] A5007/H,B ~ 10 and rN III A6583/Hc~ ~ 1, and high-ionization lines such as [He V1 A3426 and He II A4686 are present (e.g., Osterbrock 1977; Koski 1978). [O I] )6300 is considerably stronger than in H II regions, whose O and B stars produce Stromgren spheres with thin, well-defined boundaries between the zones of neutral and ionized hydrogen. Most researchers believe that the continuum arises from the vicinity of a massive black hole, and is of nonstellar origin, as described above. There is, however, a very different possibility, at least for AGNs of relatively low luminosity. In a thought-provoking paper, Terlevich and Melnick (1985) showed that ionizing radiation from evolved, very massive (M > 60 M<~) stars in metal-rich H II regions can produce emission-line intensity ratios typical of Seyfert 2 nuclei. These Wolf-Rayet stars, which they call "Warm- ers" (extreme WC or WO stars), have effective surface temperatures of (1 - 2) x 105 K; plenty of high-energy ionizing photons are therefore emit- ted. A key point is that stellar mass loss increases with increasing metal abundance in massive stars, so that very massive stars formed in high- metallici~ environments can end their lives as bare He-C-D cores with surface temperatures comparable to those of the hottest lmown nuclei of planetary nebulae but with considerably higher luminosities. Calculations show that after 3 million years, the ionizing continuum of a cluster created during a large burst of star formation (initial mass function slope = 3.0; LRqueux 1979) is nearly a power law, fit or u-i 5, with a cutoff at ~ 20 Ryd. Thus, the resulting emission-line spectrum from surrounding
HIGH-ENERGY ASTROPHYSICS 103 clouds of gas having roughly solar composition closely resembles that of Seyfert 2 galaxies. Moreover, the weak featureless continuum observed in some type 2 Seyferts may be the reddened spectrum of the ionizing cluster. As the cluster ages, the spectrum evolves into one characterized by low-ioni~tion emission lines, similar to those seen in Heckman's (1980) LINERs (low-ionmation nuclear emission-line regions). Motivated by the success of their earlier work, Terlevich et al (1987) went on to postulate that typical Seyfert 1 nuclei may also ultimately be produced by starburst activity. In their scenano, the broad emission lines are produced by supernovae (SNe), and by the interaction of SNe with a dense interstellar medium. Although there are a number of potential difficulties with this hypoth- esis, here I speculate that NGC 4395 may represent an example of a Seyfert nucleus powered by a burst of star formation; in this case, the narrow-line spectrum is produced by a cluster of Warmers. Detailed photoionization models must be computed in order to see whether the observed emission- line spectrum can be reproduced under conditions appropriate for the Warmers scenario. It is not clear, for example, whether the low ~ Ill strength reflects peculiar abundances or unusual excitation. The observed velocity of the broad-line gas Is unlikely to be produced by gravitational effects; a sufficiently compact and massive star cluster would also be very short lived, due to stellar collisions. If, on the other hand, the broad permitted lines are produced by SNe, as suggested by Terlevich et aL (1987), only one very old SN is needed to explain their low luminosity. Indeed, if the total broad-line luminosity is 1039 ergs s~: (a reasonable value, given the observed strengths of HE and H,B), the entire visible energy emitted by a typical SN (~ 1049 ergs) could power the nucleus for 300 years at a constant rate! Adding even a small fraction of the kinetic energy (105: ergs) greasy increases this estimate. Near maximum, a SN is far more luminous than the nucleus of NGC 4395; the SN in NGC 4395 must therefore be very old. Goodrich et at (1989) detected weak, broad Ha emission nearer three decades after the discovery of the more distant (d ~ 8 Mpc) SN 1961V, so a late-tune detection of a SN in NGC 4395 is not unreasonable. Note, however, that Goodrich et aL (1989) believe that SN 1961V was actually not a SN, but rather an exaggerated y-Carinae-like outburst This may be consistent with the single-star hypothesis of NGC 4395 briefly discussed in the preceding section. It might be argued that the spectral characteristics of Seyfert 1 nuclei cannot possibly be produced by SNe, since the spectra of SNe are not generally thought to resemble those of Seyfert 1 nuclei. Figure 6, however, shows Mat there are some exceptions; the spectrum of SN 1987F was quite similar to that of 3C 48 at optical wavelengths (Filippenko 1989~. There are, of course, some important differences if the spectra are examined carefully. The broad line near 5900 3, for example, is attributed to Na I
