Inward Bound: The Search for Supermassive Black Holes in Galaxy Nuclei

Annu. Rev. Astron. Astrophys. 1995. 33: 581-624
Copyright © 1995 by Annual Reviews. All rights reserved

8. CONCLUSION

Have we discovered BHs in galaxy nuclei? Rigorously, we have not. Dynamical proof requires measurement of relativistic velocities near the Schwarzschild radius, r_S appeq 2M₈ AU appeq M₈ x 10^-5 pc. Even for M31, this is 3 x 10^-6 arcsec. HST spectroscopic resolution is only 0".1. What dynamical searches have contributed is the first solid evidence for massive dark objects (MDOs) in galaxy nuclei. In Section 2, we called this step 1 of the BH search. The dynamics also show that M_BH is in the range expected for quasar engines, and they tell us a little about MDO demographics. Are MDOs BHs? The arguments are inconclusive. Steps 2 and 3 - proof that MDOs are (or are not) BHs and the study of BH statistics - are barely begun.

If the BH picture is compelling, it is because of the combination of dynamical searches and AGN evidence. A review of AGNs is beyond the scope of this paper, but it is useful to recall the observations that point most directly to BHs. (a) Superluminal jets show that AGN engines are relativistically compact. (b) AGN engines remember jet ejection directions for a long time, suggesting that they are good gyroscopes. (c) Rapid time variability and very long baseline interferometry show that AGN engines are tiny. (d) The prodigious energy output of quasars requires that AGN engines be very efficient.

These are also the strongest arguments against the best alternative to the BH picture: that AGNs are powered by starbursts (e.g., Terlevich & Melnick 1985; Terlevich et al. 1987, 1993; Filippenko 1989, 1991, 1992b, 1993; Heckman 1991; Terlevich 1992; Tenorio-Tagle et al. 1992; Terlevich & Boyle 1993; meetings devoted to this subject include Filippenko 1992a; Beckman et al. 1993). It is difficult to imagine that a starburst can produce a superluminal jet or a jet that stays collimated over a Mpc. Something like a BH is required even if AGNs are partly powered by starbursts. In addition, starburst energy conversion efficiencies are a factor of ~ 10 lower than those of BHs, so the starburst model predicts MDO masses of gtapprox 10¹⁰ M for dead quasars. If such massive MDOs existed within 15 Mpc of us, we would not be struggling to find them. Other arguments against the extreme starburst hypothesis are given in Heckman (1994). But no observation conflicts with the idea that starbursts play a role in quasar formation (Sanders et al. 1988a,b), and it is possible that some low-level AGNs are powered entirely by starbursts (see the above references; Filippenko et al. 1993). Meanwhile, AGN observations and theory, and now the results of the dynamical searches, make a compelling picture in favor of the BH paradigm.

Still, we cannot help but note that this rosy picture is not developing as a correct paradigm should. The dynamical searches and AGN results have remained almost decoupled. The galaxies with stellar-dynamical evidence for BHs are spectacularly inactive. Table 1 lists nuclear radio continuum fluxes. M31 contains the lowest-luminosity radio AGN known (Crane et al. 1992, 1993a). Sgr A* and M32 are comparably weak. Because of its inactivity, NGC 3115 has been used as an emission-line-less spectral standard in searches for low-level AGNs (Filippenko & Sargent 1985; Ho et al. 1993). Among stellar-dynamical BH cases, only NGC 4594 is somewhat active. It contains a LINER (Heckman 1980), i.e., a dwarf Seyfert 1 nucleus (H alpha width at zero intensity ~ 5400 km s^-1: Filippenko & Sargent 1985, 1987). But NGC 4594 is an order of magnitude lower in radio luminosity than M87. Even M87 is a weak radio galaxy. In contrast, many AGNs that are no farther away show no dynamical evidence for BHs. To be sure, as Dressler has said, BHs are not easily discovered when we have a searchlight in our eyes. Also, there is every reason to expect dead AGNs all around us. But our confidence is not helped by the fact that galaxies with MDOs are so inactive while AGN BHs are so elusive.

This is why the M87 and NGC 4258 results (Section 5) are so welcome. M87 epitomizes the AGN paradigm. Now the detection of a nuclear gas disk perpendicular to the jet and with velocities implying a BH of just the right mass is the first direct contact between BH searches and the rest of the paradigm. It will be important to check whether the disk is in Keplerian rotation and hence whether the velocities measure mass. But the apparent detection of a Keplerian rotation curve in NGC 4258 already increases our confidence in gas-dynamical BH searches.

