1. INTRODUCTION

Active galactic nuclei (AGNs) are galaxies whose nuclei are highly luminous, with spectra showing broad emission lines covering a wide range of ionization. AGN activity spans a broad range of luminosity, from the brightest quasi-stellar objects (QSOs) to the more common but less luminous Seyfert galaxies. Two significant facts about AGNs make them very interesting to study. The first is their ability to generate extraordinary luminosities (up to 10⁴⁷ ergs s^-1) in tiny volumes ( 2 × 10¹⁴ cm; see, e.g., Edelson et al. 1996). The second is the similarity in their spectral features over 7 orders of magnitude in luminosity, across most of the electromagnetic spectrum. No other phenomena have these unusual properties. The similarity in spectral properties over many orders of luminosity indicates that the central engine in AGNs must have the ability to scale with luminosity. It is widely believed that accretion of gas into a central supermassive black hole lies at the heart of the phenomenon (see, e.g., Rees 1984). The accretion flow is thought to be the source of the X-ray, ultraviolet (UV), and optical continuum emission, which ionizes circumnuclear gas in both the broad-line region and narrow-line legion, and may also be the launch site for winds and jets. Well outside the broad-line region (BLR) lies a dusty torus (or perhaps a warped disk) that obscures some lines of sight to the nucleus and accounts for some of the observed diversity in the AGN phenomenon (Antonucci 1993). Readers interested in a broad overview and introduction to the field of AGNs can find several good textbooks, including Peterson (1997) and Krolik (1999).

The standard picture of the central engine of AGNs now has limited observational support. On the smallest scales from Hubble Space Telescope (HST) imaging and spectroscopic observations, there is evidence for the existence of massive compact objects at the centers of galaxies (Kormendy et al. 1996a, 1996b, and 1997; van der Marel et al. 1997; Ford et al. 1994; Harms et al. 1994; Macchetto et al. 1997; Bower et al. 1997; van der Marel & van den Bosch 1998; Richstone 1998; Magorrian et al. 1998). Very broad Fe K lines observed by the Advanced Satellite for Cosmology and Astrophysics (ASCA) indicate the presence of gas moving in a relativistic potential well (Tanaka et al. 1995; Mushotzky et al. 1995; Nandra et al. 1997b). On parsec scales, HST imaging observations (Walter et al. 1996; Ford et al. 1998) and water megamaser observations of a handful of nearby AGNs show a disk structure, possibly the thick molecular torus (Nakai, Inoue, & Miyoshi 1993; Greenhill et al. 1995; Braatz, Wilson, & Henkel 1997).

Despite the fact that much of the large-scale structure has now been observationally verified, we still do not understand the central engine itself. The main difficulty in achieving this goal is that the central accretion flow, which is responsible for the observed optical, UV, and X-ray spectrum, is too small (~ 10 R_G = 5 × 10^-5 pc for a 10⁸ solar mass black hole, where R_G = GM / c² is the gravitational radius of the black hole) to resolve at the distances of even the nearest AGN. (There is one interesting exception: the Einstein Cross, which we discuss in Section 3.6 below.) Therefore, we are forced to rely on theoretical models, their predictions, and comparison with observations in order to make progress.

The theoretical accretion models are distinguished by the assumed flow geometry, the state of the plasma within the flow, and the mechanism whereby gravitational energy is converted into observed radiative and kinetic power. The simplest flow geometry is spherical, but such flows are generally very inefficient radiators unless there is some internal source of dissipation, e.g., shocks (Mészáros & Ostriker 1983). Observations on larger scales than the central accretion flow suggest that spherical symmetry is probably broken and that a well-defined axis is present in the central engine. These include the evidence for large-scale gaseous disks as discussed above, as well the presence of jets in both radio-loud and radio-quiet AGNs (Livio 1997; Blundell & Beasley 1998). The axis is generally believed to be associated with the angular momentum of the accretion flow and the hole itself. The specific angular momentum of gas in the central regions of galaxies generally greatly exceeds that of a particle in a circular orbit near the black hole (~ R_G c). To have less than this critical value of angular momentum, material 1 pc away from a 10⁸ M black hole would require nonradial velocity components less than 1 km s^-1. Centrifugal forces on material of fixed specific angular momentum vary with radius r as r^-3, while gravity varies as r^-2. Therefore, unless angular momentum is transported outward on a dynamical timescale as material flows inward from large scales down to the black hole, the flow must become rotationally supported and form an accretion disk. As first pointed out by Salpeter (1964) and Zeldovich & Novikov (1964), accretion disks can provide a high efficiency of matter-to-energy conversion in a very small volume. The outward transport of angular momentum occurs slowly through poorly understood dissipative processes ("viscosity"), thereby converting a significant fraction of gravitational energy into heat, magnetic fields, and/or outflowing kinetic energy. It is this largely theoretical reasoning that has made disks so ubiquitous as models of the central parts of the accretion flow in AGNs.

The most common assumption about the state of the plasma within the disk is that it is optically thick and thermal. This immediately implies that a substantial fraction of the bolometric power should be in the form of UV photons: a blackbody emitting at a significant fraction of the Eddington luminosity on size scales associated with supermassive black holes has a temperature in the UV range. In fact, a lot of power in AGNs is emitted in the optical/UV region of the spectrum (the Big Blue Bump), but the full spectral energy distribution (SED) is rather more complicated than that. As shown in Figure 1, the overall broadband AGN continuum is relatively flat in F and extends over nearly 7 orders of magnitude in frequency, which implies that approximately the same amount of energy is emitted per decade of frequency. For radio-quiet AGNs, this broadband spectrum can be divided into three major components: the infrared bump, the Big Blue Bump, and the X-ray region.