104 14 15 16 ._ 17 18 AMERICAN AND SOVIET PERSPECTIVES 1 ' ' I I ~I ' 1 ~ ~ 1 1 I ~1 ' ' ' ~ 1 1 ~i, I, 1 _ (a) 3C 48, 10 - 3 - 88 Ha to m] H.' Her Fe II _ Ha it' Fe II He ~ l~/iI ~f - ~ n . it _ ~ s I ~f -' Phi '-"--WOW ~ AIL _ ~ 1 1,,, 1 4000 4500 19 1 1 I I I 1 1 1 1 ~1 1 I ', . (b) SN 1987F, 12 - 26 - 87 I,,,, 1,,,, 1 1 5000 5500 6000 6500 Rest Wavelength (at) FIGURE 6 Spectra of (a) the quasar 3C 48, and (by SN 1987F in NGC 4615 roughly 9 months after discovery. The data were obtained with the Shane 3 m reflector at lick Observatory (F-llippenko 1989~. AB magnitude = - 2.5 log fit -48.6, where the unites of A, are ergs s~ 1 cm-2 Hz- ~ D in SN 1987F and to He I in 3C 48. Moreover, the centroid of the broad Ha line in SN 1987F and in other, similar SNe is blueshifted relative to the narrow component, whereas the broad Ha in Seyfert 1 nuclei tends to be redshifted at least as often as blueshifted. Finally, the continuum of SN 1987F begins to drop rapidly at near-UV wavelengths, whereas that of 3C 48 remains quite strong. Nevertheless, had SN 1987F occurred in the nucleus of a normal galaxy, a low-resolution spectrum undoubtedly would have led to a Seyfert 1 classification for the galaxy, especially if some Warmers were present to produce high-ioni~tion narrow lines and a blue continuum. If we further assume that physical conditions (e.g., a dense interstellar medium, as may have been He case for SN 1987F) in certain galactic nuclei somehow enhance the production of SNe Lee SN 1987F, then the spectrum could remain Seyfert-l~e for long periods of time, and exhibit variability as well.
HIGH-ENER~ ~TROP~ICS 105 Thus, it is possible that some catalogued type 1 Seyfert galaxies, especially those of low luminosity, owe their observed characteristics to objects such as SN 1987F; we simply do not know enough about the evolution of SNe and their remnants in dusty, dense environments. I do not want to use this spectral resemblance to argue for the above scenario as a general explanation for Seyfert 1 galaxies, and certainly not for luminous radio-loud QSOs. However, the spectral similarity of SN l987F and 3C 48 at optical wavelengths can be used to teach us a great deal about the physical processes in both types of objects. Many more comparisons must be made, over as large a range of wavelengths as possible, in order to obtain a more thorough physical understanding of AGNs. I am certain we will encounter numerous surprises along the way. ACKNOWLEDGMENTS Some of the observations reported here were taken at Palomar Obser- vatory, where I was a Guest Investigator collaborating with W. L. W. Sargent. Data were also obtained at Lick Observatory, which receives partial fund- ing from NSF Core Block grant AST 8614510, and from Ins Campanas Observatory, which is owned and operated by the Carnegie Institution of Washington. My research on AGNs and SNe has been supported by the California Space Institute, most recently through grant CS 41 - , and by NSF grants AST 8957063 (Presidential Young Investigator Award) and AST-9003~9. I thank the Once of International Affairs of the National Research Council, especially Kathleen [livers, as well as George Clark, Walter Lewin, and Rashid Sllnyaev, for organizing the successful NAS- ASUSSR Workshop on High Energy Astrophysics. I am also very grateful for the warm hospitality of our hosts in Moscow and Soviet Georgia. REFERENCES Antonucci, RRJ. 1984. Optical spectropolarimetIy of radio galaxies. Astrophysical Journal 278: 499-520. Antonucci, R.RJ., and J.S. Miller. 1985. Spectropolarimet~y and the nature of NGC 1068. Astrophysical Journal 297: 621~32. Collin-Souffrin, S., J.E. Dyson, J.C. McDowell, and JJ. Perry. 1988. The environment of active galactic nuclei-I. A two-component broad emission line model. Monthly Notices of the Royal Astronomical Society 232: 539-550. Crenshaw, D.M., and B.M. Peterson. 1986. Evidence for a low~ensity component in the broad-line region of Spyfert 1 galaxies. Publications of the Astronomical Society of the Pacific 98: 185-191. Davidson, K, and H. Netzer. 1979. The emission lines in quasars and similar objects. Reviews of Modern Physics 51: 715-766. De Robertis, MM., and D.E. Osterbroc~ 1986. An analysis of the narrow line profiles in Seyfert 2 galaxies Astrophysical Journal 301: 727-741.