Which leads us to our ``top ten'' wish list:

1. We need to address any lingering doubts about the best BH cases. This requires at least one iteration in improved spatial resolution over the discovery observations plus dynamical models that explore the boundary between extreme models with constant M/L and ones that contain BHs.
2. Theory may help us to eliminate some models. Reducing M/L requires radial anisotropy; such models may be unstable (e.g., Merritt 1987; Stiavelli et al. 1993; Bertin et al. 1994, and references therein).
3. Shortcuts for the identification of BH candidates would be helpful. For example, Kormendy (1992a, b, 1993) showed that BH candidates deviate from the core fundamental plane correlations: they have large apparent central velocity dispersions for their core parameters.
4. Further work on emission-line velocity fields is promising.
5. Our ultimate goal is to understand BH demographics. At a minimum, we want to know the frequency of occurrence and mass function of BHs as a function of Hubble type. This will be expensive.
6. Surely the most fundamental challenge is to show whether or not MDOs are BHs. Our observations are 10⁵ Schwarzschild radii away from the action. The obvious way to get closer is to look at more energetic photons. One possibility is x-ray spectroscopy. But gas will always be responsive, and gas-dynamical searches will always be uncertain.
The issue is important. Even if AGNs are not primarily starbursts, starbursts happen, and Terlevich and collaborators correctly point out that they leave remnants. In detecting MDOs, could we just be finding those dead star clusters? In fact, given our present ignorance, can we even be sure that MDOs are not concentrations of halo dark matter?
7. Therefore, work on alternatives to the BH paradigm is still desirable.
8. How did BHs form? Kormendy (1988b) points out that core relaxation times are almost everywhere so long that BH formation via dynamical evolution - e.g., core collapse - seems impossible. Rees (1984) suggests formation mechanisms that do not depend on relaxation; we do not know which (if any) are correct. The problem is particularly acute because quasars turned on so early in the history of the Universe: BHs must have formed quickly (e.g., Rees 1990, 1993). As long as BH formation remains obscure, we have a glaring hole in the AGN paradigm.
9. Should MDOs be so inactive? Predictions of flaring frequencies (Section 7) may lead to tests of the BH picture.
10. More generally, we need to look for points of contact between dynamical BH searches and the AGN paradigm. Confidence and progress both are at stake.

This review is being written at a time of unusually rapid progress. We discussed ongoing work as much as possible. But stay tuned: by the time this paper appears, HST should begin to provide stellar-dynamical results on the most important BH candidates.

ACKNOWLEDGMENTS

It is a pleasure to thank the many people who sent preprints and reprints, including those on subjects that were not covered because of space limitations. We also thank R. Bender, A. Evans, S. Faber, R. Genzel, F. Melia, J. Moran, G. Rieke, S. Tremaine, and R. van der Marel for helpful discussions or permission to quote results before publication. We are grateful to the following for providing figures or data for figures: S. Faber (data for Figure 3), T. Lauer (Figure 4), A. Dressler (data for Figure 5), R. Bacon and G. Monnet (Figure 6), R. Bender (LOSVDs in Figure 7), W. Dehnen and R. van der Marel (calculated their M32 models as seen at SIS resolution, Figure 10), R. Genzel (Figure 11), H. Ford (Figure 12), and J. Moran (data for Figure 13). Our own work on the BH search benefits enormously from efforts at the CFHT to improve the image quality. JK's work has been supported by NSF grants AST 8915021 and AST 9219221. DR is supported by NASA grant NAG 5-2758. This paper has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by JPL and Caltech under contract with NASA.

1.	We need to address any lingering doubts about the best BH cases. This requires at least one iteration in improved spatial resolution over the discovery observations plus dynamical models that explore the boundary between extreme models with constant M/L and ones that contain BHs.
2.	Theory may help us to eliminate some models. Reducing M/L requires radial anisotropy; such models may be unstable (e.g., Merritt 1987; Stiavelli et al. 1993; Bertin et al. 1994, and references therein).
3.	Shortcuts for the identification of BH candidates would be helpful. For example, Kormendy (1992a, b, 1993) showed that BH candidates deviate from the core fundamental plane correlations: they have large apparent central velocity dispersions for their core parameters.
4.	Further work on emission-line velocity fields is promising.
5.	Our ultimate goal is to understand BH demographics. At a minimum, we want to know the frequency of occurrence and mass function of BHs as a function of Hubble type. This will be expensive.
6.	Surely the most fundamental challenge is to show whether or not MDOs are BHs. Our observations are 10⁵ Schwarzschild radii away from the action. The obvious way to get closer is to look at more energetic photons. One possibility is x-ray spectroscopy. But gas will always be responsive, and gas-dynamical searches will always be uncertain. The issue is important. Even if AGNs are not primarily starbursts, starbursts happen, and Terlevich and collaborators correctly point out that they leave remnants. In detecting MDOs, could we just be finding those dead star clusters? In fact, given our present ignorance, can we even be sure that MDOs are not concentrations of halo dark matter?
7.	Therefore, work on alternatives to the BH paradigm is still desirable.
8.	How did BHs form? Kormendy (1988b) points out that core relaxation times are almost everywhere so long that BH formation via dynamical evolution - e.g., core collapse - seems impossible. Rees (1984) suggests formation mechanisms that do not depend on relaxation; we do not know which (if any) are correct. The problem is particularly acute because quasars turned on so early in the history of the Universe: BHs must have formed quickly (e.g., Rees 1990, 1993). As long as BH formation remains obscure, we have a glaring hole in the AGN paradigm.
9.	Should MDOs be so inactive? Predictions of flaring frequencies (Section 7) may lead to tests of the BH picture.
10.	More generally, we need to look for points of contact between dynamical BH searches and the AGN paradigm. Confidence and progress both are at stake.