Figure 1. Schematic representation of the broadband continuum spectral energy distribution seen in the different types of AGNs. The radio-quiet spectrum can be divided into three major components: the infrared bump, which is thought to arise from reprocessing of the UV emission by dust in a range of temperatures and at a range of distances; the Big Blue Bump, which is directly related to the main energy production mechanism and may be due to an accretion disk; and the X-ray region, which can be interpreted as the high-energy continuation of the Big Blue Bump together with a Comptonized power law with fluorescence and reflection from "cold" material.

The Big Blue Bump (BBB) continuum component in AGNs extends from the near-infrared at 1 µm to past 1000 Å in the UV and in some cases apparently all the way to the soft X-ray region of the spectrum. More than half the bolometric luminosity of an unobscured AGN is typically emitted within this spectral range, and thus its origin is directly related to the main energy production mechanism. The BBB is thought to arise from an accretion disk and is the spectral region that we will mostly be discussing in this paper. We discuss the observed properties of the BBB extensively in Section 3.

The broad infrared bump extends from ~ 100 to ~ 1 µm. Quasars show a deep minimum at 1 µm and have a sharp cutoff in the submillimeter. In Seyfert galaxies, the 1 µm minimum is rarely detected because of galaxy starlight contamination (Neugebauer et al. 1987; Sanders et al. 1989). Prior to longer wavelength observations, it was thought that the near-infrared emission in radio-quiet AGNs could be due to synchrotron emission, which could extend as a power law to higher photon energies underneath the BBB. Today, however, the infrared bump is generally thought to arise from reprocessing of the BBB emission by dust with temperature in the range of 10-1800 K and at a range of distances from the central UV source (Barvainis 1987, 1990; Sanders, Scoville, & Soifer 1988; Phinney 1989). The spectral trough at 1 µm is then naturally interpreted as being due to the finite sublimation temperature (~ 1800 K) of dust. Additional evidence for the dust interpretation is the fact that the infrared hump shows no rapid variability (see, e.g., Hunt et al. 1994) and that the long-wavelength turnover in the submillimeter is very steep (see, e.g., Hughes et al. 1993). Observations of a handful of radio-quiet AGNs in the far-infrared using the Infrared Space Observatory show strong evidence for thermal emission (Haas et al. 1998). In the radio-loud AGNs, the infrared emission is a mixture of thermal and nonthermal radiation, but in most cases one of the two components is dominant (Haas et al. 1998).

In the soft X-ray region, many objects show a "soft X-ray excess," which is emission that exceeds the extrapolation from the observed hard X-ray power-law continuum (Turner & Pounds 1989; Masnou et al. 1992). This excess is subject to a number of observational difficulties and is probably the least well understood component of the AGN's spectral energy distribution (SED). It may be that it is the high-energy continuation of the BBB, but this interpretation is not straightforward. We discuss the details in Section 3.1 below. The characteristics of the soft X-ray excess in both Seyfert galaxies and quasars are not well quantified. Radio-quiet QSOs have a mean energy spectral index (F ^_X) of _X = -1.72 ± 0.09 in the soft X-ray band (Laor et al. 1997), while Seyfert 1 galaxies have _X ~ -1.37 (Turner, George, & Mushotzky 1993; Walter & Fink 1993).

In the hard X-ray band, the AGN spectrum consists of a power law with energy spectral index ~ -0.9. This power law is almost universally believed to be due to Compton upscattering of optical/UV photons by hot or nonthermal electrons somewhere in the central engine, possibly a hot, magnetized corona above the disk. Superposed on this power law are an Fe K emission line at 6.4 keV and a Compton reflection hump above 10 keV, at least in the case of some Seyfert galaxies (Pounds et al. 1989; Nandra & Pounds 1994). These two features are thought to be due to fluorescence and reflection from "cold" material, possibly the accretion disk itself in the case of Seyfert 1 galaxies. This is supported by the fact that the Fe K emission line is observed to be very broad (Tanaka et al. 1995). Both the reflection hump and the hard X-ray power law extend up to high energies before cutting off at around several hundred keV (Gondek et al. 1996). The hard X-ray properties of high-luminosity AGNs (QSOs) are not as well understood, partly because they are relatively faint in the hard X-rays given their optical/UV luminosity compared to Seyfert galaxies. The Fe K line and reflection hump are relatively weak or nonexistent (Iwasawa & Taniguchi 1993; Nandra et al. 1995; Nandra et al. 1997c).

It is the purpose of this paper to review the current status of the observational and theoretical research effort to understand the central accretion flow and its relation to the BBB. We focus almost exclusively on radio-quiet AGNs, whose emission is dominated by the accretion flow. Radio-loud objects can also have substantial contributions to their observed emission from jets and outflows, and we refer the reader to the excellent review of Urry & Padovani (1995) for a discussion of their properties. While we will mention the important clues being provided by other regions of the electromagnetic spectrum, particularly X-rays, we concentrate here on the optical/UV BBB. We begin Section 2 of the paper by discussing the early, simple accretion disk models that were first applied to the AGN problem. These models have proved to be very inadequate, and we discuss the main observational problems in Section 3. We then turn to more recent efforts to improve these models and how they fare against the observations in Section 4. We discuss alternatives to the standard accretion disk paradigm in Section 5. Finally, we summarize our main conclusions in Section 6, pointing out the main research directions, both theoretical and observational, which we feel will lead to further progress in our understanding of the primary power source of AGNs.