106 AMERICAN AND SOVIET PERSPECTIVES di Serego Alighieri, S., L Binette, I:J.-L Couwoisier, R^.E. FosbuIy, and AN. Tadhunter. 1988. A blue, polarized continuum source near radio galaxy PKS 2152~9. Nature 334: 591-593. Elvis, M. 1984. "Microquasars" and the X-ray background. Advances in Space Research 3~10-12~: 207-209. Elvis, M., and A Lawrence. 1985. X-ray spectra of active galaxies and quasars. Pages 289-331. In: Miller, J.S. (ed.~. Astrophysics of Active Galaxies and Quasi-Stellar Objects. University Science Book$ Mill Valley, CA Berland, GJ., and GN Shields. 1985. The theory of emission-line regions in active galactic nuclei. Pages 157-184. In: Miller, J.S. (ed.~. Astrophysics of Active Galaxies and Quasi-Stellar Objects. University Science Books, Mill Valley, CA. F~lippenko, NV. 1985. New evidence for ohotoionization as the dominant excitation r- mechanism in LINERs. Astrophysical Journal 28~. 475~9. F~lippenko, ~V. 1988. Indirect evidence for massive black holes in nearby galactic nuclei. Pages 104-119. In: Kafatos, M. (add. Supermassive Black Holes. Cambridge University Press, Cambridge. F~lippenko, ~V. 1989. The "Seyfert 1" optical spectra of the type II supernovae 1987F and 19881. Astronomical Journal 97: 72~734. Filippenko, ~V., and W.LW. Sargent. 1985. A search for "dwarf" Seyfert 1 nuclei: the initial data and results. Astrophysical Journal Supplement Series 57: 503-522. F~lippenko, NV., and W.LU: Sargent. 1986. Ibe properties of "dwarf" Seyfert nuclei in nearby galaxies Pages 2145. In: Giunan, G., F. Mardirossian, M. Mezzetti, and M. Ramella (eds.~. Structure and Evolution of Active Galactic Nuclei. Infidel, Dordrecht. F~lippenko, ~V., and W.LW. Sargent. 1988. A detailed study of the emission lines in the Seyfert 1 nucleus of M81. Astrophysical Journal 324: 134153. F~lippenko, NV., and W.LW. Sargent. 1989. Discovery of an extremely low luminosity Seyfert 1 nucleus in the dwarf galaxy NGC 4395. Astrophysical Journal (Letters3 342: L11-L14. Goodrich, RW., G.S. Stnngfellow, G.D. Penrod, and A.V. F~lippenko. 1989. SN 1961V: an intragalactic eta Carinae analog? Astrophysical Journal 342: 908-916. Halpern, J.P. 1982. X-ray spectra of active galactic nuclei. Ph.D. Thesis, Harvard University. University Microfilms International, Ann Arbor, MI. Grecian, EM. 1980. An optical and radio survey of the nuclei of bright galaxies. Astronomy and Astrophysics 8?: 15~164. John, M. 1987. Formation of low ionization lines in active galactic nuclei. Astronomy and Astrophysics 184: 33~2. Keel, W.C 1983. Spectroscopic evidence for activity in the nuclei of normal spiral galaxies. Astrophysical Journal 269: 466~86. Koski, N1: 197& Spectrophotometry of Seyfert 2 galaxies and narrow-line radio galaxies. Astrophysical Journal 223: 5~73. Kunth, D., W.L~ Sargent, and G.D. Bothun. 1987. A dwarf galaxy with Seyfert characteristic. Astronomical Journal 93: 29-3Z Lequeux, J. 1979. A quantitative studier of the upper HR diagram and a new determination of the local initial mass function. Astronomy and Astrophysics 80: 3541. Malkan, M^, and NV. F~lippenko. 1983. Ibe stellar and nonstellar continua of Seyfert galaxies: nonthermal emission in the near-inffared. Astrophysical Journal Z75: 477- 492 Miller, J.S. 1989. Observational studies of the structure of active galactic nuclei and QSOs. Pages 27~283. In: Osterbrock, DE., and J.S. Miller teddy. IAU Symposium 134: Active Galactic Nuclei. Kluwer, Dordrecht. Miller, J.S., and RW. Goodrich. 1990. Spectropolanmetry of high-polarization Seyfert 2 galaxies and uniSed Seyfert theories Astrophysical Journal 355:456467. Netzer, H. 1987. Quasar discs II. A composite model for the broad-line region. Months Notices of the Royal Astronomical Society 225: 55-72. Netzer, H. 1989. Structure and nature of AGNs. Pages 69 84. In: Miller, J.S., and D.E. Osterbrock (eds.~. IAU Symposium 134: Active Galactic Nuclei. Kluwer, Dordrecht. Netzer, H., and GJ. Ferland. 1984. Some comments on models of photoion~zed nebulae. Publications of the Astronomical Society of the Pacific 96: 593-597.
