The subject of this book is astrophysical accretion, especially in those circumstances where accretion is believed to make an important contribution to the total light of an astrophysical system. Our discussion therefore centres mainly on close binary systems containing compact objects and on active nuclei. The reader is assumed to possess a basic knowledge of physics at first degree level, but only a rudimentary experience of astronomy is required. We have tried to concentrate on those features, particularly the basic physics, that are probably more firmly established; but the treatment is necessarily somewhat heterogeneous. For example, there is by now a tolerably coherent line of argument showing that the formation of an accretion disc is very likely in many close binaries, and giving a plausible picture of what such a disc is like, at least in some simple cases. In other areas, such as accretion on to the surface of a compact object, or in active nuclei, we are not so fortunate, and we must work back and forth between theory and observation. Our aim is that the book should provide a systematic introduction to the subject for graduate students. We hope it may also serve as a reference for interested astronomers in other fields, and that selected material will be suitable for undergraduate options in astronomy.

In Chapters 2 and 3 we present introductory material on fluid dynamics and plasma physics. Many excellent texts exist in these areas, but they tend to be too detailed for our needs; we have tried to extract just those basic ideas necessary for the subsequent discussion, and to set them in an astrophysical context.

Introduction

Whenever we need to consider the behaviour of a gas on lengthscales comparable to the mean free path between collisions, we must use the ideas of plasma physics. In this chapter we shall briefly introduce some of the concepts that will be important to our study of accretion.

A plasma differs from an atomic or molecular gas in that it consists of a mixture of two gases of electrically charged particles: an electron gas and an ion gas, with very different particle masses me and mi.

The electrons and ions interact with each other through their electrostatic Coulomb attractions and repulsions. These Coulomb forces decrease only slowly (∞ r-2) with distance and do not have a characteristic lengthscale. Thus, a plasma particle interacts with many others at any one instant, and this makes the description of collisions more complicated than in atomic or molecular gases, where the interparticle forces are very short-range. A further complication arises from the great difference in particle masses me and mi. Since collisions between particles of very different masses can transfer only a small fraction of the kinetic energy of order me/mi ≪ 1, it is possible for electrons and ions to have significantly different temperatures over appreciable timescales. These two properties – the long-range nature of the Coulomb force and the disparity in electron and ion masses – give the physics of plasmas its particular character. A further series of complex phenomena occurs when the plasma is permeated by a large-scale magnetic field; this is particularly relevant for the study of gas accreting on to highly magnetized neutron stars and white dwarfs.

Introduction

We have seen in the foregoing chapters that a huge variety of accretion flows are at least theoretically possible. The equations describing axially symmetric flows with gravity, pressure and rotation allow a wide array of solutions of astrophysical interest. If one includes Ω = 0 (no rotation) as a special case, then even ordinary stars and spherically symmetrical Bondi accretion are solutions. More realistically, accretion flows with low angular momentum may produce supersonic flows which shock at smaller radii if they lack pressure support, or produce settling, cooling solutions if partially supported by pressure. We have already studied in detail spherically symmetrical accretion, the standard thin disc, and thick discs, and we will study other solutions which have acquired recognition in the astrophysical literature, such as slim discs and advection dominated accretion flows (ADAFs). In this chapter we shall attempt to organize all the different solutions into a coherent picture in order to clarify their relation to each other.

The fact that black holes possess an event horizon instead of a hard surface makes the inner boundary condition for black hole accretion flows qualitatively different from that for a normal star, and allows the existence of a family of solutions in which a significant fraction of the dissipated energy is advected through the horizon. Hence, for an external distant observer, ADAFs on to black holes are characterized by a low radiative efficiency η. (In ADAFs on to objects with a hard surface the advected energy must be ultimately reprocessed and released near the surface – see Section 11.8.3.)

