Current Research Interests of James Chiang

X-ray Reprocessing in AGN Accretion Disks

It is generally believed that AGNs are powered by disk accretion onto supermassive black holes, and disk models do indeed provide a description of the spectral properties of AGNs which is roughly consistent with the observations: As expected for the surface temperature of a thin disk around a supermassive black hole, the bulk of the bolometric emission occurs in the UV; and in accord with the expectation that a hot, optically thin corona is associated with the disk, the X-ray spectrum above 1 keV consists of a power-law-like continuum which is believed to be produced by thermal Comptonization of the softer disk photons in this corona. The X-ray spectra also show features which are most likely due to reprocessing of these X-rays by relatively cold disk material --- the iron K-alpha fluorescence line at 6.4 keV and the so-called ``Compton reflection hump'' whose shape below 10 keV is determined by photoelectric absorption (e.g., Zdziarski et al. 1996). The broad, red-shifted shape of the iron line indicates that this emission is produced from regions of a thin accretion disk close to the innermost stable orbit around the central black hole (e.g., Tanaka et al. 1995).

However, when the expected spectral and temporal properties of the disk/corona system are compared in detail with the observations, significant inconsistencies are found. In particular, it is well-known that the polarization properties of the UV spectrum, particularly around the Lyman edge, cannot presently be explained by disk models, and further, the general absence of a strong Lyman absorption edge is itself not understood (Koratkar & Blaes 1999). In addition, many type 1 Seyferts have shown a strong component in the EUV/soft X-rays around 50--100 keV which cannot be accounted for by an extrapolation to higher energies of the optical/UV disk component (Magdziarz et al. 1998). In some cases, it even appears as though this soft X-ray excess may dominate the bolometric luminosity of the overall spectral energy distribution (SED).

A further puzzle is the apparent simultaneity of the optical through UV emission. Multiwavelength reverberation mapping campaigns of numerous type 1 Seyferts have shown that the variability across these wavebands are simultaneous on time scales which are several orders of magnitude shorter than the expected signal propagation time through a standard thin disk (Peterson et al. 1991; Clavel et al. 1991). Since emission at different wavelengths are associated with quite different characteristic radii in a multitemperature disk, it has been suggested that the observed variability is due to reprocessing of higher energy emission, such as the hard X-ray continuum, so that the signal propagation times are limited by the speed of light rather than by the sound speed in the accretion flow (e.g., Krolik et al. 1991). However, recent simultaneous optical/UV/X-ray observations of Seyferts have shown that this reprocessing scenario is untenable, since the X-ray and optical/UV emission are either uncorrelated or the X-rays are seen to lag the lower energy emission (Nandra et al. 1998; Chiang et al. 2000; Maoz, Edelson, & Nandra 2000).

Our recent multiwavelength observing campaign of the type 1 Seyfert NGC 5548 has shed some light on these problems and has pointed the way towards a more consistent model of radio quiet AGNs (Chiang et al. 2000). One of our principle findings was that the correlated delays between variations seen in energy bands from the EUV through the hard X-rays are consistent with Compton diffusion time scales. By measuring these temporal delays and using a spectral decomposition of the overall X-ray SED to constrain the properties of the Comptonizing corona, we were able to obtain for the first time an estimate for the physical dimensions of the putative corona. We find that the spectral and temporal variations are consistent with a central corona of approximate size 10rg (where rg = GM/c) surrounded by a cold accretion disk. This inner disk radius agrees with the value found from the spectral fits to the relativistically broadened, red-shifted iron K-alpha emission line which we see in our simultaneous ASCA spectra. It is also consistent with the transition radius between the radiation pressure supported and gas pressure supported regions of a standard thin disk (e.g., Novikov & Thorne 1974) if one uses the mass of the central black hole obtained from BLR reverberation mapping estimates (Peterson & Wandel 1999) and the observed luminosity bolometric luminosity to infer the accretion rate.

These results strongly suggest a model in which the Comptonizing corona is identified with the inner radiation pressure supported part of the disk and may be powered by the various thermal and viscous instabilities known to be associated with this region. The optical/UV emission would then be produced in the outer, cooler gas pressure supported disk which has the usual T ~ r^{-3/4} temperature dependence. The low frequency tail of a multitemperature thin disk with this temperature profile has a flux dependence F_nu ~ nu^{1/3}; and for disks which extend to the innermost stable circular orbit, the optical/UV wavebands typically sit in this part of the theoretical disk spectrum. However, the optical/UV spectra are generally seen to roll over to have a softer spectrum than this. One way of producing this roll-over is by truncating the inner disk at a radius greater than the that of the innermost stable circular orbit (see, e.g., Hubeny et al. 1999). For NGC 5548, it again turns out that inner disk radii of ~10--20rg are required to fit the optical/UV spectra. For smaller inner truncation radii, the computed spectra are brighter and harder. This then provides a possible explanation for the correlated broad band optical/UV variability: it is simply due to the transition radius of the thin disk region varying from 10 to 20rg. Since the doubling time scale for the optical/UV variability in NGC 5548 is 10 days this agrees with the thermal time scale in the disk at these radii if the viscosity parameter of the disk flow is about 0.1.

Our simultaneous EUV/hard X-ray data have also shown that the soft X-ray emission may not constitute an anomalous component in the overall SED of NGC 5548. We note that the previous observation of NGC 5548 by EUVE was not simultaneous with the hard X-ray data to which it was compared (Magdziarz et al. 1998). Our observations show that the EUV flux at about 0.1 keV is probably consistent with an extrapolation of the Comptonized hard continuum down to these energies. However, the lack of spectral coverage provided by EUVE have made a spectral decomposition at these energies difficult.