HIGH-ENERGY ASTROPHYSICS 107 Neugebauer, G., J.B. Oke, E.E. Beclclin, and K. Matthews. 1979. Absolute spectral energy distributions of quasi-stellar objects from 0.3 to 10 microns. Astrophysical Journal 230: 79-94. Osterbrock, D.E. 1977. Spectrophotomet~y of Seyfert 1 galaxies Astrophysical Journal 215: 73~745. Osterbrock, D.E., and FIG. Mathews. 1986. Emission-line regions of active galaxies and QSOs. Annual Reviews of Astronomy and Astrophysics 24: 171-203. Peimbert, M., and S. Torres-PeimberL 1981. Physical conditions in the nucleus of M81. Astrophysical Journal 245: 845~56. Peterson, B.M. 1988. Emission-line variability in Seyfert galaxies. Publications of the Astronomical Society of the Pacific 100: 18-36. Pogge, R.W. 1988. An extended ionizing radiation cone from the nucleus of the Seyfert 2 galaxy NGC 1068. Astrophysical Journal 328: 519-522. Pogge, Rev 1989. The arcumnuclear environment of nearby, noninteracting Seyfert galaxies. Astrophysical Journal 345: 730-751. Rees, MJ. 1984. Black hole models for active galactic nuclei. Annual Reviews of Astronomy and Astrophysics 22: 471-506. Salpeter, E.E. 1964. Accretion of interstellar matter by massive objects. Astrophysical Journal 140: 796~0. Sandage, A., and J. Bedke. 1988. Atlas of Galaxies Useful for Measuring the Cosmological Distance Scale. Carnegie Institution of Washington, Washington D.C Shields, GO 19 78. Thermal continuum from accretion disks in quasam. Nature 272: 706-708. Shuder, J.M. 1981. Emission-lin~continuum correlations in active galactic nuclei. Astro- physical Journal 244: 12-18. Shuder, J.M., and D.E. Osterbroclc 1981. Empirical results from a study of active galactic nuclei. Astrophysical Journal 25~. 55~5. Stauffer, J.R. 1982 A nuclear spectroscopic survey of disk galaxies. II. Galaxies with emission lines not excited by stellar photoionization. Astrophysical Journal 26Z 6~80. Sun, W.-H., and MN Malkan. 1989. Fitting improved accretion disk models to the multiwavelength continua of quasam and active galactic nuclei. Astrophysical Journal 346: 68-100. Tadhunter, ON., ROE. Fosbury, L Binette, IJ. Danziger, and A. Robinson. 1987. Detached nuclear-like activity in the radio galaxy PKS 2152~9. Nature 325: 504-507. Terlevich, R., and J. Meluic~ 1985. Ub~mer~ the missing link between starburst and Spyfert galaxies. Monthly Notices of the Royal Astronomical Society 213: 841856. Terlevich, R. J. Melnick, and M. Moles. 1987. Starburst models for AGNs. Pages 499-519. In: Khachikian, E.Ye., KJ. Fricke, and J. Mel nick (eds.~. IAU Symposium 121 : Observational Evidence of Activity in Galaxies. Reidel, Dordrecht. Veron, P. 1979. The luminosity function of Seyfert 1 galaxy nuclei and BL Lac objects, and the X-ray background. Astronomy and Astrophysics 78: 4~52. Handel, A., and A Yahil. 1985. Universal mass-luminosity relation for q~am and active galactic nuclei? Astrophysical Journal (Letters) 295: L1-LA. Weedman, D.W. 1976. Luminosities of Spyfert galaxies and QSOs Astrophysical Journal 2~: 30-36. Weedman, D.W. 1985. Evolution of active galactic nuclei and quasars Pages 497-519. In: Miller, J.S. (ed.~. Astrophysics of Active Galaxies and Quasi-Stellar Objects. University Science Books, Mill Valley, CA. Whittle, M. 1985. The narrow line region of active galaxies III. Profile compansons. Month Notices of the Royal Astronomical Society 216: 817 855. Wilson, US., MJ. Uhard, and CA. Habib. 1988. High-resolution emission-line imaging of Seyfert galaxies. II. Evidence for anisotropic ionizing radiation. Astrophysical Journal 334: 121-129. Bee, H.KC 1980. Optical continuum and emission-line luminosity of active galactic nuclei and quasars. Astrophysical Journal 241: 894-90Z Zel'dovich, Ya.B., and I.D. Novikov. 1964. Relativistic astrophysics. I. Uspekhi Fizicheskikh Nauk 84: 377~17. (Soviet Physics Uspekhi 7: 763-7~.)