Introduction

All accreting matter, like most of the material in the Universe, is in a gaseous form. This means that the constituent particles, usually free electrons and various species of ions, in teract directly only by collisions, rather than by more complicated short-range forces. In fact, these collisions involve the electrostatic interaction of the particles and will be considered in more detail in Chapter 3. On average, a gas particle will travel a certain distance, the mean free path, λ, before changing its state of motion by colliding with another particle. If the gas is approximately uniform over lengthscales exceeding a few mean free paths, the effect of all these collisions is to randomize the particle velocities about some mean velocity, the velocity of the gas, v. Viewed in a reference frame moving with velocity v, the particles have a Maxwell–Boltzmann distribution of velocities, and can be described by a temperature T. Provided we are interested only in lengthscales L ≫ λ we can regard the gas as a continuous fluid, having velocity v, temp erature T and density ρ defined at each point. We then study the behaviour of these and other fluid variables as functions of position and time by imposing the laws of conservation of mass, momentum and energy. This is the subject of gas dynamics. If we wish to look more closely at the gas, we have to consider the particle interactions in more detail; this is the domain of plasma physics, or, more strictly, plasma kinetic theory, about which we shall have something to say in Chapter 3.

In the years since the first edition of this book appeared the study of astrophysical accretion has developed rapidly. Perhaps the most fundamental change has been the shift in attitude over active galaxies and quasars: the view that accretion is the energy source is now effectively standard, and the emphasis is much more on close comparison of observation and theory. This change, and the less spectacular but still profound one which has occurred in the study of close binary accretion, have been largely brought about by the wealth of new data accumulated in the interval. In X-rays, the ability of EXOSAT to observe continuously for as much as 3 to 4 days was a dramatic advance. In the optical, new instrumentation has produced far tighter observational constraints on theory. Despite these challenges, the basic outlines of the theory are still recognizably the same.

Of course our understanding is very incomplete. As the most glaring example, we still have essentially no idea what drives disc accretion; and there are new problems such as the dynamical stability of thick discs, or the nature of fieldline threading in magnetic binaries. But it is now difficult to deny that some close binaries possess discs approximately conforming to theoretical ideas; or that some kind of anisotropic accretion occurs in active galactic nuclei. Encouragingly, accretion theory is increasingly integrated into wider pictures of the relevant systems. The process is well advanced for close binaries, particularly for the secular evolution of cataclysmic variables, and is in its early stages for active galaxies.

Introduction

In these final three chapters we come to the mechanism by which the gravitational potential energy of material accreting on to a superrnassive black hole is extracted as radiation. We start with the radio emission on kpc–Mpc scales which, where it occurs, is probably the best understood feature. To power the extended radio lobes the central engine in sources with such large-scale radio structure must turn accretion energy into directed bulk relativistic outflow. It is universally accepted that the power law radio spectrum from the lobes is synchrotron radiation from relativistic electrons. At the other extreme the X-ray emission appears to be produced on scales down to tens of Schwarzschild radii, providing the deepest possible probe of conditions near the black hole. However, the X-ray power law spectrum seems to admit any number of explanations. There is a problem here in distinguishing the primary radiation from any reprocessed components.

Since, therefore, we are not certain of either the geometry of the central source or the emission mechanism responsible for any part of the spectrum from this region, a discussion of accretion power in active nuclei contrasts sharply with our previous consideration of binary star systems. We shall present a range of partial theories each focussed on a different aspect of the problem. The thick discs expected at super- Eddington accretion rates to be considered in Chapter 10, and the electrodynamic disc theories discussed here in Sections 9.6–9.8 are primarily concerned with the production of power in the form of bulk relativistic outflows.