Accretion Disk Winds

At the Canadian Institute for Theoretical Astrophysics (CITA), Prof. Norm Murray and I developed a model for the broad emission and absorption lines in the optical and UV spectra of AGNs. In contrast to the traditional picture of AGN Broad Line Regions (BLRs) in which the line emission and absorption features are attributed to a large number of discrete clouds surrounding the central black hole, in our model, the lines are formed in a disk wind. The wind is ionized and radiatively driven from the surface of the accretion disk by the central continuum emission and by locally generated disk radiation. We have shown how broad absorption line features of BAL QSOs with blue-shifted velocities up to 100,000 km/s can be formed in these disk wind outflows ( Murray, Chiang, Grossman, & Voit 1995). Furthermore, we have also shown how broad single-peaked emission lines are produced at the interface between the wind streamlines and the disk surface (Chiang & Murray 1996; Murray & Chiang 1997). These lines possess a similar temporal response to changes in the central ionizing continuum as that seen in reverberation mapping observations of various AGNs such as NGC~5548 (Korista et al. 1995; Chiang & Murray 1996).

I am presently developing the disk wind model further by computing the wind dynamics using the hydrodynamic code ZEUS-2D. The analytic calculation of Murray et al. (1995) relied upon an approximate model of the wind dynamics in which the radial and azimuthal fluid equations were decoupled. This allowed us to solve the dynamical equations directly and perform photoionization calculations along a radial line-of-sight. Integrating the contribution to the driving due to the various lines, we were able to employ the force multiplier formalism used in the study of O star winds to determine the radiative acceleration self-consistently. In collaboration with Prof. Murray, the present numerical modeling will allow us to integrate the fully coupled equations. As before, we will use a photoionization code such as CLOUDY to determine the ionization structure of the wind and the radiative driving term self-consistently. A preliminary ZEUS-2D simulation of an AGN disk wind can be found here.

In the context of this modeling, we can address several important issues regarding AGN BLRs. The profiles for the different lines, e.g., CIV, NV, Lyman-alpha, MgII, will be simultaneously modeled, and hence we will find constraints on the elemental abundances in the disks of AGNs. We will also seek to reconcile the differences in the reverberation mapping response of Lyman-alpha versus H-beta in the disk wind model by taking into account the presence of dust in the radiative transport. These calculations may also provide the dynamical connection suggested by the apparent correlation between absorption trough structure and the shape of the accompanying emission lines in BAL~QSOs (Turnshek 1987). In addition, we will compare the observed polarization (Goodrich & Miller 1995; Ogle 1997) of the emission and absorption lines in BAL and BEL QSOs to our model which would then provide a useful test of our picture of the geometry of the BLR. These investigations are being supported by a grant from the NASA Astrophysics Theory Program.

Papers:

Gamma-Ray Bursts

During my tenure at the Naval Research Laboratory (NRL) as a National Research Council Fellow, I worked with Dr. Charles D. Dermer on models of gamma-ray bursts. Some of my most recent work there included writing a numerical code which models the propagation of a relativistic blast wave into the ambient circum-burst material (Chiang & Dermer 1999). The code is self-consistent in the sense that it accounts for radiative losses (due, for example, to synchrotron and synchrotron self-Compton processes) in determining the energy and momentum balance governing the deceleration of the blast wave material. The code is able to describe the spectral and temporal features of the afterglow emission which has been recently associated with gamma-ray bursts and which confirms their cosmological nature. It is apparent from the optical and radio characteristics of this emission that it is composed of synchrotron radiation from a relativistically expanding blast wave (e.g., Galama et al. 1998). Along with the afterglow emission and the observationally determined distance to a given burst, this numerical code can provide precise estimates of the density of the ambient medium and the total energetics of the burst.

One of the outstanding issues regarding GRBs is whether the interaction of the blast wave with ambient material also produces the prompt gamma-ray emission at early times. This is the so-called ``external shock'' model and its essential elements were originally proposed by Rees & Meszaros (1992). The basic timescales, energetics, and spectral characteristics of a majority of GRBs are roughly consistent with this interpretation. Furthermore, general equipartition arguments for the characteristic particle energies in a relativistic blast wave show that the evolution of these energies due to the deceleration of the blast wave can mimic the observed spectral softening which has been seen in the pulses of the prompt burst light curve (Fenimore et al. 1995; Katz 1994; Chiang 1998, 1999). In this model of the prompt burst emission, the main mechanism which has been posited for producing the highly variable and complex GRB light curves is the interaction of the blast wave with inhomogeneities in the ambient matter. The blast wave itself then provides a probe of the circum-burst environment via its emissions and the structure of this environment would be reflected in the burst light curve. Knowledge of this ambient material could give some insights into the more fundamental question of the nature of the burst progenitor.

There are, however, substantial problems with the external shock model as an explanation of the prompt burst emission. As Fenimore, Madras, & Nayakshin (1996) have pointed out there are problems with producing the observed light curves shapes having to do with spatial filling factors and the geometry of the ambient inhomogeneities. In accord with these findings, my own work on burst pulse fluences due to the interaction of a spherical blast wave with dense clouds as a function of cloud radial distance essentially shows that these clouds must form a spherical shell surrounding the burst central object (Chiang 1999). This geometry for the external medium places tight constraints on both the amount of variability of which can be obtained in this model and on the feasible geometries of the external medium.

Papers:

My PhD thesis (Dept. of Physics, Stanford University, 1994) :

Gamma Radiation from Rotation-Powered Pulsars and Results from the Energetic Gamma Ray Experiment Telescope Last modified: Fri Jun 23 13:34:31 MDT 2000