Introduction

In previous chapters we have discussed extensively the theory and applications of thin (H ≪ R) accretion discs. We hope the reader will by now be convinced that this theory is reasonably well understood, and that it rests on a fairly firm observational basis. The case for thick (H ∼ R) accretion discs however is less compelling as the theory is still under development and the relevant observations are few, difficult and indirect. Furthermore, since the publication of the first papers on equilibrium thick discs or tori much work has been done on extending these solutions allowing some form of poloidal flow and studying extensively their dynamical stability. The wealth of these investigations is a testimony to the interest generated by these structures. The results obtained so far virtually rule out the reality of thick discs as non-accreting toroidal equilibria but leave open the more exciting possibility of the existence of closely related accreting flows which could be of astrophysical interest. We begin here by discussing the toroidal equilibria without accretion and summarize the stability results at the end of this chapter. More recent and more general solutions with radial and other poloidal flows are discussed in Chapter 11.

The current interest in the theory of the structure, evolution and stability of thick accretion discs is due to the possibility that thick discs may be relevant to the understanding of the central power sources in radio galaxies and quasars (see Chapters 7–9).

Observations

Accretion on to stellar mass objects occurs in a wide variety of systems and yields a wide variety of observational behaviour. While there may be many arguments over detailed models, the broad basis of these differences is largely understood. Active galactic nuclei also come in many observed forms. From an observational viewpoint, they can be defined as apparently stellar sources but with non-thermal spectra, and, in cases where they can be determined, significant redshifts. Beyond this, we find a wide variety of properties, which we shall classify in more detail below. But in these cases it is not at all clear how these differences arise, or, indeed, whether one is even dealing with variants of a single basic model. We shall argue that the sources are all manifestations of accretion on to supermassive black holes (of order 108M⊙), although even this is still not universally accepted. Furthermore, for stellar-mass objects, at least in some cases, we have a complete picture of the system even if some of the details are missing. In no case do we have anything comparable for active galactic nuclei. That is not to say that there are no aspects of active galactic nuclei that are thought to be fairly well understood, but those that are do not include the mechanism of the basic energy source. Thus we have to try to extract from the available data what clues we can to the nature of the central engine.

In the decade since the second edition of this book, accretion has become a still more central theme of modern astrophysics. We now know for example that a γ-ray burst briefly emits a gravitationally powered luminosity rivalling the output of the rest of the Universe. This and other startling discoveries are a result of observational progress, driven as ever by technological advances. But these advances are also having a powerful effect on theory; modern supercomputers allow one to perform as a matter of routine calculations which were unthinkable a decade ago. This increasing capability will significantly alter the way theory is done, and indeed thought about.

The impact on accretion theory has already been profound. Most obviously, supercomputer simulations have been central in verifying that angular momentum transport in accretion discs is probably mediated by the magnetorotational instability. This opens the prospect of at last understanding how accretion is driven in the discs we see.

These changes and others make a new edition of this book timely. We are grateful for the opportunity of revising and extending the treatments of the earlier editions. As always, we have been obliged to be selective, but have tried to convey the essence of recent developments. In addition to discussing the new work on disc viscosity referred to above, we give a more thorough treatment of the thermal–viscous disc instability model now generally thought to be the basic cause of the outbursts of dwarf novae and soft X-ray transients.

Introduction

The importance of accretion as a power source was first widely recognized in the study of binary systems, especially X-ray binaries. This is still the area where the greatest progress in the understanding of accretion has been made. The reason for this is simply that, by their very nature, binaries reveal more about themselves, notably their masses and dimensions, than do other astronomical objects. This is particularly true in the case of eclipsing binaries, where we get direct information about spatial relations within the source. The importance of accretion is further manifested by the realization that probably a majority of all stars are members of binary systems which, at some stage of their evolution, undergo mass transfer.

The detailed study of interacting binary systems has revealed the importance of angular momentum in accretion. In many cases, the transferred material cannot land on the accreting star until it has rid itself of most of its angular momentum. This leads to the formation of accretion discs, which turn out to be efficient machines for extracting gravitational potential energy and converting it into radiation. This property has made accretion discs attractive candidates for the role of the central engine in quasars and active galactic nuclei; the question of whether or not this assignment is correct will be a theme of later chapters. In the context of binaries, however, their existence and importance is well attested through observation; we shall discuss the evidence for this in Chapter 5.